SKINGUARD is an innovative tool designed to assist healthcare professionals in training to detect skin cancer. Simulating the tele-dermatology process, it analyzes patient skin images and provides a “similarity score” that indicates the likelihood of a pigmented lesion being benign or malignant. By mimicking real-world workflows, SKINGUARD enhances proficiency and confidence in early skin cancer detection.
What’s new: Using deep learning algorithms we have implemented a platform to examine patient’s skin images and predict the likelihood of nine common pigmented-skin conditions including intraepithelial carcinoma, basal cell carcinoma, squamous carcinoma, and melanoma. SKINGUARD is designed to process images obtained by standard mobile devices or cameras.
Key insights: SKINGUARD is designed to mimic the skin inspection process in an unsupervised tele-dermatology setting. It takes a patient’s skin pigmented lesion image as input, returning a ranked list of benign or malignant skin conditions.
How it works: Users capture images from patients with suspected pigmented-skin conditions using a mobile app or an online platform specifically engineered for the task. An encoding algorithm within Topazium’s cloud combines images and clinical data, feeding a non-linear algorithmic framework which predicts the probability surface across each condition.
Results: SKINGUARD classifies suspected skin conditions with similar accuracy as reported across the existing reference literature for well-trained dermatologists.
Why it matters: SKINGUARD can assist on the training of dermatologists and general practitioners in areas with limited access to high-quality imaging equipment (e.g. dermatoscopy) delivering real-time inferences, which would accelerate critical delivery treatment’s times across remote regions.
HOW TO RUN THE EXPERIMENT
Access: available through a dedicated app and/or via an API.
Image upload: Upload your image by clicking the “Browse” button. The system can work with any type of image format (jpeg, tiff, png). After image upload, use the crop system to focus on the specific target lesion to be analysed and to leave other potential lesions out of the image.
Running your experiment: Expected run time for each image: less than 45 seconds
Output: A similarity score that represents the probability (in %) of the image to be classified as one of the most common different pigmented skin lesions.
AWARDS
Best HealthCare Solution Award at the IoT Solutions World Congress (Barcelona, 2024)
USE CASES
Asociación Cuerpo y Alma. During the 2023 campaign in the Impenetrable, a vast region of native forest spanning over 40,000 km² in the Chaco plains of northern Argentina, dermatologists from the Asociación Cuerpo y Alma used SkinGuard to screen the local population. Several individuals with neoplastic lesions were detected, which would have otherwise gone unnoticed. This pilot test demonstrated that SkinGuard meets its objective of aiding in the early detection of pigmented lesions in a rural setting.
Summer Campaign “Take Care of Your Skin”. Together with Medicus and Laboratorios Andrómaco, Topazium participated in the summer campaign “Take Care of Your Skin” to raise awareness among the population about proper skin care during their time at the beach. At a dermatological station in the center of the coastal city of Pinamar, a dermatologist assisted by SkinGuard was able to screen over 400 people in a single weekend, detecting several neoplastic lesions.
6th World Congress of Dermoscopy 2024. Laboratorios Andrómaco used SkinGuard at their booth to demonstrate how it can assist healthcare professionals in real time using a mobile phone.