Kulczynski Similarity Index Feature Selection based Map Estimated Rocchio Classification for Brain Tumor Disease Diagnosis

Abstract

An early discovery of brain tumor is essentialto cure the disease completely.Classification isimportantissue to be resolved in disease diagnosis. The conventional techniques designed for brain tumor disease classification finds presence of disease. But, it failed to able the diagnosis performanceof brain tumor disease.Toconquerthelimits, proposed Kul Similarity Index Feature Selection based MAP Estimated Rocchio Classification (KSIFS-MERC) Technique is introduced. The proposed technique isemployed fortumor risk factor recognitionand patient data in their disease diagnosis viaimproved accuracy and lessertime utilization. The KSIFS-MERC methodinitially performs Kulczynski Similarity Index based Feature Selection (KSI-FS) process where KulSimilarity Index is used to find out the similarity between features for analyzing the feature as relevant or irrelevant. After the feature selection process, Maximum a Posteriori Probability (MAP) estimated Rocchio Classifier is used

to perform brain tumor disease diagnosis byenhanced accuracy. MAP estimated Rocchio Classifier precisely classifies patient as normal or abnormal according to maximum a posteriori probability result.By this way, KSIFS-MERC Technique increases the risk factor identificationand brain tumor syndromeanalysis performance as compared to existingmethods. Experimental evaluation ofKSIFS-MERC methodis performedthroughEpileptic Seizure Recognition Dataseton metricsnamely tumor diagnosis accuracy, tumor diagnosis time, andfalse positive rate with number of patients.Experimental outcomeshow that KSIFS-MERC methodis toimprovetumor diagnosis accuracy as well asminimize tumor diagnosis time when compared to conventional methods.