A Spatio-Frequency Domain Anisotropic Filtering for Contrast Enhancement of Histopathological Images

Abstract

Medical images play a critical role in diagnosing of several symptomatic and asymptomatic disorders and medical conditions for early treatment to prevent and reduce fatalities. Medical image processing has played an indomitable role in processing of these medical images from age old times. Preprocessing and contrast enhancement is a critical and foremost stage in any medical image processing technology. This research article proposes a novel edge preserving spatio-frequency domain anisotropic filtering in a hybrid filtering approach to overcome the limitations of existing methods like lack of edge preserving features, significant loss of data during filtering process etc. In the proposed work, the process of filtering has been done in the frequency domain which least reflects any modifications in the time domain representation which forms the motivation behind the proposed utility. Histopathological images like red blood cells (RBC), white blood cells (WBC) and platelets whose analysis is of great significance in medical domain for detection of several abnormalities, is taken as the prime dataset for investigation in this research work. Three noise types involving Speckle, Impulse and Gaussian which normally tend to corrupt medical images has been taken for the filtering and enhancement process. Performances in terms of peak signal to noise ratio (PSNR), structural similarity index (SSIM) to analyze the edge structure deformity indicate superior performance of proposed hybrid filtering approach.