Hue-Preserved Quantile-Based Global Contrast Enhancement of Breast Thermograms for Abnormality Detection

With breast cancer cases rapidly increasing worldwide, thermography serves as a potential imaging modality for early abnormality detection. Breast thermograms are low-resolution pseudocolored images that need contrast enhancement for better differentiability between normal and abnormal breasts. Without hue-preservation, conventional contrast enhancement techniques distort the color spaces and, hence, cannot be applied for contrast enhancement of color images, such as breast thermograms. This letter proposes a novel hue-preserving quantile-based global contrast enhancement (QGCE) of breast thermograms for abnormality detection. A comparative analysis of different types of contrast enhancement techniques with and without hue-preservation is shown. An exhaustive comparison of abnormality detection performance in the presence of different types of noises is also presented. The proposed technique achieves noise resilient detection accuracy of 91%. All evaluations are done using 85 breast thermograms from publicly available datasets.