Algorithm For Automated Detection of Melanoma

Summary

Incidence of malignant melanoma (MM) are increasing in the United States, and early
detection and excision are critical due to the high tendency for MMs to metastasize. Oregon Health & Science University researchers have developed an automated cellular quantification software for use with reflectance confocal microscopy that could increase the efficiency and accuracy of MM diagnosis as well as identify premature signs of aging.

Technology Overview

Comprehensive diagnosis of MM typically requires both reliable histopathology and the trained eye of a clinician for assessment, but biopsies are invasive and often require significant turnaround time. Reflectance confocal microscopy tools make optical sectioning possible, increasing the ability to image microscopic structural details. The laboratory of Dr. Daniel Gareau has taken this technique a step further, to develop automated algorithms that can analyze stacks of confocal images to provide quantitative index of factors, all without the need for invasive biopsies. Advantages of this approach include:

• Automated detection of melanoma to bridge the gap between dermoscopy and typical biopsies;
• Computer-based detection and quantification of pagetoid melanocytes, keratinocytes and disruption at the dermal-epidermal junction to aid in melanoma diagnosis;
• Computer-based detection and quantification of keratinocytes for distinguishing squamous cell carcinomas and identifying premature signs of aging; and
• The ability to provide quantitative metrics to reflectance confocal microscopy to reduce subjectivity and enhance accuracy in visual assessments, and potentially reduce the need for unnecessary biopsies.

Publications

Gareau et al., “Automated detection of malignant features in confocal microscopy on superficial spreading melanoma versus nevi.” J of Biomedical Optics 15(2010): 061713, Link

Gareau. “Automated identification of epidermal keratinocytes in reflectance confocal microscopy.” J of Biomedical Optics 16(2011): 030502, Link

Licensing Opportunity

This technology is available for exclusive or non-exclusive licensing.