Symposium Mammographicum Conference 2021

Introduction: Stereotactic breast biopsies are performed for the assessment of indeterminate mammographic abnormalities where no target can be identified on ultrasound scanning. Methods: A retrospective audit of stereotactic biopsies performed at the Nottingham Breast Institute was undertaken to quantify the number of malignant and benign biopsy results. Patient demographics, characteristics of the mammographic abnormality and pathological diagnosis were extracted. This analysis focussed on patients recalled from the National Health Service Breast Screening Programme (NHSBSP). Results: Between November 2018 and October 2019 68 stereotactic breast biopsies were performed. The majority were for patients recalled from the NHS BSP (52, 76.5%). Mean age of patients in this cohort was 58.8 years. 33 (63.5%) malignant biopsies and 19 (36.5%) benign biopsies were obtained. Within the NHSBSP recall group 40 (76.9%) patients were recalled for indeterminate microcalcifications (M3/M4). Within this group 26 (65%) malignant and 14 (35%) benign biopsy results were obtained. Mean age of patients in the malignant and benign biopsy group were 57.2 and 61.1 years respectively. Mean size of microcalcifications in the malignant and benign biopsy group were 30.4mm and 18.9mm respectively. Histograms demonstrate overlap between groups with the majority of microcalcifications measured between 10mm and 20mm. Malignant calcifications tend to be spread over a greater distance and there is wider frequency distribution in this group. Conclusion: Two thirds of stereotactic biopsies performed for NHSBSP recall patients for indeterminate calcifications yield a malignant diagnosis in our practice.

Background: Early diagnosis of breast cancer saves lives [1,2] and is the aim of the NHSBSP [3] . However, mammographic screening programmes result in disputed levels of both over-diagnosis and underdiagnosis [4,5] . Breast MRI is better at finding cancers but scanning takes longer and needs senior medical staff to interpret it. The aim of this paper is to estimate the budget impact to the NHS of the use of MRI in breast cancer population screening. The objectives are: To estimate the cost per extra detected breast cancer of (a) using MRI alongside mammography vs. mammogram alone, and (b) MRI alone vs. mammography alone. Methods: A decision analytical model taking the perspective of the NHS in England using a three-year time horizon and price year 2019 was developed. Model parameters were taken from literature. Results: The analysis showed increased costs associated with the use of MRI both for the inclusion of MRI with mammography vs. mammography alone and for MRI alone vs. mammography. The cost per extra detected cancer was between £22,170-£14,642 and £17,144-£9,859 respectively. Conclusions: To reduce breast cancer mortality, we need a costeffective screening test that preferentially finds aggressive cancers that are not well seen on mammogram and works well for all women. MRI is effective but the cost may be prohibitive. Further research, to determine the cost effectiveness of alternative versions of breast MRI, such as FAST MRI, which reduce scanning times and can be interpreted by the same professionals who interpret mammograms, is needed.
Funding: NIHR RfPB (ISRCTN 16624917) [1]. Saadatmand S, Bretveld R, Siesling S, Tilanus-Linthorst MMA. Influence of tumour stage at breast cancer detection on survival in modern times: Population based study in 173 797 patients. BMJ. 2015;351. [2]. Massat  Purpose: To investigate the expression of the gadoliniumtransporting solute carriers SLCO1B1 and SLCO1B3 in breast cancer. Both transporters were previously thought to be liver-specific and are known to influence the enhancement characteristics of liver lesions on MRI. If these transporters are also expressed in breast cancer cells it may influence the enhancement characteristics of breast cancers on MRI. Methods: Transporter expression was investigated in two breast cancer cell lines, MCF-7 (hormone responsive) and MDA-MB-231 (hormone non-responsive), and in tumour samples from four patients (all ER positive invasive ductal carcinoma). Semi-quantitative and realtime PCR (sq-PCR and RT-PCR) was used in both cell lines followed by bulk-cell RNA/single-cell RNA sequencing. Protein expression was investigated by immunocytochemistry. In the tumour samples, expression was investigated by sq-PCR and RT-PCR. Results: SLCO1B1 and SLCO1B3 showed differential expression between the MCF-7 and MDA-MB-231 cell lines, with considerably higher expression in MDA-MB-231 cells than MCF-7 cells both at the gene and protein levels. Expression of SLCO1B1 was detected in all four patient tumours at variable levels. SLCO1B3 was detectable in 2 of 4 patient tumours, and when present it was detected at high levels.
Conclusions: Confirming the presence and differential expression of SLCO1B1 and SLCO1B3 in breast cancer cells suggests that these gadolinium transporters may play a role in tumour enhancement on MRI. Further characterisation of expression levels in different tumour types, such as invasive lobular carcinomas, may help to substantiate this and provide a basis to develop studies to investigate its potential relevance in clinical practice. Objectives: To identify the onset of fatigue during DBT reading through the use of eye tracking and analysing blinking behaviour. The multiple image slices associated with DBT makes reading a longer and potentially more strenuous process when compared with FFDM. The aim is to provide an informed recommendation regarding the appropriate amount of time to spend reading DBT cases to reduce fatigue onset. Methods: Participants read a set of 40 DBT cases in one session while their visual search behaviour was recorded using a non-intrusive eye tracker. Our focus was on analysing the 'Eyelid Opening' signal generated by the eye tracker and detecting blinks through the development of a dedicated software. This allowed us to identify certain aspects of blinking behaviour, namely blink duration. Blinks were classified by their duration into ultra-short, short and long blinks and micro-sleeps and then correlated with different states of vigilance and fatigue. Results: A statistical statistically significant difference in the average of blink durations for the 20th cases (Mean: 143 ms) and the 40th cases (Mean: 228 ms,) p = 0.0013) across participants was found. Moreover, analyses Analysis of the eyelid opening aperture across participants showed that readers' tend towards a smaller eyelid aperture gets progressively smaller (< 10 mm) in the third hour of reading compared with larger eyelid apertures (>12 mm) in the first two hours of reading. Conclusions: Increased blink durations and reduced eyelid apertures were found after 20 cases were read, indicating that individuals begin to exhibit signs of visual fatigue.

O3.1
Smoking is an independent negative predictor of mammography attendance in women eligible for breast screening Naoise C Synnott 1 , Patricia Fitzpatrick 2 Correspondence: Patricia Fitzpatrick Background: BreastCheck is the national breast screening programme in Ireland; in 2014-2015 the eligible age group was 50-64. The aim of this study was to identify behavioural lifestyle predictors of mammography attendance in Irish women aged 50 -64. Methods: Data from the Irish Longitudinal Study on Ageing (TILDA) Wave 3, conducted in 2014-2015, was used. Multivariable logistic regression was employed to identify independent predictors of mammography attendance. Results: There were 3,575 female participants in TILDA Wave 3; 1,750 were eligible for population breast screening at the time of data collection. 73% reported high activity levels, 16% were current smokers and 12% had an alcohol problem (CAGE score > 2). A significantly lower proportion of smokers (74%) than non-smokers (87%) attended their last mammogram (p<0.001). There was a significant inverse relationship between number of cigarettes smoked/day, and mammogram attendance (p<0.001). In contrast, activity levels, BMI or an alcohol problem did not predict mammography attendance. Using the multivariable model to adjust for co-variates including sociodemographic, health and lifestyle variables, being a current smoker (OR 0.57, 95%CI 0.43-0.74), and smoking ≥20 cigarettes per day (OR 0.49, 95%CI 0.34-0.7), remained negatively associated with mammography attendance. Conclusion: Smoking was the only lifestyle factor in our study associated with mammography attendance in Irish women eligible for population screening. Smoking is higher in lower socioeconomic groups, in whom screening uptake is poor. Smoking cessation advice and courses could potentially include information about breast screening, with complementary cancer prevention effect. Background: Demand on breast services has grown with increasing awareness of breast health and technological advances. However, given the workforce crisis, it is important that the current workforce is supported to work to and maintain high standards. As part of Breast Screening Quality Assurance, we review our film reader statistics (FRQA) annually, including individual recall rates (RR). Over a number of years, it was noted that readers settled into 2 groups; one with a consistently low <5% RR and the second with higher RR (>5%) which did not decrease despite mentoring and regular interventions. Using adult educational learning theory, we reviewed the trends in reading practises for each group to identify ways to reduce the recall rate. Methods: We identified that the low RR group led screening assessment clinics. In accordance with Kolb's experiential learning