Attention-Guided Curriculum Learning for Weakly Supervised Classification and Localization of Thoracic Diseases on Chest Radiographs

Abstract

In this work, we exploit the task of joint classification and weakly supervised localization of thoracic diseases from chest radiographs, with only image-level disease labels coupled with disease severity-level (DSL) information of a subset. A convolutional neural network (CNN) based attention-guided curriculum learning (AGCL) framework is presented, which leverages the severity-level attributes mined from radiology reports. Images in order of difficulty (grouped by different severity-levels) are fed to CNN to boost the learning gradually. In addition, highly confident samples (measured by classification probabilities) and their corresponding class-conditional heatmaps (generated by the CNN) are extracted and further fed into the AGCL framework to guide the learning of more distinctive convolutional features in the next iteration. A two-path network architecture is designed to regress the heatmaps from selected seed samples in addition to the original classification task. The joint learning scheme can improve the classification and localization performance along with more seed samples for the next iteration. We demonstrate the effectiveness of this iterative refinement framework via extensive experimental evaluations on the publicly available ChestXray14 dataset. AGCL achieves over 5.7% (averaged over 14 diseases) increase in classification AUC and 7%/11% increases in Recall/Precision for the localization task compared to the state of the art.

A self-attention based deep learning method for lesion attribute detection from CT reports

“…To reduce manual annotation burden, some researchers leveraged the rich information contained in associated radiology reports. Disease-related labels have been mined from reports for classification and weakly-supervised localization on X-ray and CT images [3]- [6].…”

Kernel Joint Sparse Representation Based on Self-Paced Learning for Hyperspectral Image Classification

“…Rather than using all training samples for learning, curriculum learning (CL) or self-paced learning (SPL) adopts a gradual learning strategy to select samples from easy to complex to use in training [17,[27][28][29][30][31]. Both CL and SPL share a similar conceptual learning paradigm but differ in the derivation of the curriculum.…”

Fusion High-Resolution Network for Diagnosing ChestX-ray Images

“…A series of studies was conducted to classify thoracic disease using this dataset. Existing CXR image diagnosis with deep learning [26][27][28][29][30][31][32][33] was used to resize or down-sample the high-resolution or original highpixel images and eliminate most of the pixels in the images, with the hope that useful disease information would not be lost. The mainstream framework of a CNN for diagnosing thorax disease is shown in Figure 1, in which the input size of the CXR image is normally set to 224 × 224 × 3.…”