Indian Journal of Science and Technology

Abstract

Objectives: Manifestation of thermal inconsistencies in skin lesions could be a potential pointer towards malignancy. In this work, an attempt was made to characterize skin cancer lesions through Thermography and Image processing. Methods/ Analysis: In this work, the Region Of Interest (ROI) was extracted from each of the thermal images of skin cancer lesions followed by their colour based clustering to characterize their thermal properties. Histograms and first order statistical parameters of both these image types were obtained and studied. Findings: With respect to the RGB model, for certain parameters, the content of red colour (considered to be indicative of warmer areas) was found in maximum percentage as compared to green and blue where blue colour (considered to be indicative of coolest areas) was at minimum. It was observed that the maximum values for red, green and blue ranged from 193 to 255, 110 to 255 and 39 to 255 respectively; minimum values of red in two cases was above 100 while minimum values for green were below 100. The values of blue were 0 or close to 0. The range of mean values was highest for red colour (137-190). Results hint towards certain characteristics in thermal images which can be potential indicators of malignancy of a lesion. Novelty /Improvement: These findings indicate the need to further experiment in this area with larger database for obtaining more robust results with diagnostic markers.

Keywords: Infrared Imaging, ImageJ, Image Processing, Skin Cancer Detection, Thermography