Otsu’s thresholding technique for MRI image brain tumor segmentation

• Published in: Multimedia tools and applications Vol. 81; no. 30; pp. 43837 - 43849
• Main Authors: Nyo, Myat Thet, Mebarek-Oudina, F., Hlaing, Su Su, Khan, Nadeem A.
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.12.2022; Springer Nature B.V
• Subjects: Bins; Brain; Brain cancer; Computer Communication Networks; Computer Science; Data Structures and Information Theory; Image processing; Image segmentation; Magnetic resonance imaging; Medical imaging; Multimedia Information Systems; Special Purpose and Application-Based Systems; Tumors; BRATS; Otsu’s thresholding; MRI; Median filter
• ISSN: 1380-7501; 1573-7721
• DOI: 10.1007/s11042-022-13215-1

MRI image segmentation is very challenging area in medical image processing. It is implemented with the low contract of MRI scan. In terms of certain input features or expert information, the major objective of medical image segmentation is to isolate and describe anatomical constitutions. In MRI image segmentation, brain tumor segmentation is more difficult because of its complex structure. The Otsu’s thresholding method is well-known method in image segmentation. In this paper, choosing the classes or bins of Otsu’s thresholding are analyzed on MRI image brain tumor segmentation. As a preprocessing, the 2D MRI images are convert the grayscale image and resized to the same size. And then, median filter is utilized to eliminate the noise from MRI image. In MRI image segmentation, the varieties of classes or bins of Otsu’s thresholding are utilized to segment the brain tumor from MRI images. Then, the morphological operation is used to achieve the accurate tumor regions. All of the experiments are tested on 2015 BRATS dataset. As segmentation quality validation metric, Jaccard similarity index, true positive rate (Sensitivity), true negative rate (Specificity) and accuracy are used to validate the segmented results and their ground truth. According to the results, level 4 or class 4 got 68.7955% in true positive and 95.5593% in accuracy. Class 4 is the best or suitable for MRI image segmentation according to experiments.