Detection of tumours from MRI scans using Segmentation techniques
Renuka Devi M N, Cauvery Raju, T. M. Rajesh · 2021 Second International Conference on Electronics and Sustainable Communication Systems (ICESC) · 2021
Medical Image segmentation is complex and challenging task due to the distinct features of biomedical images. MRI scan is commonly used in treating brain, prostate, uterus cancer, ankle and foot cancers. The proposed research work utilizes the real medical images of uterus cancer, brain tumour and throat cancer images obtained from Sagar hospital. MRI images are likely to suffer from the noises such as Gaussian noise, salt pepper noise and speckle noise. Henceforth, noise removal is a very important task to make the image clear with high accuracy and suitable for further diagnosis process. In our paper, the proposed tumour image processing involves three stages 1) converting RGB images to grayscale, 2) pre-processing the images using Gaussian, median and anisotropic filters for noise removal and image enhancement 3) Tumour segmentation using watershed and active contour method. The proposed research work has compared the results of watershed and active contour method. Also, it leverages the ability to find the area and perimeter values of tumour, which indicates the size of brain tumours that provide more accuracy compared to other existing algorithms. Watershed algorithm gives better results, when compared to active contour method, since tumour area using watershed segmentation is more accurate than active contour method. Keywords-