cycle [1] , to develop a skill there needs to be regular feedback and reflection to allow learners to refine their skills. We asked readers with a higher RR to identify, review and reflect on images that they recalled but were subsequently read normal by the 2nd & 3rd reader. Results: Following establishment of this intervention, the overall RR for this group has fallen significantly and been maintained over 2 years. In the period 2015-2017 the average RR was 5.28/5.6 (1st/2nd reader). Following interventions the average RR over 2017-2019 was 4.7 and 3.9. Conclusion: Reduction in RR was achieved and maintained by introducing an intervention incorporating adult learning theory. [1]  Background and Purpose: Male breast cancer is rare whereas gynaecomastia is very common. Only asymmetrical gynaecomastia require breast imaging and focal lumps are amenable to clinical core biopsy. So the use of ultrasound in the assessment of male breast should be limited. The aim of this study is to audit the referral indications and ultrasound outcomes in male breast US (MBUS) patients against local guidelines. Methods: In the last 5 years, 968 patients were referred for MBUS in our Trust. This audit includes the patients between 02/01/2019-04/ 12/2019. The duplicate patients and follow-ups were excluded from the study. In total, 197 patients were analysed (mean age: 58 (8-90) retrospectively. Referral diagnosis, age, US grading and clinical outcomes were noted. Results: Of the 197 patients, 79% were gynecomastia (133), lipoma (21) or fat necrosis (2), and 15% (30) were normal. There was 1 chest wall lymphoma and 1 DCIS, and 9 (5%) patients had benign breast disease (fibroepithelial lesions, abscess, papilloma, sebaceous cysts, haematoma). In 122 patients (62%) clinical grade was not given, 66 had P2, 8 had P3, 1 had P5. 2 patients were scored as U4 and 4 patients as U3. Conclusions: These results clearly show that 99% of the patients referred to MBUS were benign. And also 95% of the patients were clinically benign or not assessed. The excessive use of MBUS without a clinical indication leads to patient anxiety, increased waiting times and might delay the proper imaging to the patients who should have the priority in terms of clinical indication. Careful clinical assessment before ultrasound referral is mandatory for better care. [1]. Best practice diagnostic guidelines for patients presenting with breast symptoms. NICE Nov 2010 https://www.evidence.nhs.uk/ document?id=2013590&returnUrl=search%3Fq%3Dhc11%26sp%3 Don&q=hc11 [2]. Shrewsbury and telford hospitals clinical guidelines for the management of breast cancer. June 2019 Qualitative research exploring the perceptions of Advanced Practitioner Radiographers (APRs) in delivering biopsy results within a single unit to NHS Breast Screening Programme (NHSBSP) assessment patients. A significant lack of published research in this area provides the rationale for this research, combined with an identified serviceneed and the increasing pressures on breast radiologists. [1,2] Method: A grounded theory research design was used to interview six APRs individually in a single breast unit, followed by a focusgroup with five of the APRs to acquire additional data and explore the identified themes using the Burnard constant comparative approach. [3] Results: Five core themes emerged from the data; role of the APR, patient experience, efficiency, role boundaries and delivering results. The findings demonstrate the extension of the role is found to be within the scope of practice for APRs, providing they obtain the appropriate training and skills to deliver breast-biopsy results. Advanced practice experience is considered essential with the potential to benefit the radiography profession. Patient expectations were carefully considered along with communication difficulties. The SPIK ES communication model is taught through NHS Connected Advanced Communication Skills Training and a requirement of the NHS Breast Screening Programme. [4,5] There is potential to improve efficiency within the breast-screening service; however emotional impact requires consideration. Effective implementation has successfully changed practice within this unit and may be adopted by other NHSBSP units to address service-need. Further guidance is needed from professional bodies. Overall, with appropriate training and peer-support APRs feel able to effectively deliver results with a patient-centred approach. [1]. Department of Health. 'The NHS Cancer Plan ', 2000. pp. 1-98. [2]. Royal College of Radiologists. Clinical radiology UK workforce census 2015 report.
[3]. Burnard P. A method of analysing interview transcripts in qualitative research. Nurse education today. 1991 Dec 1;11 (6) The multi-centre, multi-modality, BRAID Trial includes an evaluation of the benefit of Contrast-Enhanced Mammography (CEM) for women with radiologically dense breasts. Imaging equipment used for breast cancer screening must undergo periodic testing to ensure that it is performing within expected parameters. These tests are normally defined in guidance published by PHE and IPEM however these are not yet available for CEM. It was necessary therefore to develop a physics and user quality assurance procedure to monitor the CEM systems, ensuring that image quality and radiation dose performance remains constant throughout the study.
Following the spirit of the current national standards for screening [1][2][3] and informed by published research [4][5], quality assurance protocols were written for physics and user testing. Both protocols outline measurements performed using a test object containing iodinated details of varying concentrations to assure that system is acquiring images with expected levels of signal-to-noise-ratio and detail conspicuity. For this purpose, medical physics teams at several of the study sites have purchased dedicated CEM phantoms (CIRS Inc., Norfolk, VA) to perform the QA while radiographers have been provided with laminated test objects created by Schofield et al. While initially prescribed by the lead site, these QA and QC protocols are working documents to be improved upon over the course of the study based on feedback from local physics teams and the utility of the data collected. It is hoped that this work across multiple sites and vendors will inform CEM QA and QC protocols which can be adopted nationally. [1]  Background: Most MR radiomics studies to date have used "pure" datasets accrued from single-vendor, single-field-strength scanners. This does not reflect the ultimate generalisability of AI models and there is no way of determining how such models will be for input data from different sources. Methods: 156 patients with pathologically proven breast cancer underwent multi-contrast MRI prior to neoadjuvant chemotherapy and/or surgery. From these, 92 patients were identified for whom T2weighted, diffusion-weighted and contrast-enhanced T1-weighted sequences and clinicopathological variables were available. Regions of interest were drawn on the images, and from these semantic and calculated radiomics features were derived. Model fits were generated for four different types of classification model (support vector machine, random forest, extreme gradient boosting and naïve Bayes) to predict pathological nodal status. In parallel work, survival modelling was performed using random survival forests. Conclusions: Potentially useful radiomics signatures for disease classification may be derived from heterogeneous "real world" input data, despite clear confounding information being present. [1]. Papanikolaou, N., C. Matos, and D.M. Koh, How to develop a meaningful radiomic signature for clinical use in oncologic patients. Cancer Imaging, 2020. 20: p. 1-10. [2]. Zwanenburg, A., et al., The image biomarker standardization initiative: standardized quantitative radiomics for high-throughput imagebased phenotyping.Radiology, 2020. 295(2): p. 328-338. Background: The Micrima MARIA® system is an ultra-wideband (3-8GHz) radar imaging system sensitive to electromagnetic permittivity contrast in breast tissue. Using a hemispherical array of 60 antennas, 1770 channels are measured over 101 frequencies and focused to form a 3D image of scattering. This phantom study investigates the feasibility of focused spectra for classification of lesions. Methods: Two lesion classes were included; benign cyst, modelled as water-filled glass spheres (diameter = 1-2 cm, relative permittivity ε r = 78 @ 3GHz) and malignant lesion, modelled as spiculated ovoids consisting of polyurethane, graphite, and carbon black (diameter = 1-2 cm, ε r = 29 @ 3GHz). Lesions were imaged on two MARIA® systems while embedded in an analogue of adipose breast tissue (liquid paraffin and water emulsion) at multiple locations in 3 breast volumes (460cc, 700cc, and 1000cc). In total 288 images were acquired. Focused spectra from lesion ROIs were analysed to assess classification performance. Results: Multivariate analysis of variance of 10 frequency bands indicated statistically significant effect of lesion class, Pr(>F) <2.2e-16, with strongest effect for the 3-3.5GHz band. Leave-one-out crossvalidation of ROI classification based on thresholding the mean value of this band achieved sensitivity 0.75 (0.67-0.82) and specificity 0.79 (0.71-0.85). Conclusions: Results indicate that focused spectra provided by MARI A®, particularly in the 3-3.5GHz band, are sensitive to differences in breast lesion characteristics associated with the likelihood of malignancy in vivo. The response in this frequency band when used as a decision variable yields promising classification performance, warranting further clinical investigation.

