Visual Explanations for Convolutional Neural Networks via Latent Traversal of Generative Adversarial Networks

TL;DR

This paper introduces a novel method combining GANs and CNNs to generate detailed visual explanations of how CNNs interpret COVID-19 chest X-ray images, enhancing interpretability beyond traditional techniques.

Contribution

It proposes a GAN-based framework that disentangles lung features from COVID-19 indicators, enabling fine-grained visualization of CNN responses through latent space interpolation.

Findings

GAN framework effectively visualizes lung feature transitions

Provides detailed insights into CNN decision-making process

Enhances interpretability of medical image classification

Abstract

Lack of explainability in artificial intelligence, specifically deep neural networks, remains a bottleneck for implementing models in practice. Popular techniques such as Gradient-weighted Class Activation Mapping (Grad-CAM) provide a coarse map of salient features in an image, which rarely tells the whole story of what a convolutional neural network (CNN) learned. Using COVID-19 chest X-rays, we present a method for interpreting what a CNN has learned by utilizing Generative Adversarial Networks (GANs). Our GAN framework disentangles lung structure from COVID-19 features. Using this GAN, we can visualize the transition of a pair of COVID negative lungs in a chest radiograph to a COVID positive pair by interpolating in the latent space of the GAN, which provides fine-grained visualization of how the CNN responds to varying features within the lungs.