SBIR-STTR Award

The study addresses the need for better physician training by providing specific accurate clinical descriptions of pigmented lesions including malignant melanoma. In Phase 1, we will develop clinical algorithms to precisely describe malignant melanoma. In this phase, our aims include 1) location of early melanomas within the melanoma color band, definition of variegation, precise color description and quantization, 2) automatic border detection using global image features and automatic induction, 3) seeking optimal measures for analysis of asymmetry and irregularity and apply to pigmented lesion images, and 4) showing the importance of these features. Phase I will demonstrate what are the relative weights of critical features to be combined to generate a simple clinical rule for melanoma diagnosis. Phase 11 will add one more analytic feature: computer-determined visual texture. What texture measures best help to distinguish melanoma from irregular seborrheic keratoses? Analysis as in Phase I will be performed for benign pigmented lesions and will include development of software for computer-assisted instruction for improved clinical diagnosis of pigmented lesions. Future application of computer vision diagnostic assistance systems might enable low-cost screening of large groups for pigmented lesions. These could include routine hospital admissions, nursing home patients at annual intervals and public skin cancer screenings.Awardee's statement of the potential commercial applications of the research: If successful, this research will provide algorithms to enable development of pigmented lesion identification software and pigmented lesion computer-aided instruction (CAI). American and foreign medical schools, large clinics and private practitioners comprise a large and growing market, with growing melanoma incidence and public and physician interest.

Thesaurus Terms:
computer assisted diagnosis, computer assisted instruction, computer program /software, computer system design /evaluation, diagnosis quality /standard, melanoma, neoplasm /cancer diagnosis, pigmentation disorder artificial intelligence, biomarker, digital imaging, erythema, morphology statistics /biometry National Cancer Institute (NCI)

A significant number of malignant melanomas, especially early melanomas curable by excision, are not diagnosed correctly in the clinic. Available teaching materials do not allow recognition of the critical features of melanoma, especially for early lesions. The Phase I digital image analysis effort has resulted in a set of clinical rules that may permit higher accuracy in diagnosis of early melanomas. Digital image analysis will be continued in Phase II to improve diagnostic accuracy. The Melanoma Detection CD-ROM will be developed for advanced medical students and primary care physicians. The tutorial will be tested and modified by both target groups prior to general release. The CD-ROM will incorporate some new computer instruction techniques and an atlas with descriptor and thumbnail image indexing capability, to allow "best- match" lookup. This CD-ROM will utilize results available only recently measuring diagnostic performance based on image features, successive pairs in discrimination, and diagnosis of atypical melanomas. The success of the module will be measured using medical student diagnostic performance with slides and primary care clinical outcomes. The semi- automatic classification system from Phase I will be modified and tested on the larger set of images available in Phase II using ROC curves. PROPOSED COMMERCIAL APPLICATIONS: Specific features of melanoma will be presented quantitatively to medical students and primary care physicians in a CD-ROM tutorial. No similar product exists, and sales are expected to be significant. The semi- automated diagnostic system will have add-on potential for digital dermatoscopic systems.

Thesaurus Terms:
computer assisted diagnosis, computer assisted instruction, computer program /software, computer system design /evaluation, interactive multimedia, melanoma, neoplasm /cancer diagnosis diagnosis design /evaluation, education evaluation /planning, medical education, optical data storage, pigment human subject