Towards Better Dermoscopic Image Feature Representation Learning for Melanoma Classification

TL;DR

This paper introduces a novel framework combining GAN-based data augmentation and implicit hair denoising to improve melanoma classification from dermoscopic images, addressing data imbalance and artifact interference.

Contribution

It proposes a new implicit hair denoising method and a large-scale annotated dataset, enhancing feature representation learning for melanoma detection.

Findings

The framework outperforms existing methods in melanoma classification accuracy.

The hair denoising strategy effectively reduces artifact interference.

The augmented dataset improves model training and robustness.

Abstract

Deep learning-based melanoma classification with dermoscopic images has recently shown great potential in automatic early-stage melanoma diagnosis. However, limited by the significant data imbalance and obvious extraneous artifacts, i.e., the hair and ruler markings, discriminative feature extraction from dermoscopic images is very challenging. In this study, we seek to resolve these problems respectively towards better representation learning for lesion features. Specifically, a GAN-based data augmentation (GDA) strategy is adapted to generate synthetic melanoma-positive images, in conjunction with the proposed implicit hair denoising (IHD) strategy. Wherein the hair-related representations are implicitly disentangled via an auxiliary classifier network and reversely sent to the melanoma-feature extraction backbone for better melanoma-specific representation learning. Furthermore, to train the IHD module, the hair noises are additionally labeled on the ISIC2020 dataset, making it the first large-scale dermoscopic dataset with annotation of hair-like artifacts. Extensive experiments demonstrate the superiority of the proposed framework as well as the effectiveness of each component. The improved dataset publicly avaliable at https://github.com/kirtsy/DermoscopicDataset.