Three Step Authentication of Brain Tumour Segmentation Using Hybrid Active Contour Model and Discrete Wavelet Transform

Abstract

An innovative imaging research is expected in the medical field due to the challenges and inaccuracies in diagnosing the life-threatened harmful tumours. Brain tumor diagnosis is one of the most difficult areas of study in diagnostic imaging, with the maximum fine for a small glitch given the patients survival rate. Conventionally, biopsy method is used to identify the tumour tissues from the brain's soft tissues by the medical researchers (or) practitioners and it is unproductive due to: (i) it requires more time, and (ii) it may have errors. This paper presents the three-stage authentication-based hybrid brain tumour segmentation process and it makes the detection more accrual. Primarily, tumour area is segmented from a magnetic resonance image and after that when comparing a differentiated segment of an image to the actual image, an improved active contour model is employed to achieve a good match. In addition, discrete wavelet transform is used for the features extraction which leads to improve the accuracy and robustness in the tumour diagnosis. Finally, RELM classifier is used for precise classification of brain tumours. The most effective section of our method is checking the status of the tumour through finding the tumour region. The results are evaluated through new dataset, and it demonstrates that the suggested approach is more efficient than the alternatives as well as provides 96.25% accuracy.