P05
Coding accuracy following the introduction of electronic coding: a single-centre experience from a breast imaging department Background: Accurate coding of procedures is essential for reimbursement within radiology departments. Under a new system within the Imperial College Healthcare NHS Trust, income will soon only be generated if procedures are coded within the Cerner© electronic patient record (EPR), with radiologists and advanced practitioners responsible for inputting the code. However, issues related to familiarity with EPR and time pressures within clinics can affect the accuracy of coding. The aims of this study were to investigate the rate of non-coding for biopsy procedures, and subsequently how much income would be lost through non-coding. Methods: All biopsies performed during a sample month of August 2019 were identified using the weekly MDT lists, which captures all specimens sent to histopathology. Ultrasound guided core biopsies were included, and the EPR was checked for coding of the biopsy. Coding accuracy was cross-checked against the Radiology Information System (RIS). Results: 97 ultrasound guided core biopsies were included in the study. 7.2% were axilla biopsies, 3.1% biopsied both breast and axilla, and 89.7% were breast biopsies. Only 16.5% of biopsies were coded on EPR. Cost analysis revealed £11,826 loss of income in one month due to procedural non-coding. Income gained from biopsy coding was £2336. Conclusions: Overall a large volume of income was lost through non-coding. Many radiologists find the current method of electronic coding cumbersome and time-consuming during clinics, which is likely contributing to low coding rates. An automated system linking RIS entries to EPR would generate a huge cost saving for the department. BreastCheck is the national breast screening programme in the Republic of Ireland. In late 2015, age-range extension was launched for women ≥65 years. A small number of these women had never been invited before, perhaps due to immigration/return to Ireland or recent self-registration. However, the majority had previously been invited for screening so the age extension resulted in a higher number of subsequent and a small number of initial women ≥65 being invited. Data is routinely collected on all women invited and screened through BreastCheck [1] . The current age range extension is being rolled out on a phased basis. By 2021, all women aged 50-69 will be invited. In 2017-18 608 women aged 65+ were invited for the first time, with 87 women screened (14.7%). 19,615 older women were invited for subsequent screening, and 16,650 attended (84.9%). Among these 16,650, the recall rate was similar to women aged 55-64; the benign biopsy rate (BOBR) was lower than for any other age groups, while the cancer detection rate (CDR) was higher ( Table 1). * Per 1,000 women screened Early results show good uptake and, as anticipated, higher CDR and lower benign biopsy rate among women in the age extension. As BreastCheck has a two year interval, CDR is lower than the corresponding NHS figures with a three-year interval. [2] [ Background and Objectives: National [1,2] and international guidelines [3][4][5][6][7] require women diagnosed with breast cancer at an early age to undergo numerous annual mammograms beyond diagnosis until screening age. This despite younger women having dense breast tissue [8] with reduced mammographic sensitivity for detection of abnormalities [3] , and the lifetime risk for developing radiationinduced cancer being highest in younger women [9] . A current UK trial (Mammo-50 [10] ) investigates targeted surveillance for patients over 50 years, but there is no known trial seeking change for younger women. Ultimately: Could there be a targeted approach for follow-up of younger age women breast cancer groups? This study aims to investigate how many younger women are affected by nontargeted screening. Methods: Younger women under 45 years who have undergone more than 5 years of annual mammograms are investigated for: grade of cancer diagnosis, family history, breast tissue density, whether original cancer was mammographically occult, number of mammograms since diagnosis, and whether discharged from annual surveillance (including reason).

P08
Results: UK-wide results suggests 9% of new breast cancers were diagnosed in women below 44years in 2014-2016 [11] . Sensitivity of surveillance mammography for detection of recurrence was 64-67% with a specificity from 85-97% 3 , while sensitivity was reduced in Background: Chest wall perforator flaps (CWPFs) are a volume replacement technique which permits breast conserving surgery in selected cancer patients where mastectomy is the only alternative [1] . CWPFs are predicated on a blood supply from the lateral chest wall behind the breast. These flaps are based on a perforating artery which courses through the soft tissue into the sub-dermal plexus to perfuse the tissue. Examples includes the lateral intercostal and Lateral Thoracic artery perforators (LICAP and LTAP respectively) [2] . The origin, distribution and calibre of these perforator vessels is mapped both pre-operatively and intra-operatively using hand-held acoustic doppler but can be ambiguous. We therefore aim to determine the potential role of MRI in mapping the vascular anatomy to aid surgical planning.
Methods: We retrospectively reviewed our entire cohort of breast cancer patients who had CWPF surgery. We conducted radiological review of breast MRI performed as part of routine pre-operative imaging to determine perforator anatomy.
Results: Of 43 patients who underwent CWPFs, 22 (51%) had MRI imaging of their breasts prior to surgery. Evidence of perforator vessels on MRI were noted and formal analysis of our results will be completed and presented in full at the symposium. Conclusion: Although cross sectional imaging is not a pre-requisite for performing CWPFs a significant proportion of breast cancer patients already undergo breast MRI for oncological purposes. MRI may provide useful additional information regarding the origin, location and calibre of perforator vessels to aid in surgical planning for these flaps. [1]. Soumian S, Parmeshwar R, Chandarana M, Marla S, Narayanan S, Shetty G. Chest wall perforator flaps for partial breast reconstruction: Surgical outcomes from a multicenter study. Arch Plast Surg. 2020;47 (2) Purpose: To share our unit experience in using the Radiofrequency Tag, (RFT) with a view to replace the conventional wire localisation and improve patients' preoperative experience. Methods: From August 2019 to January 2020, 12 patients with 13 non-palpable breast lesions (one was bilateral) underwent preoperative lesion localisation using the RFT (LOCalizer™). The device consists of a small radioactive free tag with unique number and a polypropylene cap to prevent migration, a 12gauge needle applicator, a radiofrequency reader and surgical reader probe, which is an 8mm pencil size allowing for small incisions. Results: 10/13 lesions were found to be invasive cancers, 1/13 DCIS and 2/13 B3 lesions. 46.2% were micro-calcifications and 53.8% were masses with average mammographic size of 11.5mm (5-20mm). The tag was placed satisfactorily in 12/13 lesions. For 1 lesion a second tag had to be inserted due to misplacement of first one. 11/13 lesions underwent wide local excision and 2/13 excision biopsies. 3 intraoperative re-excisions were done in 2 patients. The average weight of the specimen was 38.75. Average time of operation was 17minutes (4-35 minutes). Average ease of operation was 4.7 (3 to 6). Only 1 patient needed re-excision on a later day (7.8%). The average time between localisation and operation was 2.2 days (0-6 days).
Conclusions: Despite small number of cases, RFT is an accurate, easy and wire free technique which can be inserted up to one month prior to surgery and can potentially be used as an alternative to wires and improve patients' experience. Background: To validate the feasibility of training mammogram readers in FAST MRI interpretation [1] we developed display software (MedXViewer) containing ground truth information for each FAST MRI case to train and assess mammogram readers from NHSBSP sites.

Complication rates following vacuum biopsy and implications for lesion localisation
Methods: A per breast analysis of the frequency of the results against the true outcome was obtained overall and for each reader. Differences in accuracy, sensitivity and specificity across reader groups (group 1 = mammogram readers experienced in breast MRI interpretation; group 2 = mammogram readers with no previous experience in breast MRI interpretation) were analysed using a multilevel generalised mixed model to account for multiple readers per case. Results: 37 NHSBSP mammogram readers (17 in group 1 and 20 in group 2) attended the training day (6 NHSBSP sites) and completed the reading task of 125 cases (250 breasts) (total=9250 reads). The 83% accuracy achieved by Group 2 (4129/5000 (95% CI 82-84%)) was significantly lower than that by Group 1 (3814/4250 (90%; 89-91%); p<0.0001) but differed by only 7%.
Accuracy improved for the group 2 readers from the first 55 cases to the remaining 70 cases (p=0.02), whereas there was no significant improvement for the expert readers of group 1 (p=0.81).
Conclusions: This study validates the feasibility of training mammogram readers to interpret FAST MRI with only one day of training. Improvement in performance with experience by the group 2 readers indicates a learning curve and suggests the performance gap between the two groups might be narrowed by further training. [ Background: Breast cancer screening using mammography has poorer sensitivity in dense tissue. [1,2] Retrospective studies have shown higher sensitivity of Molecular Breast Imaging (MBI) in this group. [3,4]  Background: There is a lack of UK guidelines regarding the most appropriate age for symptomatic mammography [1,2]  Purpose/Background/Objectives: To explore variation in current reporting and decision-making strategies nationally for recalls within breast screening services. To seek to correlate findings with unit performance, based on specific criteria from published national service data (KC62), and produce recommendations for future effective use of arbitration processes. To explore the role of Artificial Intelligence (AI) in this setting.
Methods: A mixed-method, explanatory sequential study with methodological integration of two national surveys (49 Units), qualitative interviews with professionals in varying roles (n=21), and specific performance metrics (KC62). Thematic analysis, including triangulation to compare and contrast data.
