WonDerM: Skin Lesion Classification with Fine-tuned Neural Networks

TL;DR

This paper introduces WonDerM, a neural network pipeline that leverages segmentation data and ensemble methods to classify seven skin lesion types with high accuracy, aiding non-invasive skin cancer diagnosis.

Contribution

The work presents a novel pipeline combining image resampling, fine-tuning with segmentation data, and ensemble classification for skin lesion diagnosis.

Findings

Achieved 89.9% validation accuracy

Achieved 78.5% test accuracy

Effective multi-class skin lesion classification

Abstract

As skin cancer is one of the most frequent cancers globally, accurate, non-invasive dermoscopy-based diagnosis becomes essential and promising. A task of the Part 3 of the ISIC Skin Image Analysis Challenge at MICCAI 2018 is to predict seven disease classes with skin lesion images, including melanoma (MEL), melanocytic nevus (NV), basal cell carcinoma (BCC), actinic keratosis / Bowen's disease (intraepithelial carcinoma) (AKIEC), benign keratosis (solar lentigo / seborrheic keratosis / lichen planus-like keratosis) (BKL), dermatofibroma (DF) and vascular lesion (VASC) as defined by the International Dermatology Society. In this work, we design the WonDerM pipeline, that resamples the preprocessed skin lesion images, builds neural network architecture fine-tuned with segmentation task data (the Part 1), and uses an ensemble method to classify the seven skin diseases. Our model achieved an accuracy of 0.899 and 0.785 in the validation set and test set, respectively.