PHG-Net: Persistent Homology Guided Medical Image Classification

TL;DR

PHG-Net introduces a topological feature integration method using persistent homology to enhance medical image classification, capturing anatomical structures often overlooked by traditional CNNs and Transformers.

Contribution

The paper presents a novel persistent homology guided approach that effectively incorporates topological features into deep neural networks for improved medical image classification.

Findings

Significant accuracy improvements over state-of-the-art methods.

Effective integration of topological features into CNNs and Transformers.

Lightweight PH module compatible with various architectures.

Abstract

Modern deep neural networks have achieved great successes in medical image analysis. However, the features captured by convolutional neural networks (CNNs) or Transformers tend to be optimized for pixel intensities and neglect key anatomical structures such as connected components and loops. In this paper, we propose a persistent homology guided approach (PHG-Net) that explores topological features of objects for medical image classification. For an input image, we first compute its cubical persistence diagram and extract topological features into a vector representation using a small neural network (called the PH module). The extracted topological features are then incorporated into the feature map generated by CNN or Transformer for feature fusion. The PH module is lightweight and capable of integrating topological features into any CNN or Transformer architectures in an end-to-end fashion. We evaluate our PHG-Net on three public datasets and demonstrate its considerable improvements on the target classification tasks over state-of-the-art methods.