Brain tumor segmentation in MRI images using Chan-vese technique in MATLAB

Over the last couple of decades, image processing, computer vision and machine learning applications have received considerable attention in biomedical and medical imaging research. Segmentation of tumors from medical images is considered as one of the major challenge due to the overlapping of tissues. The accurate knowledge of these segmented regions play an important role in the diagnosis and treatment of malignant tumors. In this study, we have presented an improved image segmentation technique based on total variation methods for accurately segmenting the tumors from brain MRI image data-set. The technique is based on Chan-Vese active contour without edges. In this methodology, a force shrinks the contour and another force expands the contour. These two forces get balanced when the contour reaches the boundary of our interest objects, and thus allow to find the contour of the object resulting in its precise segmentation. The efficiency of the algorithm has been tested and verified on various brain MRI images. The proposed method can be successfully applied to to detect and segment the tumor and its geometrical dimensions. The approach will benefit scientific community and and healthcare researcher to achieve better diagnosis and can be extended to various other imaging modalities for further research.