17-Year-Old Develops Early Diagnosis Tool for Melanoma

The current medical recommendation for visually inspecting suspicious melanoma patches is the "ABCD" of pigmented spots:

A: Asymmetry of shape (asymetry)

B: Irregular edges (irregular border)

C: Varied colors (color varigation)

D: Larger than 6mm in diameter (diameter > 6mm)

Elizabeth Zhao's research aims to develop an early diagnostic tool to maximize treatment opportunities. This project uses image processing algorithms to perform edge detection and image segmentation on suspicious skin spots, identifying their characteristics. The data is then statistically analyzed and processed, and the next stage of learning is conducted by "Artificial Neural Networks" (ANN). The learning process used data from 350 cases to train the ANN, establishing a database to determine whether the spots are benign or malignant, aiding in the final diagnosis.

Zhao stated that even for asymmetrical, irregular-edged benign melanomas that current computer diagnostics find difficult, her diagnostic tool achieves an accuracy rate of 75%.