AI Skin Cancer Tool Offers Diagnostic Edge

Introducing PanDerm A Leap Forward in Skin Disease Detection

Detection of melanoma and a range of other skin diseases is set to become faster and more accurate thanks to a new artificial intelligence (AI) powered tool. This innovative system, known as PanDerm, analyses multiple types of medical images simultaneously, offering a significant advancement in dermatological diagnostics.

University of Queensland researchers including Professors H. Peter Soyer and Monika Janda were part of a Monash University-led team that developed the tool, which improved skin cancer diagnosis accuracy by 11 per cent when used by dermatologists in a reader study.

How PanDerm Works and Its Key Advantages

PanDerm analyses multiple types of images, including close-up photos, dermoscopic images, pathology slides and total body photographs.

A series of evaluations showed the model helped non-dermatologist healthcare professionals improve diagnostic accuracy on various other skin conditions by 16.5 per cent. It could also support clinicians to detect skin cancer early, identifying potentially concerning changes before clinician detection.

Professor Soyer said the tool could be particularly valuable in busy or resource-limited settings, or in primary care where access to dermatologists may be limited. “Differences in imaging and diagnosis techniques can arise due to different levels of resources available in urban, regional and rural healthcare spaces,” Professor Soyer said. “The strength of PanDerm lies in its ability to support existing clinical workflows.

“We have seen that the tool was also able to perform strongly even when trained on only a small amount of labelled data, a key advantage in diverse medical settings where standard annotated data is often limited.”

The Power of Multimodal Data and Extensive Training

Trained on more than two million skin images, data for the model was sourced from 11 institutions in multiple countries, across 4 types of medical images.

First author and PhD student Siyuan Yan from Monash University said the multimodal approach was key to the system's success. "By training PanDerm on diverse data from different imaging techniques, we've created a system that can understand skin conditions the way dermatologists do; by synthesising information from various visual sources," Mr Yan said.

With skin conditions now impacting 70 per cent of the global population, early and accurate diagnosis is crucial and can lead to better treatment outcomes.

Broader Applications and Clinical Endorsements

Unlike current models, which are trained to perform a single task, PanDerm was evaluated on a wide range of clinical tasks such as skin cancer screening, predicting the chance of cancer returning or spreading, skin type assessment, mole counting, tracking lesion changes, diagnosing a wide range of skin conditions, and segmenting lesions.

Alfred Health Victorian Melanoma Service Director, Professor Victoria Mar, said PanDerm showed promise in helping detect subtle changes in lesions over time and provide clues to lesion biology and future metastatic potential. “This kind of assistance could support earlier diagnosis and more consistent monitoring for patients at high risk of melanoma,” Professor Mar said.

“In real-world hospitals or clinic settings, doctors use diverse ways and different types of images to diagnose skin cancer or other skin conditions.”

Current Status and Future Directions

Though showing promising research results, PanDerm is currently in the evaluation phase before broader healthcare implementation.

The team plans to establish standardised protocols for cross-demographic assessments and further investigate the model's performance in varied clinical settings, with a particular focus on ensuring equitable performance across different patient populations and healthcare environments.

The research led by AI and machine learning experts at Monash University also included collaboration with researchers and clinicians at Princess Alexandra Hospital in Brisbane, Alfred Health, Medical University of Vienna, NVIDIA AI Technology Centre in Singapore, University of Florence, Royal Prince Alfred Hospital, NSW Health Pathology and Hospital General Universitario de Alicante in Spain.

The research is published in Nature Medicine.