Results: Four main themes were identified: (1) Service Variation (2) Culture/implementation climate (3) Planning the service/standardisation (4) Task shifting and PHE arbitration guidance (5) Artificial Intelligence Conclusions Key recommendations: 1: Blinded double reading recommended to obtain the best insight into individual reader performance and standardisation of practice. 2: Blinded arbitration (anonymisation of the reporter) -to obtain independent non-biased opinions. 3: Careful selection of arbitrators -BSIS data to support delegation of solitary third reader arbitration/consensus leads. 4: NBSS updates -to improve usability, enable optimisation of consensus groups and to facilitate true blind reading/arbitration. 5: Consider centralisation/independent arbitration (but internal to the NHSBSP) -may normalise arbitration and provide sufficient arbitrations per individual to allow more accurate performance monitoring. 6: AI could potentially tackle some of the current challenges in breast screening, including improved accuracy of detection, increased efficiency, and advance detection of early cancers. [1,2,3] Further research is needed on optimising human/AI decision-making.  Background: Mammography is associated with pain/discomfort and this is mainly due to positioning and the compression applied to the breast. [2] The aim of the research is to develop an evidencebased protocol that may help reduce pain/discomfort. The angle of image receptor (IR) on the mediolateral oblique (MLO) projection plays a vital role in the distribution of pressure through the breast. When the IR angle is perpendicular to the sternum during compression, there should be an even pressure balance and increased breast coverage. [1] Method: A phantom study was conducted on a model torso with breast attachment. A digital inclinometer was used to take the angle of model's sternum before it was positioned for MLO. Xsensor pressure mat was secured to the surfaces of the compression paddle and IR to read and record pressure distribution applied on the breast phantom. Compression of 10daN was applied to breast phantom and pressure readings and breast foot print were recorded with the IR at various angles in the multiples of 5 from 40 0 to 70 0 . Numerical pressure data recorded on the mat was transferred onto excel and analysed.
Results: IR angles at 55 0 to 65 0 produced a more even pressure and area balance. The recorded sternal angle of model was 60 0 .Conclusion: When the IR angle is parallel or close to the angle of the sternum, there is an even distribution of pressure and area. A study in human female volunteers using this method is in progress. [1]. Background: Regular breast self-examination (BSE) is a simple, costeffective way of early detection of breast cancer, particularly for women outside the eligible age bracket for mammography and for interval cancers. [1] BSE is routinely recommended but compliance is generally lower than desired and highly variable. [1,2] The aim of this study was to identify socio-demographic, lifestyle and health-related factors associated with BSE practice among women in Ireland.
Methods: The Irish Longitudinal Study of Ageing (TILDA) wave 4 was used; TILDA collects information on all aspects of health, economic and social circumstances from people aged 50 and over in a series of data collection waves once every two years. Background: Adjuvant endocrine therapy (AET) substantially reduces risks of recurrence and mortality in women with ER-positive breast cancer. However, adherence is poor. No effective interventions to support adherence exist. SWEET is developing and evaluating an intervention to improve adherence, improve health-related qualityof-life (HRQoL) and reduce long-term recurrence. Method: SWEET is split into six work streams (WS) spanning six years. WS1 will iteratively develop a person-centred, evidence-based, theoretically informed intervention to support adherence. Including: a tailored consultation with a trained health professional; an app/website including a symptom monitoring tool and other support mechanisms; a three-month follow-up consultation; and regular email/text contact. WS2 will assess the interventions feasibility and acceptability. WS3 will deliver a RCT, with internal pilot and process evaluation. 1018 women at medium/high risk of recurrence will be randomised to usual care or intervention+usual care. Adherence and cancer specific HRQoL will be measured at 6, 12 and 18-months. WS4 will assess intervention cost-effectiveness. WS5will use theory, qualitative research and stakeholder involvement to inform potential NHS scaleup. If WS3 improves adherence, WS6 will assess effectiveness in reducing recurrence at five years. PPI will be integral throughout.
Results: Initial work started in May 2020 and is focusing on intervention design and development, using remote approaches to engage patients and health professionals in co-design.
Conclusion: This is the first trial of an intervention to improve AET adherence which is powered to detect effects on recurrence. It offers real potential to reduce breast cancer recurrences and deaths thereby benefiting patients, the NHS and society.

P22
Could psychological and perceptual learning based theories have the potential to transform mammography interpretation Purpose: The interpretation of a mammogram is a complex visual discrimination task that incorporates both a cognitive and perceptual element to the decision-making process. Current learning strategies involve exhaustive review of thousands of mammograms until levels of expertise are acquired. Recent advances in learning sciences suggest the potential for improving medical learning and performance. Often neglected, perceptual learning is a fundamental contributor to expertise. Difficult to teach pattern recognition skills can be systematically accelerated using techniques of perceptual learning. Aside from the "practice makes perfect approach" cognitive science offers a theoretical platform from which to formulate meaningful experiments that could lead to novel training strategies to improve the accuracy and efficiency of training.
Method: Learning theorists have demonstrated the benefits of training on easy perceptual discrimination tasks followed by more difficult cases known as the transfer along the continuum. Novice participants will be divided into two groups. Mammograms are digitally displayed according to the level of diagnostic difficulty cascading from easy to hard cases and hard to easy cases. Decisions of normal versus abnormal in a post-test set of 60 mammograms are recorded. Results could show a benefit of training when obvious features are learned first followed by less salient features that require more effort and an extended period of learning. Further interventions that include re presentation of learning items after a predetermined interval of time or spaced repetition is planned.
Conclusion: Acquiring expertise in mammographic interpretation could be accelerated if mammograms are displayed in a specific order.

P23
Correlation between the shape and size of grade 3 (G3) invasive ductal carcinoma ( The study demonstrated screen-detected G3 IDC were most commonly ill-defined at time of diagnosis, while the expectations were to find rounded masses as the most common once. The maximum diameter at time of presentation was between 10-30 mm. The highest incidence of G3 cancer was in 61-70 age group.

P24
To determine the radiological and histological characteristics of biopsy proven DCIS upgraded to invasive cancer after surgery Results: Of the 136 women diagnosed with DCIS on biopsy, 20 cases were upgraded to invasive cancer post-surgery. 75% displayed micro-calcifications ,20% had mass-like morphology, and 5% had architectural distortion. Equal proportions of comedonecrosis (50%) and cribriform pattern (50%) was seen on biopsy. 90% of invasive malignancies were non-specific type malignancies (NST), followed by mucinous (5%) and mucinous papillary cystic (5%). Majority of cases (50%) were invasive grade 2. No nodal spread and only one case of vascular invasion was recorded. Majority of invasive cancers were ER (85%) and PR (60%) positive, with 15% HER-2 positive. Conclusion: Unfortunately, our results showed no reliable factors to enable us to identify likely cases of invasion amongst DCIS-diagnosed patients. Similar results have been obtained by other studies which failed to identify common characteristics for upgrade [1,2,3 Purpose: To investigate characteristics of interval cancers (ICs) and compare them with characteristics of screen detected cancers (SDCs) to add to current evidence and draw inferences on implied prognostic outcomes.
Methodology: This was a single centre study, using a non-probability sampling technique to identify the ICs. ICs were age matched with a SDC group and inclusion and exclusion criteria applied. Data on specific characteristics were collected. ICs were sub categorised into true intervals, minimal signs cancers, false negative cancers or occult cancers. Results: There were 106 patients in each group. Forty one percent of ICs presented in the third year after screening. Twenty-nine percent of cancers found in the first year were mammographically occult, 67% of these were lobular carcinomas. Upon further classification of ICs into sub categories the occult group had a larger percentage of dense breast tissue (31%) compared to the true interval group (5%). Thirty two percent of the SDCs were < 10 mm in tumour size compared to 9% of the ICs. Fifty four percent of the ICs were > 20 mm compared to only 12% of the SDCs. We found positive lymph node status in the IC group in 24% of the cases and 6% in the SDC group. Conclusion: The incidence of ICs in the third year after screening suggests that a 2 yearly screening programme in the UK may be beneficial. The IC group had less favourable prognostic features in comparison to the SDC group in line with previous studies. Background: Radiographic practice in the NHS breast screening programme is rigorously monitored. The acceptable standard for repeat rate (RtR) at screening is <3%. Repeat images at assessment are not recorded and the rate unknown. Also, there is little knowledge of communication/decision making between radiographer and responsible assessor (RA) during the assessment process.Methods: Questionnaires were developed for stereo/vacuum biopsy assessment cases (S/VBCs) and non-stereo assessment cases (NSCs) where radiographic repeats were taken. The survey covered local radiographic practice including abnormality identification, location, imaging required and associated interactions with the RA. The number of repeats per case was requested including contributory factors. Questionnaires were disseminated to a small number of NHS breast screening units (BSUs) as a pilot.Results: Between 5.8.19 and 15.11.19 3424 assessment cases were undertaken in 10 BSUs. For NSCs involving repeats, 90% radiographers identified the abnormality, location and imaging required from documentation on client sheet and/or imaging. In 62% cases there was no discussion with the RA and 74% radiographers decided autonomously to undertake repeats. For S/ VBCs 32% radiographers identified the abnormality, location and imaging required from documentation but 58% discussed this with the RA. In 82% S/VBCs it was the RA decision to repeat. RtR was ≤3% in 70% participating BSUs. Sixty percent repeats were NSCs, 40% S/ VBCs; difficult abnormality location/client habitus contributed.Conclusion: RtR is higher at assessment compared to routine screening relating to a more complex case mix. Consultation with RAs correlated with a lower repeat rate and is recommended for all Objectives: NHSBSP QA standards [1] state that small cancers should account for 55% of the overall cancers detected. We aim to determine if small cancer detection rate can be improved by reviewing the characteristics of cancers arising from discordant read (where one film reader did not identify the lesion). A secondary outcome will be to develop a case collection of images that demonstrate subtle findings associated with discordant reads for training purposes.

