Due to the increasing incidences of malignant melanoma, there is a rising demand for assistive technologies for its early
diagnosis and improving the survival rate. The commonly used visual screening method is with limited accuracy as the
early phase of melanoma shares many clinical features with an atypical nevus, while conventional dermoscopes are not
user-friendly in terms of setup time and operations. Therefore, the development of an intelligent and handy system to
assist the accurate screening and long-term monitoring of melanocytic skin lesions is crucial for early diagnosis and
prevention of melanoma. In this paper, an advanced design of non-invasive and non-radioactive dermoscopy system was
reported. Computer-aided simulations were conducted for optimizing the optical design and uniform illumination
distribution. Functional prototype and the software system were further developed, which could enable image capturing
at 10x amplified and general modes, convenient data transmission, analysis of dermoscopic features (e.g., asymmetry,
border irregularity, color, diameter and dermoscopic structure) for assisting the early detection of melanoma, extract
patient information (e.g. code, lesion location) and integrate with dermoscopic images, thus further support long term
monitoring of diagnostic analysis results.
A clinical trial study was further conducted on 185 Chinese children (0-18 years old). The results showed that for all
subjects, skin conditions diagnosed based on the developed system accurately confirmed the diagnoses by conventional
clinical procedures. Besides, clinical analysis on dermoscopic features and a potential standard approach by the
developed system to support identifying specific melanocytic patterns for dermoscopic examination in Chinese children
were also reported.
|