GenMix: Combining Generative and Mixture Data Augmentation for Medical Image Classification

TL;DR

GenMix is a novel data augmentation method that combines generative models and mixup techniques to improve medical image classification, especially in limited data scenarios, by enhancing data diversity and boundary learning.

Contribution

This paper introduces GenMix, a new two-stage data augmentation approach that integrates generative models with mixup to address limitations of each in medical imaging.

Findings

GenMix improves classification accuracy on FLL CT images.

Textual Inversion with GenMix outperforms other generative methods.

The method enhances synthetic data quality and diversity.

Abstract

In this paper, we propose a novel data augmentation technique called GenMix, which combines generative and mixture approaches to leverage the strengths of both methods. While generative models excel at creating new data patterns, they face challenges such as mode collapse in GANs and difficulties in training diffusion models, especially with limited medical imaging data. On the other hand, mixture models enhance class boundary regions but tend to favor the major class in scenarios with class imbalance. To address these limitations, GenMix integrates both approaches to complement each other. GenMix operates in two stages: (1) training a generative model to produce synthetic images, and (2) performing mixup between synthetic and real data. This process improves the quality and diversity of synthetic data while simultaneously benefiting from the new pattern learning of generative models and the boundary enhancement of mixture models. We validate the effectiveness of our method on the task of classifying focal liver lesions (FLLs) in CT images. Our results demonstrate that GenMix enhances the performance of various generative models, including DCGAN, StyleGAN, Textual Inversion, and Diffusion Models. Notably, the proposed method with Textual Inversion outperforms other methods without fine-tuning diffusion model on the FLL dataset.