[Paper Review] A Web-based Mpox Skin Lesion Detection System Using State-of-the-art Deep Learning Models Considering Racial Diversity

Shams Nafisa Ali, Md. Tazuddin Ahmed+6 more|arXiv (Cornell University)|Jun 25, 2023

Poxvirus research and outbreaks10 citations

TL;DR

The paper presents a web-based mpox skin lesion detector that benchmarks multiple deep learning models on a diverse dataset (MSLD v2.0) and uses color-space augmentation and transfer learning to improve performance, achieving 83.59% accuracy.

ABSTRACT

The recent 'Mpox' outbreak, formerly known as 'Monkeypox', has become a significant public health concern and has spread to over 110 countries globally. The challenge of clinically diagnosing mpox early on is due, in part, to its similarity to other types of rashes. Computer-aided screening tools have been proven valuable in cases where Polymerase Chain Reaction (PCR) based diagnosis is not immediately available. Deep learning methods are powerful in learning complex data representations, but their efficacy largely depends on adequate training data. To address this challenge, we present the "Mpox Skin Lesion Dataset Version 2.0 (MSLD v2.0)" as a follow-up to the previously released openly accessible dataset, one of the first datasets containing mpox lesion images. This dataset contains images of patients with mpox and five other non-mpox classes (chickenpox, measles, hand-foot-mouth disease, cowpox, and healthy). We benchmark the performance of several state-of-the-art deep learning models, including VGG16, ResNet50, DenseNet121, MobileNetV2, EfficientNetB3, InceptionV3, and Xception, to classify mpox and other infectious skin diseases. In order to reduce the impact of racial bias, we utilize a color space data augmentation method to increase skin color variability during training. Additionally, by leveraging transfer learning implemented with pre-trained weights generated from the HAM10000 dataset, an extensive collection of pigmented skin lesion images, we achieved the best overall accuracy of $83.59\pm2.11\%$. Finally, the developed models are incorporated within a prototype web application to analyze uploaded skin images by a user and determine whether a subject is a suspected mpox patient.

Motivation & Objective

• Address mpox (Monkeypox) diagnostic challenges due to similarity with other rashes.
• Develop a web-based tool for rapid, non-PCR-assisted screening using deep learning.
• Create and utilize the Mpox Skin Lesion Dataset Version 2.0 (MSLD v2.0) with mpox and non-mpox classes.
• Mitigate racial bias through data augmentation that increases skin color variability.
• Leverage transfer learning from HAM10000 to boost model performance.

Proposed method

• Benchmark multiple CNN architectures: VGG16, ResNet50, DenseNet121, MobileNetV2, EfficientNetB3, InceptionV3, Xception.
• Use color space augmentation to increase skin color variability during training.
• Apply transfer learning with pre-trained weights from the HAM10000 pigmented skin lesion dataset.
• Train and evaluate models on MSLD v2.0 (mpox, chickenpox, measles, hand‑foot‑mouth disease, cowpox, healthy).
• Integrate the best-performing model into a prototype web application for uploaded image analysis.

Experimental results

Research questions

• RQ1 Can state-of-the-art CNN models effectively classify mpox versus other infectious skin diseases on a diverse dataset?
• RQ2 Does color space data augmentation reduce racial bias and improve performance across skin tones?
• RQ3 To what extent does transfer learning from HAM10000 improve mpox-related classification accuracy?
• RQ4 How does the best model perform in a web-based prototype setting for user-uploaded skin images?

Key findings

• Best overall accuracy achieved: 83.59% with a standard deviation of ±2.11%.
• The MSLD v2.0 dataset includes mpox and five non-mpox classes (chickenpox, measles, hand-foot-mouth disease, cowpox, healthy).
• Color space augmentation was employed to increase skin color variability during training to mitigate racial bias.
• Transfer learning using HAM10000 pre-trained weights contributed to improved performance.
• The models were integrated into a web application prototype that analyzes uploaded skin images for suspected mpox.