Transfer Learning with Ensembles of Deep Neural Networks for Skin Cancer Detection in Imbalanced Data Sets

TL;DR

This paper introduces an ensemble CNN approach that combines pre-trained and trained models with metadata to improve skin cancer detection accuracy on imbalanced datasets, demonstrating superior performance over benchmarks.

Contribution

A novel ensemble CNN architecture integrating multiple models and metadata, enhancing detection accuracy on limited, imbalanced skin cancer datasets.

Findings

Outperforms seven benchmark methods across multiple metrics.

Effectively handles limited and imbalanced data.

Improves F1, AUC-ROC, and AUC-PR scores.

Abstract

Several machine learning techniques for accurate detection of skin cancer from medical images have been reported. Many of these techniques are based on pre-trained convolutional neural networks (CNNs), which enable training the models based on limited amounts of training data. However, the classification accuracy of these models still tends to be severely limited by the scarcity of representative images from malignant tumours. We propose a novel ensemble-based CNN architecture where multiple CNN models, some of which are pre-trained and some are trained only on the data at hand, along with auxiliary data in the form of metadata associated with the input images, are combined using a meta-learner. The proposed approach improves the model's ability to handle limited and imbalanced data. We demonstrate the benefits of the proposed technique using a dataset with 33126 dermoscopic images from 2056 patients. We evaluate the performance of the proposed technique in terms of the F1-measure, area under the ROC curve (AUC-ROC), and area under the PR-curve (AUC-PR), and compare it with that of seven different benchmark methods, including two recent CNN-based techniques. The proposed technique compares favourably in terms of all the evaluation metrics.