Skip to main content
Fig. 4 | Breast Cancer Research

Fig. 4

From: A machine learning model that classifies breast cancer pathologic complete response on MRI post-neoadjuvant chemotherapy

Fig. 4

Relative importance of the radiomics features and molecular subtype as selected by the recursive feature elimination random forest (RFE-RF) classifier for the a radiomics only model and b radiomics with molecular subtype model. “af” corresponds to post-neoadjuvant chemotherapy (NAC), “bef” to pre-NAC, and “diff” to difference between post-NAC and pre-NAC radiomics features. “Pre” corresponds to pre-contrast MRI, post1 to the first-post contrast, post2 to the second post-contrast, and post3 to the third post-contrast of the multi-phase DCE-MRI sequence. Gab0 corresponds to Gabor edge feature computed at 0°, while Gab90 to the Gabor edge feature computed at 90°. A bandwidth of 1.414 was chosen for the Gabor textures for all orientations. Feature importance corresponds to the Gini importance measure used by the random forest model

Back to article page