Various forms of melanoma detection technologies consisting of optical methods are under development in the lab. A brief description is as follows:
Natural phenomena such as skin cancer have fractal geometrical characteristics. The fractal principal of self-similarity suggests that comprehensive optical diagnostic testing could target similarly dysplastic characteristics on multiple size scales. Three melanoma detection systems under development target atypia on various scales. On the ultrastructure scale, hyperspectral dermoscopy implements diffuse light to image from the ultraviolet range (superficial basal layer dysplasia) to the infrared range (dermal involvement). On the cellular scale, confocal reflectance microscopy implements near-infrared ballistic photon scattering to enable detection of pagetoid melanocytes and the roughness of the basal layer. On the level of proteins and lipids, Raman scattering from a confocally-gated micro-probe detects the presence of cancer-associated molecules. These three technologies may form a powerful 4-stage screening sequence: (1) a cellphone camera-enabled broad screening app, (2) hyperspectral dermoscopy for general care practitioners, (3) dermatologist screening with the additional aid of confocal microscopy and (4) specialized screening that interrogates morphologically suspicious cells with Raman spectroscopy. In each stage, pre-processing steps transform raw spectra and/or images into quantitative metrics of malignancy. Then, machine learning maximizes diagnostic accuracy. At near 100% sensitivity, these technologies are expected to yield increasing specificity for the increasing cost of their increasing complexities.
The following video is something we put together to show the translation of the RGB/mAID technology. It's basically hyper spectral dermoscopy with pathological computer vision:
The confocal microscopy can best be explained in the rap music
video I put together for the Science Genius Rap Battle
Here is the -->LINK<--