1887

Abstract

Background & Objectives Aneurysm is a condition in which an abnormal dilation occurs in the wall of an artery or vein. Since cerebral aneurysm morphology is considered as a potential surrogate of rupture, clinicians often desire its segmentation before any pre-operative or interventional planning is performed. Also, segmentation of such images offers the potential of identifying the right treatment for a wide range of medical conundrums. This paper presents a simple and fast automatic algorithm based on active contour technique for cerebral aneurysm segmentation. Since the performance of active contour is highly dependent on the placement of the initial contour, the objective of this work is to automate this placement process to get the desired contour quickly. Methods: The active contour approach is a prime candidate for practical exploitation because active contours make effective use of specific prior information about objects and this makes them inherently efficient algorithms. In this work, we utilize gradient vector flow (GVF) snake, which is an extension of the active contour because it efficiently converges to boundary concavities. However, computational time remains a concern; therefore, we try to place the initial contour as close to the desired boundary of the object in the medical image as possible. For this, we use Otsu's adaptive thresholding method that is based on the histogram of the intensity distribution of the input image. As a post-processing step, the extracted contour is smoothened using linear regression. Results: The proposed method has been implemented on angiography datasets obtained from two different sources. The angiography CT datasets are collected from University of College of London and Hamad Medical Corporation (512x512x405 pixels at circa 0.5mm resolution). The original, ground truth and segmented intracranial aneurysms are shown in Figure 1, which clearly reflects the potential of the proposed algorithm. The average time taken for segmentation is 2 s per slice (without optimization) on MATLAB 7.5 on a PC with Pentium 4, 3 GHz dual core processor. Conclusion: Although our work might be a simplified approach, we envision that it could represent a necessary first step towards improving the performance of active contour that is dependent on the initial contour in a clinical setting.

Loading

Article metrics loading...

/content/papers/10.5339/qfarf.2013.BIOP-0119
2013-11-20
2019-10-24
Loading full text...

Full text loading...

http://instance.metastore.ingenta.com/content/papers/10.5339/qfarf.2013.BIOP-0119
Loading
This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error