An interpretable classifier for high-resolution breast cancer screening images utilizing weakly supervised localization
This work addresses the challenge of efficient and accurate medical image analysis for breast cancer screening, offering a model that reduces computational costs and improves performance over existing methods, though it is incremental as it builds on prior frameworks.
The paper tackles the problem of classifying high-resolution breast cancer screening images by proposing an interpretable classifier that uses weakly supervised localization, achieving an AUC of 0.93 for malignant findings, outperforming baseline models and radiologists by 0.11 in AUC while being 4.1x faster and using 78.4% less GPU memory.
Medical images differ from natural images in significantly higher resolutions and smaller regions of interest. Because of these differences, neural network architectures that work well for natural images might not be applicable to medical image analysis. In this work, we extend the globally-aware multiple instance classifier, a framework we proposed to address these unique properties of medical images. This model first uses a low-capacity, yet memory-efficient, network on the whole image to identify the most informative regions. It then applies another higher-capacity network to collect details from chosen regions. Finally, it employs a fusion module that aggregates global and local information to make a final prediction. While existing methods often require lesion segmentation during training, our model is trained with only image-level labels and can generate pixel-level saliency maps indicating possible malignant findings. We apply the model to screening mammography interpretation: predicting the presence or absence of benign and malignant lesions. On the NYU Breast Cancer Screening Dataset, consisting of more than one million images, our model achieves an AUC of 0.93 in classifying breasts with malignant findings, outperforming ResNet-34 and Faster R-CNN. Compared to ResNet-34, our model is 4.1x faster for inference while using 78.4% less GPU memory. Furthermore, we demonstrate, in a reader study, that our model surpasses radiologist-level AUC by a margin of 0.11. The proposed model is available online: https://github.com/nyukat/GMIC.