Skin Lesion Diagnosis using Ensembles, Unscaled Multi-Crop Evaluation and Loss Weighting

TL;DR

This paper presents an ensemble-based deep learning approach with unscaled multi-crop evaluation and loss weighting for skin lesion classification, achieving second place in the ISIC 2018 challenge using only public data.

Contribution

It introduces a combination of ensemble models, unscaled multi-crop evaluation, and loss weighting to improve skin lesion diagnosis accuracy.

Findings

Achieved second place in ISIC 2018 challenge

Effective handling of class imbalance with loss weighting

Utilized unscaled multi-crop evaluation for better performance

Abstract

In this paper we present the methods of our submission to the ISIC 2018 challenge for skin lesion diagnosis (Task 3). The dataset consists of 10000 images with seven image-level classes to be distinguished by an automated algorithm. We employ an ensemble of convolutional neural networks for this task. In particular, we fine-tune pretrained state-of-the-art deep learning models such as Densenet, SENet and ResNeXt. We identify heavy class imbalance as a key problem for this challenge and consider multiple balancing approaches such as loss weighting and balanced batch sampling. Another important feature of our pipeline is the use of a vast amount of unscaled crops for evaluation. Last, we consider meta learning approaches for the final predictions. Our team placed second at the challenge while being the best approach using only publicly available data.