Discriminant boundary features based classification and staging of chronic liver disease

Abstract

Chronic liver disease is a continuous mortal disease that causes more damages to human life. In most of the growing countries, the temporality and morbidity amount has enlarged due to CLD. However, the virus moves frequently inside the human body due to its dynamic behavior. Prevention from CLD is a hard task due to its dynamic movement of virus in the environment. When compared to other diseases, preventive mechanisms are highly required for the CLD because of the entry of virus that are affecting the environment activities in the society.CLD is an acute disease that spreads among the world wide. In order to check the CLD, liver function test should be needed. Apart from that physical examination, laboratory tests, tissue analysis and imaging tests are developed to find out the CLD. Further, invasive and non invasive approaches are developed for observing CLD. CLD identification is based on the collection of image features from the patients through ultrasound imaging. The early diagnosis rate of CLD is still low, which is partially due to asymptomatic phenomena and neglect by the patients. Ultrasound screening is still the first choice that is applicable for primary healthcare hospitals, whereas CT and MRI is more accurate but costly. The main aim of this research work is to design a detection mechanism for detecting and identifying the CLD from the abnormal patients and to enhance prevention from all other diseases in an effective way. To achieve this goal, a detection mechanism is designed into three phases. A novel ensemble learning machine and Hough histogram oriented gradient based feature extraction mechanism is proposed in the first phase for the classification of CLD. The objective of this proposed method is to detect edge of the region. The most important characteristics of proposed mechanism is that it mitigates CLD through the dynamic development of multiple detection schemes to detect the CLD quickly as possible and provides better accuracy by evaluating the classification performance. More significantly, the proposed mechanism can effectively handle both the detecting and classifying of these multi detection schemes for the detection of CLD. Essentially, these multi detection methods in proposed mechanism are used to improve the accuracy and minimize the needs of biopsy using ELM classifiers. In proposed mechanism, each classification stage highly optimizes with respect to classifier and features set by avoiding inaccurate classification and drastically increasing the computer aided diagnosis in all hospitals. In this proposed method ultrasound images, clinical detecting, and laboratory investigating are used to evaluate the stages of CLD newline