Automatic Detection of Critical Dermoscopy Features for Basal Cell Carcinoma

DESCRIPTION (provided by applicant): SkinScan digital dermoscopy skin cancer detection software, developed by Missouri's SandA, can now detect critical features of early melanoma. However, there is also a need for diagnostic help for the other 90+% of skin cancers. The need for diagnostic improvement in screening for skin cancers may be greatest in those nurse practitioners who now see the majority of elderly patients in some underserved areas. Underserved clinical arenas with a greater than average incidence of skin cancer and significant numbers of nurse practitioners include both civilian and military clinics in the rural Midwest, where SandA is located. This innovative software is a timely development designed to solve problems every healthcare consumer has encountered- too long a wait to get specialty care, uncertainty about the diagnosis when one does get the care, and too much overall expenditure in providing the care. The proposed Phase I research addresses the steps needed to develop the SkinScan automatic detection modules for basal cell carcinoma. The BASAL features, first described by Stoecker and Stolz, will be programmed during Phase I and incorporated in our early detection system. Additional work during Phase I will allow acquisition of more clinical and dermoscopy images, will allow training of the first nurse practitioner, and will allow development of a hierarchical neural network for diagnosis of basal cell carcinoma. With over a decade of development using thousands of images from dermatology practices in the United States and Europe, SkinScan at the end of the Phase II research period will have demonstrated effectiveness in an early cancer detection clinical trial. This trial provides a bridge to market for a product that will result in increased access to high quality care by providing automated diagnostic assistance for the most deadly skin cancer, melanoma, and the most common skin cancer, basal cell carcinoma, by detecting key early features of these skin cancers. The software is designed to guide the non-specialist nurse practitioner or physician assistant, helping to make the biopsy/no biopsy decision, the most critical decision in early detection of skin cancer. PUBLIC HEALTH RELEVANCE: SandA proposes the SkinScan Training System for Skin Cancer Screening, to meet the needs of nurse practitioners in finding skin cancers. SandA offers practitioners an integrated database system that allows even those without specialized dermatology training to identify skin cancers at a very early stage, before they become deadly. The SandA SkinScan Training System is comprised of 3 parts: Dermoscopy cancer detection software with searchable atlas, DermLite D3 cross-polarized light dermoscope, and Melanoma and Mimics Clinical Tutorial. Phase I of this research proposal contains key milestones in development of the most advanced software dedicated to identifying signs on dermoscopy images of the most common skin cancer, basal cell carcinoma. SkinScan is in an excellent position to compete in the marketplace. Key personnel at SandA bring decades of experience in software development and a record of introducing innovative software to the dermatology market. The SkinScan system is the first such product to be marketed to nurse practitioners, physician assistants, and family practitioners. This proposal concentrates on nurse practitioners, the largest single group of non-physician practitioners seeing primarily adults in a general practice setting. Phase I and Phase II of this proposal should allow us to reach the software sale stage. After that, integration with strategic partners will allow market expansion.