What is it about?

Melanoma is one of the deadliest types of skin cancer and can be difficult to treat when it's advanced. To reduce mortality rates, early detection is key. In order to do this, computer-aided systems have been developed to help dermatologists diagnose the condition. To make it more accessible to the public, researchers are working on creating portable, at-home diagnostic systems. An Android-based smartphone application utilizing image capture, preprocessing, and segmentation was developed to extract Asymmetry, Border irregularity, Color variegation, and Diameter (ABCD) features from skin lesions. Using these feature sets and support vector machines, the application can accurately classify malignant and benign cases. Processing an image takes under a second and the system's performance metrics (sensitivity, specificity, accuracy, and AUC) are competitive with or better than current methods. What's more, the user-friendly application is easy to download and navigate, which is key to making medical diagnosis more democratic.

Featured Image

Why is it important?

Melanoma is a particularly dangerous type of skin cancer and is hard to treat in its later stages. Therefore, early detection is key in reducing mortality rates. In order to assist dermatologists in doing this, computer-aided systems have been designed for desktop computers. However, there is a desire for the development of mobile, at-home diagnostics for melanoma risk assessment.

Perspectives

Here, we introduce a smartphone application that captures images and extracts ABCD features to classify skin lesions as either malignant or benign. The algorithms used are adaptive to make the process light and user-friendly, as well as reliable in diagnosis. Images can be taken with the phone's camera or imported from public datasets. The entire process of taking the image, performing preprocessing, segmentation and classification is completed on an Android smartphone in a short time. Our application is evaluated on a dataset of 200 images, and achieved either comparable or better performance metrics than other methods. Additionally, it is easy-to-download and easy-to-navigate for the user, which is important for the widespread use of such diagnostics.

Prof Santosh Pandey
Iowa State University

Read the Original

This page is a summary of: Skin Cancer Diagnostics with an All-Inclusive Smartphone Application, Symmetry, June 2019, MDPI AG,
DOI: 10.3390/sym11060790.
You can read the full text:

Read

Contributors

The following have contributed to this page