Gaze-Guided Class Activation Mapping: Leveraging Human Attention for Network Attention in Chest X-rays Classification

TL;DR

This paper introduces GG-CAM, a lightweight method that incorporates radiologists' visual attention into CNNs to improve chest X-ray classification accuracy and interpretability without requiring human attention during inference.

Contribution

The paper presents a novel gaze-guided class activation mapping (GG-CAM) method that leverages human attention to regulate network focus in CNNs for medical image classification.

Findings

Significant increase in AUC metrics for ResNet50 and EfficientNetv2 with GG-CAM.

Enhanced interpretability and pathology localization in CNNs.

Lightweight extension with only 3 additional trainable parameters.

Abstract

The increased availability and accuracy of eye-gaze tracking technology has sparked attention-related research in psychology, neuroscience, and, more recently, computer vision and artificial intelligence. The attention mechanism in artificial neural networks is known to improve learning tasks. However, no previous research has combined the network attention and human attention. This paper describes a gaze-guided class activation mapping (GG-CAM) method to directly regulate the formation of network attention based on expert radiologists' visual attention for the chest X-ray pathology classification problem, which remains challenging due to the complex and often nuanced differences among images. GG-CAM is a lightweight ($3$ additional trainable parameters for regulating the learning process) and generic extension that can be easily applied to most classification convolutional neural networks (CNN). GG-CAM-modified CNNs do not require human attention as an input when fully trained. Comparative experiments suggest that two standard CNNs with the GG-CAM extension achieve significantly greater classification performance. The median area under the curve (AUC) metrics for ResNet50 increases from $0.721$ to $0.776$. For EfficientNetv2 (s), the median AUC increases from $0.723$ to $0.801$. The GG-CAM also brings better interpretability of the network that facilitates the weakly-supervised pathology localization and analysis.