Iterative Attention Mining for Weakly Supervised Thoracic Disease Pattern Localization in Chest X-Rays

Abstract

Given image labels as the only supervisory signal, we focus on harvesting, or mining, thoracic disease localizations from chest X-ray images. Harvesting such localizations from existing datasets allows for the creation of improved data sources for computer-aided diagnosis and retrospective analyses. We train a convolutional neural network (CNN) for image classification and propose an attention mining (AM) strategy to improve the model's sensitivity or saliency to disease patterns. The intuition of AM is that once the most salient disease area is blocked or hidden from the CNN model, it will pay attention to alternative image regions, while still attempting to make correct predictions. However, the model requires to be properly constrained during AM, otherwise, it may overfit to uncorrelated image parts and forget the valuable knowledge that it has learned from the original image classification task. To alleviate such side effects, we then design a knowledge preservation (KP) loss, which minimizes the discrepancy between responses for X-ray images from the original and the updated networks. Furthermore, we modify the CNN model to include multi-scale aggregation (MSA), improving its localization ability on small-scale disease findings, e.g., lung nodules. We experimentally validate our method on the publicly-available ChestX-ray14 dataset, outperforming a class activation map (CAM)-based approach, and demonstrating the value of our novel framework for mining disease locations.

Learning to Recognize Chest-Xray Images Faster and More Efficiently Based on Multi-Kernel Depthwise Convolution

“…That is the reason that the Chest X-ray 14 dataset keeps 1024 × 1024 bitmap images [34] to preserve details, which exceeding the 512 × 512 images in OpenI dataset [3]. Similarly, most global image-based CNN methods adopt large images as inputs, 512 × 512 or even 1024 × 1024 in [2], [18], [37]. And some local image-based CNN methods use 224 × 224 inputs [7], because the 224 × 224 inputs are large enough to take local details for local images.…”

Deep learning applications in pulmonary medical imaging: recent updates and insights on COVID-19

“…Cai et al [126] proposed an attention mining AM strategy to improve CNN's sensitivity or saliency to disease patterns. Moreover, the ResNet CNN model was modified to include multi-scale aggregation (MSA) to improve the localization of small-scale disease findings.…”

Weakly Supervised Brain Lesion Segmentation via Attentional Representation Learning

“…On the other hand, weakly supervised segmentation approaches which only require the coarse-grained (e.g. image-level) labels point to a promising direction [4,3,6,7,10]. These approaches aim to exploit the minimum level of expert annotations, while being able to generate the fine-grained segmentation automatically.…”

“…Many of these weakly supervised approaches leverage the Class Activation Maps (CAMs) [14] as a key step for segmentation. For example, the work in [7] uses multi-layer CAMs to detect histological features of glioma in CLE images, while [3] proposes an iterative mining pipeline to localise lesion in different areas. However, a major drawback of CAMs is that they can be very noisy in practice, and thus it is often challenging to derive good initial segmentation proposals from them, especially for 3D images.…”