Synthetic Generation of Dermatoscopic Images with GAN and Closed-Form Factorization

TL;DR

This paper introduces a novel unsupervised data augmentation method using GANs and latent space factorization to generate diverse dermatoscopic images, improving skin lesion classification accuracy and model explainability.

Contribution

It presents a new GAN-based augmentation technique with semantic control over generated images, enhancing model performance on dermatological datasets.

Findings

Improved skin lesion classification accuracy on HAM10000 dataset.

Set a new benchmark for non-ensemble models in dermatology.

Enhanced model explainability through generated data analytics.

Abstract

In the realm of dermatological diagnoses, where the analysis of dermatoscopic and microscopic skin lesion images is pivotal for the accurate and early detection of various medical conditions, the costs associated with creating diverse and high-quality annotated datasets have hampered the accuracy and generalizability of machine learning models. We propose an innovative unsupervised augmentation solution that harnesses Generative Adversarial Network (GAN) based models and associated techniques over their latent space to generate controlled semiautomatically-discovered semantic variations in dermatoscopic images. We created synthetic images to incorporate the semantic variations and augmented the training data with these images. With this approach, we were able to increase the performance of machine learning models and set a new benchmark amongst non-ensemble based models in skin lesion classification on the HAM10000 dataset; and used the observed analytics and generated models for detailed studies on model explainability, affirming the effectiveness of our solution.