Noninvasive prediction of skin precancer severity using in vivo cellular imaging and deep learning algorithms. - nonmelanoma skin cancer (nmsc) represents the most common form of cancer in the human…

Noninvasive prediction of skin precancer severity using in vivo cellular imaging and deep learning algorithms. - nonmelanoma skin cancer (nmsc) represents the most common form of cancer in the human body and causes twice as many fatalities each year as melanoma. The method for diagnosing and treating nmscs requires a skin biopsy that is processed and stained for analysis on a standard optical microscope. This process is painful for patients, and the invasiveness of biopsy introduces a delay into nmsc detection, which contributes to patient morbidity and adds substantial cost to the healthcare system. Enspectra health's mission is to bring digital oncology diagnostics to the point-of-care for earlier cancer detection where the healthcare cost and burden to patients is minimal. Cancer is an inherently cellular disfunction, and yet modern medicine still lacks the basic ability to view cellular histology without biopsy. This widespread clinical need is the core motivation for enspectra and its innovations. Enspectra aims to address this unmet clinical need for a better method to detect nmscs earlier. This direct to phase II application builds on the progress of awarded phase i, phase II, and phase IIB projects (R43CA221591, R44CA221591). In these projects, enspectra has progressed from concept, through technical feasibility, and into clinical trials, recently completing enrollment in a pivotal trial for submission to the FDA for 510(k) approval (clinicaltrials.GOV: NCT05619471). Enspectra has created the first portable, fiber coupled, combined multiphoton microscopy (MPM) and reflectance confocal microscopy (RCM) system for in vivo imaging of nmsc. In this direct to phase II proposal, enspectra aims to leverage the analytic power of its multimodal data and extend its reach to actinic keratosis (AK), a precancerous lesion that can progress to nmsc. Enspectra will build a large-scale digital database of histopathology in AKS on patients before topical therapy. AKS that do not respond to therapy are more likely to progress to nmsc and are clinically of higher risk. Using the therapy outcome as an indicator of AK severity, enspectra will train a deep learning algorithm to predict which AKS would be unresponsive solely on pathologic features in our noninvasive images. The ability to identify problematic AKS before they become malignant should improve surveillance of high-risk patients, hasten detection of nmsc, and lessen the burden of surgical intervention to low-risk patients.