Supervised Classification Approaches for Brain Tumour Classification Using Fused Wavelet Features

Shailendra Kumar Mishra
Hiran Kumar Singh

Indian Journal of Forensic Medicine & Toxicology
Vol. 15 No. 2 (2021): Indian Journal of Forensic Medicine & Toxicology

In this study, an efficient pattern recognition technique is developed for Brain Image Classification (BIC) intonormal or abnormal. Wavelet transform features with supervised classification have a potential role to playin bringing Magnetic resonance Images (MRI) of the brain into practical clinical use.The developed patternrecognition technique uses Discrete Wavelet Transform (DWT), Dual tree M-band Wavelet Transform(DMWT), and Stationary Wavelet Transform (SWT) for feature extraction, k-Nearest Neighbour (kNN)and Naive Bayes (NB) for classification and is considered as an effective and accurate tool for brain imageanalysis for cancer classification. Also, the predominant coefficients are chosen from the combined featurespace by rank features of statistical feature selection approach. Results show that the proposed system actsas a pre-treatment predictor for BIC with an accuracy of 88.5% for kNN and 95.5% for NB classification.