Abstract:Medical image datasets often exhibit long-tailed distributions due to the inherent challenges in medical data collection and annotation. In long-tailed contexts, some common disease categories account for most of the data, while only a few samples are available in the rare disease categories, resulting in poor performance of deep learning methods. To address this issue, previous approaches have employed class re-sampling or re-weighting techniques, which often encounter challenges such as overfitting to tail classes or difficulties in optimization during training. In this work, we propose a novel approach, namely \textbf{S}aliency-guided and \textbf{P}atch-based \textbf{Mix}up (SPMix) for long-tailed skin cancer image classification. Specifically, given a tail-class image and a head-class image, we generate a new tail-class image by mixing them under the guidance of saliency mapping, which allows for preserving and augmenting the discriminative features of the tail classes without any interference of the head-class features. Extensive experiments are conducted on the ISIC2018 dataset, demonstrating the superiority of SPMix over existing state-of-the-art methods.

What problem does this paper attempt to address?

This paper aims to solve the long - tail distribution problem in medical image datasets, especially in the skin cancer image classification task. The long - tail distribution means that in the dataset, the number of samples of some common disease categories is much larger than that of rare disease categories. This unbalanced data distribution will lead to a significant decline in the performance of deep - learning methods when dealing with low - frequency categories. To meet this challenge, the paper proposes a new method - Saliency - guided and Patch - based Mixup (SPMix). Specifically, the SPMix method combines saliency maps to guide the mixing of images of different categories to generate new tail - category images. This method can preserve and enhance the discriminative features of the tail category without introducing interference from the head - category features. In this way, SPMix can not only effectively solve the data imbalance problem but also improve the model's ability to recognize the tail category. The main contributions of the paper include: 1. Proposing a new supervised contrastive learning framework that combines saliency - guided and patch - based mixing strategies for long - tail classification tasks. 2. Introducing saliency maps into the mixing process, enabling the generated tail - category samples to preserve diagnostic features. 3. Different from traditional mixing strategies, SPMix uses lesion - aware mixing ratios and flexibly specifies the mixing ratio of each patch according to the saliency map, which is more suitable for the characteristics of medical images. 4. The experimental results on the ISIC2018 dataset show that SPMix outperforms the existing state - of - the - art methods in long - tail medical image classification tasks.

Saliency-guided and Patch-based Mixup for Long-tailed Skin Cancer Image Classification

CCMix: Curriculum of Class-Wise Mixup for Long-Tailed Medical Image Classification.

Mixed Re-Sampled Class-Imbalanced Semi-Supervised Learning for Skin Lesion Classification

Semi-supervised medical image classification based on CamMix

Flexible Sampling for Long-tailed Skin Lesion Classification

PCSalmix: Gradient Saliency-Based Mix Augmentation for Point Cloud Classification

Mix from Failure: Confusion-Pairing Mixup for Long-Tailed Recognition

QMix: Quality-aware Learning with Mixed Noise for Robust Retinal Disease Diagnosis

Saliency detection-guided for image data augmentation

EdgeMixup: Improving Fairness for Skin Disease Classification and Segmentation

Gradsalmix: gradient saliency-based mix for image data augmentation

Gradient Saliency-aware CutMix for Semi-Supervised Medical Image Segmentation

Out-of-Distribution Detection for Long-tailed and Fine-grained Skin Lesion Images

Multi-Pooling Attention Learning for Melanoma Recognition.

SUMix: Mixup with Semantic and Uncertain Information

Enhanced Long-Tailed Recognition with Contrastive CutMix Augmentation

Lesionmix Data Enhancement and Entropy Minimization for Semi-Supervised Lesion Segmentation of Lung Cancer

PCLMix: Weakly Supervised Medical Image Segmentation via Pixel-Level Contrastive Learning and Dynamic Mix Augmentation

Self-contrastive Feature Guidance Based Multidimensional Collaborative Network of Metadata and Image Features for Skin Disease Classification

Text-Guided Mixup Towards Long-Tailed Image Categorization

ECL: Class-Enhancement Contrastive Learning for Long-tailed Skin Lesion Classification