The Depth Estimation and Visualization of Dermatological Lesions: A Novel Methodology (Preprint)

Abstract

BACKGROUND Thus far, considerable research has been focused on classifying a lesion as benign or malignant. OBJECTIVE To propose a novel methodology for the depth estimation and visualization of skin lesions. METHODS We start by doing the same using a CNN, followed by using Explainable AI (XAI) to localize the image features responsible for the CNN output. We apply computer graphics for depth estimation and developing the 3D structure of the lesion. Our novel method, called the red spot analysis, measures the degree of infection based on which a conical hologram is constructed. Physicians can study this hologram via Mixed Reality headsets. RESULTS The neural model achieves an accuracy of 85.61%. We successfully obtained 3D representations of lesion depth using the method stated above. CONCLUSIONS When we map the CNN outputs (benign or malignant) to the corresponding hologram, we observe that a malignant lesion has a higher concentration of red spots (infection) in the upper and deeper portions of the skin. CLINICALTRIAL We do not perform RCT for this study.