Comparative Evaluation of Medical Thermal Image Enhancement Techniques for Breast Cancer Detection

Abstract

Thermography is a potential medical imaging modality due to its capability in providing additional physiological information. Medical thermal images obtained from infrared thermography systems incorporate valuable temperature properties and profiles, which could indicate underlying abnormalities. The quality of thermal images is often degraded due to noise, which affects the measurement processes in medical imaging. Contrast stretching and image filtering techniques are normally adopted in medical image enhancement processes. In this study, a comparative evaluation of contrast stretching and image filtering on individual channels of true color thermal images was conducted. Their individual performances were quantitatively measured using mean square error (MSE) and peak signal to noise ratio (PSNR). The results obtained showed that contrast stretching altered the temperature profile of the original image while image filtering appeared to enhance the original image with no changes in its profile. Further measurement of both MSE and PSNR showed that the Wiener filtering method outperformed other filters with an average MSE value of 0.0045 and PSNR value of 78.739 dB. Various segmentation methods applied to both filtered and contrast stretched images proved that the filtering method is preferable for in-depth analysis.