P26
Methods: Review of all screen detected discordant cancers from 01.04.2018-31.03.2020. Each case was reviewed for mammographic abnormality, lesion size, breast density, one/two view presentation, prevalent/incident read, histological subtype, hormone and lymph node status, reader experience and the results recorded.
Results: Of 104 screen detected cancers diagnosed from a discordant read, 65(63%) were small cancers. Lesion characteristics were: spiculate/ ill-defined mass (54%), microcalcifications (38%) and other (8%). 70% were invasive,the majority G2 NST. Lymph nodes were involved in 11/104 cases. The majority of discordant cases (70%) were detected by readers with >5 years of experience. 72% of the mammographic abnormalities were demonstrated on both views. 69% were identified in mixed fatty dense breast tissue. Conclusions: Our results highlight the importance of double reading and film reader experience in maximising small cancer detection rate [2] .This has implications for which film readers should be paired and the importance of having experienced film readers undertaking arbitration/ consensus. [1]  Despite NHSBSP [1] guidance, Breast Care Nurse (BCN) resource at assessment may be erratic, service evaluation aimed to demonstrate BCN consult effect on anxiety levels during assessment, reinforcing BCN importance. Aims: Demonstrate anxiety level differences in women, dependent on BCN consult via repeated measures using Spielbergers State Trait Anxiety Index [2] (STAI) questionnaire. Assess BCN consultation and therefore value to clients via patient questionnaire. Methods: Mixed methods single centre study recruiting 57 women attending assessment clinics over three month period. State/Trait anxiety levels measured on arrival using full STAI form, followed by repeated measures State anxiety using 6 question short anxiety questionnaire [3] . Responses to questions are scored and totalled, differences between scores compared. Women placed in 2 groups dependant on BCN availability. Group A had pre assessment BCN consult, group B did not. Measurements of anxiety from all women taken on arrival and post assessment. Group A had additional measurement post BCN consult. Comparison of scores between groups undertaken using Wilcoxon signed rank test. Results: Group A -State anxiety significantly decreased post consult compared to baseline, 6.7% (p = 0.036). Both groups -State anxiety significantly reduced post assessment; A-20.7%, (p = 0.002), B-17.7% (p = 0.008). A demonstrated 3% greater reduction than B. Biopsy increases state anxiety post assessment, independant of consult. Group B-60% of women asked wished to have seen a BCN before assessment, highly significant, p=0.003. Conclusions: BCN consult significantly reduces anxiety in assessment women, therefore all women attending assessment would benefit from BCN consult, as recommended in NHSBSP guidance. Purpose: As per NHSBSP guidelines [1] , departmental protocol is to perform Vacuum Assisted Biopsy/Excision (VAB/E) as a second line biopsy. This is following all B3 stereotactic core biopsies and for other selected cases after MDT discussion. The impact of VAB/E on patient outcome and non-operative diagnosis rates was reviewed over 5 years. Method: All patients undergoing VAB/E during the 5 year period were identified. The 14-gauge core biopsy results and the follow-up 9-gauge VAB/E results were compared. The surgical or non-surgical outcome for these patients was recorded. Results: 113 patients underwent VAB/E: 94 VAB and 19 VAE. Prior to VAB/E 81 had initial B3 core biopsies and 12 B4. Post VAB/E 50 were returned to routine screening. 29 required annual mammographic surveillance, 21 women were upgraded to B5a or B5b and only 4 required diagnostic excisions in theatre. The screening unit's nonoperative diagnosis rate for invasive cancers remained within 1%. However, the non-operative diagnosis rate for non-invasive cancers increased from 79% to 95%. Conclusion: We cannot say how many of these patients would have had a different outcome if the standard 14-gauge stereotactic core biopsy had been repeated. However, all of these patients would have had surgery if no second line biopsy was performed. 70% of patients were reassured with a benign result or increased surveillance. Noninvasive non-operative diagnostic rates have improved since the introduction of VAB/E in this screening unit. [ Case report describing management of a large lactational breast abscess during the height of the Covid-19 outbreak utilising Vacuum Assisted Biopsy (VAB) device to aid aspiration, thus avoiding surgical Incision and Drainage. Background: Abscesses are commonly treated with antibiotics and aspiration in an attempt to avoid surgical intervention if possible. The purpose of treatment is the removal of infection and pus as rapidly as possible to avoid further complications [1]. This case report describes the process of VAB aspiration of a persistent large lactational breast abscess. Surgical colleagues were unwilling to take the patient to theatre due to Covid-19. Presentation: A breast feeding woman in her twenties, 15 weeks post-partum was admitted with left UOQ breast lump and accompanying pink skin with a swollen warm firm breast consistent with lactational abscess [2]. Patient was systemically unwell, spiking temperature, unable to feed/express from symptomatic breast. Administered IV antibiotics (flucloxacillin) with analgesia. US demonstrated a 100x60mm collection, repeated aspirations had variable success with needle blockages problematic. To avoid surgery, VAB aspiration under US guidance was performed. Conclusion: VAB successfully aspirated over 200ml of material from the abscess which facilitated patient recovery. This was the first time our institution had utilised VAB in this manner. Literature search found 2 studies that utilised VAB in this way, both concluding VAB to be a safe and viable alternative to surgical intervention in the case of larger (over 5cm) abscesses [3]. This new procedure has been audited to assess outcomes, highlight complications and guide safe expansion to vacuum excision as a service improvement. Statistics were collected, including lesion size, distance from nipple and skin, number of cores, biopsy results, final histology and outcomes. Results of the 70/76 cases that were second line biopsies have been analysed. 16% (n=11) of cases were fibroadenomata, benign breast change or sclerosing lesions. 30% (n=21) were papillomata with no atypia and avoiding surgery. 27% (n= 19) were upgraded to cancer pre-surgically. 27% (n=19) were other non malignant lesions of which 5 cases were ADH or papilloma with atypia. 11% had less than 6 cores, done in the early years or were halted procedures; 68% had 7-17 cores, and 21% had 18+ cores, now done as the routine minimum. Five complications were minor only of pain and haematoma; only one of these was 0.5cm from the nipple and 0.8cm from the skin. Sizes ranged from 3-80mm 65% were well defined masses. Conclusion: 25% of cases became cancer ensuring therapeutic surgery and 67% cases were discharged avoiding excisional surgery. Safe distance from the nipple and skin are important considerations for this procedure. Ultrasound VAB procedures are now routinely undertaken, for lesions easily identified, reducing the need for stereotactic approach. VAE is slowly evolving, protocols in place including moving to 7g. The poster will illustrate the safety and accuracy of this efficient procedure. 15/26 were cysts. In 11/26 patients a biopsy was performed which identified 6 FAs and 5 extra foci of cancer. In 9/11 biopsies, the lesion could be identified on the initial mammogram. 2 lesions identified solely on WBUS were benign following biopsy. Despite WBUS, 16 patients still required a 2nd look US following MRI scan. In 10/16 patients this was of the contralateral breast. All lesions were identified and biopsied and malignant in 4/10. 7/16 patients required a 2nd look of the ipsilateral breast, of which 5/7 were malignant on biopsy. One of these patients had multifocal cancer diagnosed despite an initial normal WBUS. Conclusion: 14 extra foci of cancer were identified in 249 patients with symptomatic breast cancer. These were all identified by mammography or MRI. WBUS did not identify any additional malignant lesions not seen on other modalities. WBUS does not add any diagnostic value for multifocal breast cancer diagnosis; it can be falsely reassuring and increase unnecessary biopsies. Background: Magnetic resonance imaging (MRI) is known to be the most sensitive imaging modality for examining the breast. Due to it's low specificity breast MRI is often used as a problem solving tool, to examine those at high risk of developing breast cancer or to monitor those on neo-adjuvant treatment regimes. Targeted breast ultrasound is the imaging modality of choice in this centre to further evaluate additional lesions detected on breast MRI. Although playing an essential role in the development of a treatment plan-these additional imaging stages inevitably slow down the diagnostic process, and there is a feeling that referrals for breast MRI may be increasing unnecessarily. An audit was performed to establish the frequency and outcome of 2nd look ultrasound following breast MRI in those with an already diagnosed breast malignancy, as well as the clinical indication for breast MRI. Methods: A retrospective audit was performed of all breast MRI performed in this centre within a 2 year period on women with at least 1 biopsy proven malignancy. The referral type was plotted against those recommended by the European Society of Breast Imaging (ESOBI) [1]. Those exams requiring 2nd look ultrasound were reviewed with ultrasound results and additional biopsy rates noted. Results: 87 breast MRI were reviewed from the 2 year audit period. 