Delineation of brain tumours for radiotherapy patients using image segmentation techniques

Abstract

Author(s): Yousif Abdallah*

Background: Precise radiation segmentation of brain tumours is essential for the definition of Gross Tumour Volumes (GTVs). For sensitive GTV detection, the most comprehensive data may be provided by MRI images. This study was conducted to delineate malignant brain tumours in radiotherapy patients using image segmentation techniques. Materials and Methods: MRI images of 10 patients with astrocytoma were used in this study. A new method of watershed-based segmentation techniques was used to segment the GTVs (region of interest, ROI) in T1- and T2-weighted MRI images. These techniques were used to delineate the GTVs morphologically and accurately and were compared with manually delineated GTVs. To analyse the segmentation technique quantitatively, the Dice similarity coefficient, sensitivity and segmentation specificity were calculated. The images were processed with the MATLAB image processing toolbox. In MRI images, brain tumours can be easily detected if the objects have a sufficient contrast background. Results: The experimental study and the new method achieved quality segmentation, Dice similarity coefficient, sensitivity and specificity values of 0.92 ± 0.09, 0.86±0.03, 0.94 ± 0.06 and 0.90 ± 0.09, respectively, were achieved. For GTV volume segmentation, image detection and filter morphology were conducted by reading the image, completing brain detection, image dilation, image filling, edge removal, and brain smoothing. This study led to an alternative way of showing an object in a divided brain. This method can help remove unwanted background information and improve diagnosis via brain MRI. Conclusion: The new watershed segmentation method allows the semiautomatic segmentation of GTVs. Anatomical and functional MRI images can create a new way to identify radiation therapy goals and methods.

Share this article