Analysis and Comparison of Accuracy in Brain Tumor using Berkeley Wavelet Transform and Robust Principal Component Analysis

Abstract

The main objective of this study is to compare Berkeley wavelet transform (BWT) and robust principal component analysis (ROBPCA) techniques in tumor analysis to improve the accuracy of image processing. Based on the sample sizes of BWT (N=16) and ROBPCA (N=16), tumor MR pictures of various brain tumor illnesses have been gathered. Image segmentation has been finished, and textural features have been retrieved using image processing methods. The accuracy and sensitivity of the parameter are taken into consideration by both organizations when evaluating tumor detection and evaluation. The sample size for each group could be determined by maintaining the enrollment ratio at 1, the threshold alpha at 0.05, the g power at 80%, and the confidence interval at 95%. The absence of a statistically significant difference (p = 0.182) between the two groups was verified using an Independent Sample T-test. The accuracy numbers in BWT are 81.5%, while 84% is the accuracy value in ROBPCA. When it comes to brain tumor detection and analysis, ROBPCA has performed well when compared to BWT.