19 were referred for 2nd look ultrasound, with 12 requiring additional biopsy. 4 new malignancies were identified. No women were referred for MRI biopsy. Referrals for breast MRI were in line with those from ESOBI guidance. Background: Mammography is the gold standard for investigation of breast abnormalities but is limited by overlapping tissue obscuring real lesions or mimicking malignancy [1,2]. The traditional 90°lateral projection and Digital Breast Tomosynthesis (DBT) provide similar information when investigating possible lesions identified on standard mammography. This study aims to investigate the frequency of use and usefulness of the 90°lateral projection in breast screening assessment clinics where DBT is available. Methods: A self-administered questionnaire was designed and piloted. Clinicians within a screening organisation were invited to complete one questionnaire per case in assessment clinics during a 6-week period. Results: Twelve clinicians from two of the three invited regions participated. 231 questionnaires were included in the dataset. Lateral projection and DBT were used frequently, in 81.8% (n=189) and 83.5% (n=193) of cases respectively. They were used to complement each other as evidenced by varied indications. Lateral projections and DBT were reported 'very useful' or 'useful' in most cases, 65% and 79.3% respectively. Conclusions: The lateral projection remains frequently used when DBT is available. Radiation dose of performing both is justified by varied indications for use and both were reported useful in most cases. Increased use of the lateral DBT projection could combine the benefits of an orthogonal projection to indicate lesion location and those of DBT in characterising lesions. Purpose: In 2018 digital breast tomosynthesis (DBT) equipment was installed in the department. Published literature claimed DBT a superior imaging technique for perioperative specimen imaging [1] . A combination of DBT and full field digital mammography (FFDM) was introduced and audited. Method: All surgical specimens in a 9 month period were imaged using FFDM/DBT or in theatre using a specimen x-ray cabinet (SXC). Cases were separated into 3 month groups, before, during and after equipment changeover. All FFDM/DBT images were reviewed by the radiology team and verbally reported to theatre. Data regarding imaging method and specimen margin status was collected. Results were discussed within the multidisciplinary team and a further 3 month period of data collection was agreed upon. Results: During the initial 9-month audit 236 specimens were imaged. 40 needed surgical re-excision. 183 specimens were imaged using FFDM/DBT and 53 using SXC. The SXC images showed a higher percentage of re-excisions (26.42% versus 14.2% in FFDM/DBT). Post introduction of DBT (88 specimens) the re-excision rate for FFDM/ DBT fell from 15.9% to 12.7%, suggesting DBT may be beneficial, Post MDT discussion showed a reduction of specimens imaged using SXC, (21.7% to 13.6%), due to imaging method selection by surgeons. Conclusion: There has been a reduction in re-excision rates since the introduction of DBT. Due to significant differences in re-excision rates between FFDM/DBT versus SXC, auditing has prompted a change of surgical practice and imaging protocols. However, non-calcifying DCIS could be a confounding variable to this Purpose: Axillary lymph node metastasis is seen as a key prognostic factor for breast cancer patients. [1] Pre-operative diagnosis of axillary lymph node metastasis can ensure patients receive the appropriate axillary surgery and can prevent the need for further surgery. [2] This study assessed the accuracy of ultrasound, MRI and ultrasound guided core biopsy in diagnosing axillary lymph node metastasis pre-operatively. The results will aim to refine our current clinical practice. Methods: Ultrasound and MRI data was retrospectively analysed from breast cancer cases diagnosed between January 2017 and December 2019. The results were correlated to the final histological outcomes from the surgery. Results: Two hundred and fifty eight cases were included in the study, 107 (41.5%) had evidence of lymph node metastasis on final histology. Ultrasound was compared to MRI to establish which imaging modality was most accurate at detecting lymph node metastasis. Ultrasound was demonstrated to have a sensitivity of 67%, specificity of 87%, PPV 79%, NPV 79% and an Accuracy of 79%. MRI was demonstrated to have a sensitivity of 76%, specificity of 84%, PPV 77%, NPV 83% and an Accuracy of 80.6%. 103 of the cases had ultrasound guided core biopsy of an abnormal node to establish lymph node metastasis. The overall sensitivity of lymph node core biopsies was 86%, specificity was 100%, PPV 100%, NPV 68% and accuracy 89.3%. Conclusion: There was no statistical difference between the performance of ultrasound to MRI in the detection of lymph node metastasis. No change in clinical practice. [1] Valente SA, Levine GM, Silverstein MJ, Rayhanabad JA, Weng-Grumley JG, Ji L, Holmes DR, Sposto R, Sener SF. Accuracy of predicting axillary lymph node positivity by physical examination, mammography, ultrasonography, and magnetic resonance imaging. Annals of surgical oncology. 2012 Jun 1;19 (6) Purpose: Patients presenting to the breast clinic with clinically assessed skin lesions are frequently referred for additional diagnostic US scan. 8 months of radiology activity were audited to assess whether US contributed to the management of these lesions. Methods: The reports of US performed in the rapid diagnostic clinic between 1/01/2020 and 31/08/2020 were reviewed retrospectively. A case was included in the audit if the clinical details or the imaging details suggested that the presenting symptom was due to a skin lesion. Patients who presented with a malignant mass involving the skin were excluded. Results: 167 cases were identified out of 4442 patients referred to the breast rapid diagnostic clinic in our institution. Most cases were clinically assessed as due to benign causes. In the large majority of cases there was no further action following US imaging. There was one adverse event: a lesion thought to be an epidermoid cyst at presentation was later proven to be an invasive carcinoma (increasing size). Another skin lesion that appeared very vascular on US in a patient who had previously been treated with radiotherapy was biopsied and proved to be an angiosarcoma. Discussion: US features of skin lesions are reviewed. We discuss the contribution of US and its influence on further management. Background: Axillary ultrasound (AUS) sensitivity in detecting axillary lymph node (ALN) metastases is low with no standard cortical thickness threshold to indicate nodal sampling. [1,2,3,4] The RCR suggests a sensitivity of 50% for AUS in detecting ALN metastases. Nodal disease confirmed preoperatively results in a 'fast-track' to axillary node clearance (ANC). Following a local departmental audit in 2017, we changed our cortical thickness threshold from 2mm to 3mm based on a predicted increase in specificity with only a modest drop in sensitivity. Methods: Retrospective analysis of 100 screen detected and symptomatic breast cancers in 2019 where final axillary staging with sentinel lymph node biopsy or ANC was available. Findings were correlated with AUS and core biopsy results. Those receiving neoadjuvant chemotherapy were excluded. Results: The change in the cortical thickness threshold reduced the sensitivity from 65.7% to 60%, which is still well above the expected RCR standard of 50%. The specificity and accuracy improved from 83.9% to 95.7% and 78.1% to 85% respectively. There were 12 false negative AUS cases: 8 were invasive lobular carcinoma (ILC), of which 5 had low burden disease (1-2 nodes involved), and 4 were invasive ductal carcinoma, all of which had low burden disease. Conclusion: The recommendation is to maintain a cortical thickness threshold of 3 mm, quantify disease burden on AUS and accept a slightly lower sensitivity but higher specificity. Further work needed to investigate whether a 'second look' AUS following breast MRI for ILC is of any value in detecting ALN metastases. Introduction: The use of mobile digital specimen radiography systems (DSR) is advocated following wire localisation of impalpable lesions, to confirm total target excision and reduce re-excision rates [1]. Aim: Compare image quality and accuracy for depicting various targets (mass, distortion, microcalcification, marker clip) with DSR (Faxitron) to conventional digital mammography (CDM, GE) using compression and magnification, as the reference standard.Assess DSR and CDM specimen radiographs in predicting complete excision with histological correlation as the reference standard. Method: Retrospective audit of specimen radiographs taken between August 2018 -January 2020 with DSR and CDM, scored by 3 teams (breast radiologists and experienced film reader) for: image quality, target visibility and complete excision, on a 3-point scale.

