An evolutionary framework for detecting protein conformation defects

In the medical community, the demand for detecting texture abnormalities related to pathological Protein Conformation Diseases (PCD) has been increasing for years. In this paper, we use the value (V) channel of an HSV (Hue, Saturation, Value) image, which provides a measurable Detection Region of the V Channel (DRVC) for isolating muscular Intranuclear Inclusions (INIs) indicative of Oculopharyngeal Muscular Dystrophy (OPMD), one variety of PCD. Derived from the Cumulative Distribution Function (CDF) computation over DRVC (CDF-DRVC), the robust cumulative histogram presents the INIs-texture-carrying intensity data as non-decreasing distributions, leading to an accurate estimation of intensities in the texture feature extraction. Beyond the proposed CDF-DRVC methodology, we introduce a methodology of feature synthesis to enhance the quality of feature representations. The experiments are conducted on irregular microscopic images (where variations in size and shape occur) in the binary classes of healthy and sick cases related to OPMD. The proposed framework achieves a higher accuracy rate for the detection as compared to other methods, and exhibits a detection strategy for other types of PCD in medical applications.