Analysis of Positron Emission Tomography Data for Tumour Detection and Delineation

Zhang, W. and Ray, S. (2022) Analysis of Positron Emission Tomography Data for Tumour Detection and Delineation. 14th SINAPSE Annual Scientific Meeting, Glasgow, UK, 13-14 June 2022.

Abstract

Recent developments in statistical image analysis and machine learning are culminating towards developing innovative tools for automatic analysis of three-dimensional radiological images — PET (Positron Emission Tomography) images. Statistical imaging together with other machine learning techniques are the epitome of digitalizing healthcare and offers many opportunities for providing patients with personalized medicine/therapy and reducing the cost of diagnosis/treatment. However, the three major challenges in radiology are: (1) increasing demand for medical imaging; (2) decreasing turnaround times caused by mass data; (3) diagnostic accuracy that should lead to a quantification of images. To address these challenges along with ethical issues regarding the use of Artificial Intelligence in patient care, there is a need to develop a new framework of statistical analysis which can be readily used by clinicians and can be trained with a relatively lower number of samples. In this project, we have developed a kernel-based method on PET image segmentation which will be more direct for tumour detection, delineation, monitoring and radiotherapy planning. We are currently working on combing other complementary information, such as CT(Computed tomography) images. The kernel-based method is a non-parametric regression technique which can corporate 3D information for a set of images. It is also computational efficient and can produce reproductive and robust results compared with other statistical methods. A huge advantage is that our kernel-based method builds up a surface over images which produces continuous contour-based results rather than traditional binary results, thus mimicking human observers’ behaviour. Another advantage of the kernel-based methods is that there is a great potential for developing a probabilistic approach with uncertainty measurement along with the segmentation.

• Item Type: Conference or Workshop Item
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Zhang, Wenhui and Ray, Professor Surajit
• Authors: Zhang, W., and Ray, S.
• College/School: College of Science and Engineering; College of Science and Engineering > School of Mathematics and Statistics > Statistics