Results
Cohort: 56 specimen radiographs depicting 63 targets. Image quality: scored as good or moderate of 85% DSR vs 95% CMD radiographs. Target Visibility: 90% of all targets were 'present' or 'probably present' with both modalities. 10% were not seen with either. 14% were mammographically occult (US visible only or marked by clip, post neo-adjuvant treatment). Some readers scored clips as targets. Complete Excision: Margins were scored as positive on 27.5% of DSR and 23% of CDM radiographs (average for all readers taken). 22% of margins were confirmed positive on specimen histology [2]. Conclusion: Excellent target visibility and comparable margin assessment is achievable with DSR and CDM. Experience and use of image optimization tools can improve confidence in DSR interpretation. DSR, based in theatre, can provide 'point-of-care' assessment of excision specimens. [1]  Background: One stop breast clinics are a high pressure, busy environment, where the likelihood of missing an abnormality is high. Current practice at our DGH one stop clinic is that imaging is reported by a single consultant radiologist or radiographer. The aim of this audit was to identify whether there would be a benefit to second reporting of symptomatic clinic mammograms. Methods: All one stop clinic mammograms during a 6 month period were second reviewed to see if there was any discordance with the original report. If a discrepancy was identified, a third review was performed. If the second and third review agreed there was a discrepancy, further details were obtained regarding further investigations, follow up and biopsy results. Results: 1452 patients underwent mammography in the 6 month time period. 13 discrepancies were identified between the original mammogram report and second/third review. Of these, 4 have not currently been followed up, 2 patients have died, and 7 have undergone further follow up and biopsy. Of the 7 who have had biopsies, 4 of these were found to have invasive cancer, 1 a diagnosis of DCIS and 2 had benign disease. 3 patients had a delay in the diagnosis of DCIS or invasive cancer (range 3 months -14 months delay). Conclusions: Our review suggests double reporting of imaging in symptomatic clinics could reduce error and increase sensitivity. It may lead to a reduction in missed cancers and earlier detection of incidental cancers, a benefit to both patients and the trust.  2010) recommends ultrasound as the primary imaging tool in women aged <40, where a triple assessment approach should be adopted where deemed necessary [ ¹ ] . RCR (2019) states a scale of U1-5 should be adopted to indicate level of concern, U1 representing normal breast tissue and U5 showing a high level of suspicion [ ² ] . With clinical demand increasing into the breast clinic, can imaging referrals be further streamlined in this younger cohort of patients where a normal ultrasound report may be associated with a normal clinical breast exam. Aims To identify the ultrasound classification associated with P1 examinations.
Were these P1 cases true representation of a normal clinical examination.
Methods: The Computerised Radiology Information System (CRIS) system generated a list of women who had attended clinic between October 2018-April 2019 aged between 24-29 who were referred for an ultrasound and assigned a clinical grade of P1. Results: 140 women met the inclusion criteria. Of these 85/140 (61%) were classified as P1. These were further subdivided into their ultrasound classification; Normal (U1) was 74/85 (87%) Benign (U2) 8/85 (9%) and finally 3/85 (4%) (U3) strictly indeterminate. However, 11 cases were not strictly P1, most included lump and skin thickening. Of the true P1 patients 100% had a U1 ultrasound. Conclusion: Where a clinical indication of P1 was assigned this was followed by a normal ultrasound classification of U1 in 87% of cases. Care should be taken in this age group to ensure concordance of the clinical level of suspicion with the ultrasound classification. [1] Willet et al. (2010) Best practice diagnostic guidelines for patients presenting with breast symptoms.
[online] Accessed 27th January 2020. Available from URL: https://www.rcr.ac.uk/system/files/publication/field_publi-cation_files/bfcr199-guidance-on-screening-and-symptomatic-breastimaging.pdf Purpose: To quantify the CT and MRI susceptibility artefact of a 5mm Magseed breast marker constructed of surgical grade stainless steel [1] . This will determine whether a patient with this marker in situ can undergo diagnostic breast imaging. Methods: Phantom testing was done on MRI [2] and CT [3] . Additionally a healthy volunteer was scanned using MRI with the marker on the skin surface. All MRI was performed at 3T using our clinical breast protocol which included spin echo, gradient echo, DWI and dynamic sequences. CT scanning was performed with filtered back projection and iterative reconstruction to compare techniques. Results: On MRI spin-echo sequences the artefact diameter was 4.1cm in-vivo, consistent with the manufacturer guidance 1 . For gradient echo and DWI the artefact ranged from 4.7 cm to 5.8 cm. Fat saturation was unsuccessful with the marker in dynamic imaging resulting in a much larger artefact of 9.5 cm. 3D surface rendering of the in-vivo MRI showed a butterfly distribution of the artefact. On CT the artefact was limited to the size of the marker and remained unaffected by reconstruction technique. Conclusion: Non-breast MRI with the marker in situ can be performed safely while retaining diagnostic quality of images. The presence of the marker in breast MRI causes significant artefact up to 9.5cm from the marker rendering anatomy within this region nondiagnostic. Patients with this marker in situ are unlikely to benefit from an MRI breast exam until after the marker has been removed. CT imaging is unaffected by the presence of the marker. [1]. Endomagnetics Ltd, Cambridge. Magseed for use with Sentimag. Instructions for Use, Page 1. [2]. Inc., Supertech. ACR MRI Phantom -ACR-PH1 -ACR-PHE. supertechx-ray. Background: Breast cancer screening programmes have led to an increase in the detection of non-palpable breast lesions which traditionally have been localised with wire guidance, placed under ultrasound or stereo taxis on the day of surgery. This relies on close liaison with surgical and radiology teams, which with ever increasing Fig. 1 (abstract P46). See text for description. demands on these teams can prove difficult. These difficulties have led to cancelled operations and service delays. Furthermore complications with wires such as the potential for them to move once placed within the breast may lead to incomplete excision. The Magseed is a relatively new localisation technique that can be placed up to 30 days prior to surgery, and has the potential to increase radiology and surgical flexibillity and thus improve the service offered to our patients. Method: The Magseed was trialled within our NHS Trust for a short period in 2019 and here we present an audit of the trial results, with a case study pictorial review. Conclusion: The Magseed localisation device allowed for easy and accurate placement in radiology, with complete surgical margins in all post operative specimens. The increased flexibility with placement time will allow for improved access for localisation and surgery.

P48
Would it be safe to avoid tissue sampling in ultrasound benign breast lesions (U2)  Background: Histopathology is the gold standard for diagnosis of breast masses, but it is an invasive procedure and places considerable stress on a system with limited resources. Best practice diagnostic guidelines for symptomatic breast imaging, widely followed in the NHS, state that benign lesions in women <25 years old do not warrant needling. We carried out this study to determine if this age limit can be safely increased to 30 years. Accurate ultrasound grading is crucial for this; therefore, we also studied grading variability between ultrasound operators within the breast imaging department. Methods: We correlated ultrasound grading and pathological reports for tissue sampling performed on all women <30 years old in our Breast Unit over a year period. Also, a blinded lesion characterisation assessment was completed by breast radiologists who regularly participate in breast ultrasound, to establish if the Stavros criteria applied were accurate. Results: 81 ultrasound-guided samplings (48 core biopsies and 33 fine needle aspirations (FNAs) were performed for women <30 years old (39 U2 and 41 U3 lesions). None of the U2 lesions were malignant on pathology. 11/ 33 FNAs were non-diagnostic and patients were recalled for a biopsy, all of which proved benign. 39/ 41 U3 lesions were benign on pathology. Conclusion: Tissue sampling can be safely avoided for U2 lesions in women less than 30 years old, provided strict ultrasound criteria are adhered to and hence, we propose that the current national age limit of 25 years be reviewed and consideration be given to it being increased to 30 years. Breast carcinomas can take on a multitude of appearances that display the features of a true carcinoma. These being: shape, orientation, echo pattern and posterior features. However, some benign entities can mimic these appearances on ultrasound. This pictorial review demonstrates some of the great mimickers that display the worrying features.
