Identification of cancerous lesions from healthy skin using an infrared, optical filter based approach

Abstract

Skin cancer, in all its various forms has been identified by the American Cancer Society as the most common form of cancer in both the United States and worldwide. In fact, more Americans will be diagnosed with skin cancer than all other forms of cancer combined. Typically, initial skin cancer diagnosis relies on the asymmetry, border, core, diameter, and evolution (ABCDE) test. As a subjective test, the ABCDE method relies on the skill of the practitioner to provide an accurate diagnosis. Therefore, a diagnostic instrument that could be used as an initial screening tool to determine whether a skin lesion is potentially cancerous would have significant application. This need is especially acute in under-served communities without ready-access to skilled practitioners. It also could have widespread applicability at the point-of-care to reduce the strain on clinical dermatologists. We have recently demonstrated preliminary results showing the capability to discriminate between healthy skin and cancerous lesions using an infrared method inspired by human color vision. This technique relies on identifying variations in the infrared (IR) spectra of healthy and cancerous tissue; however, unlike standard IR-spectroscopic sensing methods it does not require the use of a spectrometer. Rather, this approach utilizes the response through three IR optical filters; similar to the method that human eye uses to discriminate between hundreds of thousands of colors in the natural world. Here, we report on calculated studies demonstrating the selectivity of this approach for various types of cancerous lesions including melanomas. We explore various potential optical filter sets to discriminate not only between healthy and cancerous tissue, but also to determine whether it may be possible to discriminate between various types of melanoma lesions using this bioinspired method. These results demonstrate that this approach has potential as a low-cost, widely deployed sensing method for detection of cancerous skin lesions, prior to examination by a trained specialist.