Saliency of breast lesions in breast cancer detection using artificial intelligence

Abstract

The analysis of mammograms using artificial intelligence (AI) has shown great potential for assisting breast cancer screening. We use saliency maps to study the role of breast lesions in the decision-making process of AI systems for breast cancer detection in screening mammograms. We retrospectively collected mammograms from 191 women with screen-detected breast cancer and 191 healthy controls matched by age and mammographic system. Two radiologists manually segmented the breast lesions in the mammograms from CC and MLO views. We estimated the detection performance of four deep learning-based AI systems using the area under the ROC curve (AUC) with a 95% confidence interval (CI). We used automatic thresholding on saliency maps from the AI systems to identify the areas of interest on the mammograms. Finally, we measured the overlap between these areas of interest and the segmented breast lesions using Dice's similarity coefficient (DSC). The detection performance of the AI systems ranged from low to moderate (AUCs from 0.525 to 0.694). The overlap between the areas of interest and the breast lesions was low for all the studied methods (median DSC from 4.2% to 38.0%). The AI system with the highest cancer detection performance (AUC = 0.694, CI 0.662-0.726) showed the lowest overlap (DSC = 4.2%) with breast lesions. The areas of interest found by saliency analysis of the AI systems showed poor overlap with breast lesions. These results suggest that AI systems with the highest performance do not solely rely on localized breast lesions for their decision-making in cancer detection; rather, they incorporate information from large image regions. This work contributes to the understanding of the role of breast lesions in cancer detection using AI.

Figure 1

Saliency analysis. (a) Mammogram with a manually segmented lesion. (b) Saliency map for an AI system. (c) Relevant region (in green) obtained by thresholding the saliency map in (b).

Figure 2

Experimental methodology. (a) Analysis of the detection performance of AI systems using the area under the ROC curve (AUC). (b) Saliency analysis of breast lesions using Dice’s similarity coefficient (DSC).

Figure 3

Saliency of AI systems for screening mammograms. From left to right: END2END, DMV-CNN, GMIC, and GLAM. The first two columns show the saliency maps for the best-performing systems in our study. It is clear that saliency shows a high value in a large area within each mammogram, regardless of lesion location and size.