Diabetic Mastopthy -Disease with fibro inflammatory processes of the breast. Features can include hard, irregular hypoechoic mass with posterior shadowing. Fat Necrosis -Known as a benign non-supparitive inflammatory process. This process occurs due to breast trauma. Some of the common causes are: radiotherapy, surgery or trauma. Sonographically, the appearance may display an irregular complex mass, edge shadowing or a hyper-echoic irregular mass. Tuberculosis -The most frequent mode of infection is spread from the axillary nodes. Appearances can present as nodular, diffuse and sclerosing. The nodular type of Tuberculosis can manifest as an ill defined hypoechoic mass. The diffuse type can simulate inflammatory carcinoma, and the sclerosing type can be associated with areas of architectural distortion. Granular Cell tumour -A benign neoplasm derived from perineural scwann cell of peripheral nerves. On ultrasound, the appearances can present as an irregular or ill defined mass with posterior acoustic shadowing. Fibromatosis -This is a benign tumour that can occur in the breast. The definitive eitology is unclear but can be associated with Gardeners' syndrome. On Ultrasound, this can present as an irregular hypoechoic mass with a thick echogenic rim and posterior shadowing. [  Teaching and training are key components of everyday work for most of the breast imagers in order to provide a high quality and well trained work force for the future. They are also often the most rewarding aspects of one's career. Unfortunately, COVID 19 has put an unexpected spin on how training has been delivered over the past few decades. Need to limit human contact meant all 'face to face' and 'being on the shop floor' sessions had to come to an abrupt stop. As the pandemic hit, attention was focused predominantly on essential service delivery whilst training had to be put on hold for a brief time. However, most trainers quickly realised the detrimental effect this was having on an entire generation of work force and started to look for alternative options. A new world of virtual platforms and virtual interaction has started becoming the new norm for delivery of training. The aim of this poster is to highlight various freely available online platforms and wed based resources that can help in to deliver training in a safe, effective and time efficient manner. We share our experience in using the more popular platforms with an emphasis on the pros and cons to help the reader choose the most appropriate ones for their purpose. We also share alternative ideas on how to effectively deliver training in a clinical setting during COVID restrictions. The injectable breast filler Polyacrylamide hydrogel (PAAG) was widely used in China since the 1980s with as many as 300,000 women subjected for cosmesis and reconstruction following cancer . [1] The procedure requires no anaesthesia, often injected by nonmedical professionals. No safety clinical trials were conducted and in 2006 the Chinese State Food and Drug Administration prohibited the clinical application following significant evidence of neurotoxic and teratogenic monomers residual in the synthesis of PAAG . [2] Although now withdrawn, many patients are developing on-going associated complications of PAAG and presenting worldwide to surgeons unfamiliar with the treatment, necessitating complex surgery. Management of PAAG is not standardised and often directed by the radiological appearances. A case study is reported discussing the associated challenges of PAAG with recommendations gathered from the literature. Radiological imaging can mimic malignancy with inflammatory appearances, whilst simulating silicone implants and the features of Breast Implant Associated -Anaplastic Large Cell Lymphoma. Glandular atrophy and encapsulation can develop potentially delaying cancer diagnosis. [3] Surgically, migration of the gel almost always prevents complete removal and often requires extensive reconstructive techniques. Considering PAAG may have potential toxicity and radiological interpretation is significantly compromised, careful assessment is required to understand how best to manage this group of patients often presenting with multiple complications; the long term implications are yet unknown.

P53
A planned feasibility and pilot study to compare two mammographic imaging protocols in women with breast implants Alex Coltart University Hospital Crosshouse, Crosshouse, United Kingdom Breast Cancer Research 2021, 23(Suppl 1):P53 Background: Scant literature on optimal mammographic imaging for women with breast implants exists beyond that published by Eklund et al in 1988 [1] , so there is little evidence on which to base best practice. This study aims to establish the feasibility of a larger study to compare two "Eklund-based" mammography protocols for screening women with breast implants, specifically: 1. whether acceptable levels of intra-rater reliability (single consultant radiographer rater) on the outcome measures can be achieved and 2. participant recruitment rate. Pilot data will be collected to enable a sample-size calculation for a future definitive study comparing the amount of breast tissue demonstrated and the mean glandular dose received by the two mammographic protocols. Methods: The two protocols to be compared are 1. Standard series -medio-lateral oblique (MLO) and cranio-caudal (CC) implant-in-field, plus implant-displaced versions of both projections (current standard practice in Scotland -8 views total); 2. Alternative seriesstandard MLO implant-in-field and MLO implant-displaced, plus implant-displaced medially-and laterallyextended CC (i.e. no standard CC views); also 8 views total. Following NHS Research Ethics Service and management approvals, a repeated measures design will be used, i.e. participants as their own controls. With written informed consent, 20 participants will have the Scottish standard 8-view mammographic series, plus two extended implant-displaced CC views to one breast. The standard versus alternative series of views for the 20 single breasts will be compared for: area of displaced tissue demonstrated, depth of tissue visualised between the nipple and the chest wall, radiation dose.
[1] Eklund, G.W., Busby, R.C. et.al (1988). Improved Imaging of the Augmented Breast. Background: Evidence shows that women with a learning disability are more prone to developing cancer than the general population. [2] A lack of reasonable adjustments can be a barrier to accessing healthcare settings. [1]. It is important that a breast screening service makes adaptations to assist women with a learning disability access screening. Method: A review of a Breast Screening service practice and the adaptations that have been made within the service to facilitate women with a learning disability access screening was carried out. Adaptations included developing a database and improving the process of recording client activity and asking for assistance from general practitioners in identifying women with a learning disability. Changes to the actual screening process were also made. The screening uptake figures for women with a learning disability compared uptake in 2016-17 with the uptake in 2018-19 to see if there had been any improvement. Results: The breast screening uptake for women with a learning disability improved in 2018-19 compared with 2016-17 which indicated that the interventions had been beneficial. The new interventions which were introduced became part of a set of work instructions incorporated into the clinical practice of the unit. Conclusion: In reviewing the adaptations that had been made to facilitate screening for women with learning disabilities we identified measures that appeared to make a difference according to the uptake figures. [1] Ali, A., Scior, K., Ratti, V., Strydom, A., King, M., & Hassiotis, A. (2013). Discrimination and other barriers to accessing health care: perspectives of patients with mild and moderate intellectual disability and their carers. PloS one, 8(8). [2] Heslop, P., Blair, P., Fleming, P., Hoghton, M., Marriott, A., & Russ, L. (2013). Confidential Inquiry into premature deaths of people with learning disabilities (CIPOLD). Bristol: Norah Fry Research Centre. Background: Due to the immense pressure to provide capacity for women with breast symptoms, to be seen within two weeks, a new innovative role has been created to provide increased capacity. Introduction: The breast services see many women with conditions that are benign and easily identified upon ultrasound. The majority of these conditions occur in women under the age of 40years. The role of an advanced clinical practitioner was created to answer a service need. This role requires a highly specialised cohort of skills combining breast image interpretation, breast ultrasound and breast biopsying alongside a range of clinical competences enabling autonomous practice within clear governance. Methods: A new clinic was created for under 40 aged women only requiring only a breast clinical specialist and an advanced clinical practitioner, using ultrasound for assessment. Unexpected findingssuspicious upon ultrasound-would be redirected to the next consultant led clinic for full imaging assessment and biopsy. Results: Increased capacity was achieved, without increased costs. Anxiety levels were reduced due to these patients seen within these clinics and more specialist skills could be directed to more complex cases in the traditional cancer clinics. Conclusion: The use of this specialist role has proven to be innovative and specialised in answering capacity issues within the workforce. The ACP role is utilised as a support to all clinics working alongside consultant radiographers as well as in an autonomous role, thereby freeing up the consultants for cases requiring specialist skills. The stability of the breast service has been ensured. The Problem: The shortage of breast imaging staff is both a national and local issue; in 2016 a Public Health England survey [1] reported a national vacancy rate for breast imaging practitioners of 15%. A Solution? In an attempt to overcome this issue locally, our breast unit has implemented an ongoing training programme to train 9 mammographers in the 3½ year period to September 2021, utilising both the post graduate route for qualified radiographers and the new Mammography Associate apprenticeship. In addition, two experienced mammographers have been supported to qualify as advanced practitioners in mammography, providing additional support for our radiologists. This has required a strategic and collaborative approach, involving all staff groups. In order to ensure that all trainees achieve competency and receive a quality training experience the following strategies have been employed: Full staff engagement Dedicated mentors and practical trainers Locum staff for backfill Close collaboration with the breast screening administrative team to provide dedicated training lists Individualised timetables for trainees to maximise training opportunities and ensure an equitable training experience for all Timetabled study leave -to minimise students feeling overwhelmed Where are we now? At the time of writing 5 trainees have successfully completed their training and a further 2 radiographers and 2 mammography assistant practitioners are due to complete their studies within the next twelve months. This will mean that the unit has a full compliment of mammographic staff to support our expanding service and will allow more experienced staff to develop into advanced practitioner roles. [1] Public Health England. NHS Breast Screening Programme National radiographic workforce survey. 2016. Available from: https://www5. shocklogic.com/scripts/jmevent/Abstract_2.asp?Client_Id= ' P P ' & P r o j e c t _ I d = ' S YM P 2 0 A B ' & C r y p t = U G V y c 2 9 u X 0 l k P T Q 5 ODYyMjYmRm9ybV9JZD0mRm 9yb V9OdW1iZXI9MiZMYW5 ndWFnZV9Db2RlPSZBPQ== [Accessed 07 October 2020].