A Fast and Efficient Region based Aneurysm Segmentation Model for Medical Image Segmentation

Abstract

Aneurysm and blood vessel delineation from medical images facilitates efficient diagnosis of the Aneurysm and vessels (Stroke or Hemorrhage and Stenosis or malformations) and registration of patient images obtained at different times. Computer-aided diagnosis and detection of Aneurysms via Segmentation algorithms is a complex and multi-faceted issue in medical image processing, as adjoining vessels are the high-intensity structures whereas aneurysms are of low contrast and intensity. Obviously, segmentation is essential to identify the disease severity by change monitoring and also to know further Haemo-dynamic situation in critical cases. Change detection and further analysis gives the complete picture of the case of interest. For Brain tumor detection and analysis, there are several segmentation algorithms but only few are suitable for aneurysm detection and analysis. There is a necessity to provide an efficient segmentation model for aneurysm analysis, change detection and delineation which overcomes the limitations on speed and accuracy of other models. The objective of this paper is to first, apply local binary fitting (LBF), chan-vese (CV) models to aneurysm analysis. Then perform Region based Aneurysm Segmentation model frame work on data sheets of MR Angiography of brain. It is a perfect level set based Active contour model which converges in short span without requirement of any stability and termination criterions. The key feature of this model is that delineation is independent of choice of mask dimensions. Promising results are obtained with the proposed model