Intelligent MRI Liver Images based Cirrhosis Disease Identification using Modified Learning Principle

Now-a-days, the lifestyle of people is totally changed and many of them are using alcohol, unhealthy foods, energy drinks, self-medication and so on. And some of them keep these things as a regular day-to-day activity. This will cause a huge destruction to human liver and makes suffer a lot. Especially alcohol consumption is a cause and leads many people to get liver failures as well as their life is risky enough to lead. The main problem in the case is late identification of such disease and providing a treatment after the liver is completely affected. In that stage it is really a complex thing to provide treatment and cure the disease completely. Earlier detection of such liver disease will preserves human life in rich manner as well as the physicians feel comfortable to treat the patients and preserve their valuable life. In medical industry lots of innovative technologies and association of smart devices are entered as well as these systems provide a huge support to physicians to exactly predict the disease structure with the help of computer programs. The logic of machine learning based data analysis provides a huge support to medical field to exactly predict the diseases based on digital images, in which the digital image may be in any format such as Magnetic Resonance Imaging (MRI), Computer Tomography Images (CT) and so on. This paper is intended to design a new machine learning approach called Learning based Disease Prediction Logic (LBDPL), in which it is admired from the conventional learning technique called Support Vector Machine (SVM). The implementation parameters of the classical SVM logic is modified by using the hyper parameter tuning logic and attain the new proposed strategy called LBDPL. This paper considers the problem of Non-Alcoholic Fatty Liver Disease (NAFLD) prediction using machine learning methodology. The propose methodology of LBDPL is cross-validated by using the conventional algorithms such as Logistic Regression (LR) Classifier, Support Vector Machine (SVM) and k-Nearest Neighbor algorithm (kNN). The proposed approach identifies the liver disease with respect to four different categories such as: Normal, Steatosis, Steatohepatitis and Cirrhosis. All these specifications are clearly demonstrated over the further sections. Several literatures are considered in this paper to analyze the best prediction model to provide a good support to physicians and patients to identify the liver disease on earlier stages. The resulting section provides an efficiency of the proposed approach with proper graphical representations as well as the performance metrics of theproposed scheme LBDPL is clearly illustrated with graphical proofs.