Event Title
X-Means Clustering Implementing the Gap Statistic for Multiple Positron Emission Particle Tracking
Faculty Mentor
Dr. Arthur Ruggles
Department (e.g. History, Chemistry, Finance, etc.)
Nuclear Engineering
College (e.g. College of Engineering, College of Arts & Sciences, Haslam College of Business, etc.)
College of Engineering
Year
2018
Abstract
The most efficient and accurate method for clustering Coincidence Lines (CL) for Positron Emission Particle Tracking (PEPT) is undetermined. A number of methods have been created to perform this task. A novel clustering method featuring x-means is presented using an automated k-value estimator for k-means clustering. Gap statistic is used to select the best k-value. Geant4 Application for Tomographic Emission is used to simulate particles of 50 µCi and the detection and electronic chain associated with the Siemens Inveon Pre-Clinical Scanner. The simulation produces an array of coincident lines (CL) in a format consistent with scanner output. The CL are used to reconstruct particle positions every 100 ms. The position reconstruction is compared to earlier methods.
X-Means Clustering Implementing the Gap Statistic for Multiple Positron Emission Particle Tracking
The most efficient and accurate method for clustering Coincidence Lines (CL) for Positron Emission Particle Tracking (PEPT) is undetermined. A number of methods have been created to perform this task. A novel clustering method featuring x-means is presented using an automated k-value estimator for k-means clustering. Gap statistic is used to select the best k-value. Geant4 Application for Tomographic Emission is used to simulate particles of 50 µCi and the detection and electronic chain associated with the Siemens Inveon Pre-Clinical Scanner. The simulation produces an array of coincident lines (CL) in a format consistent with scanner output. The CL are used to reconstruct particle positions every 100 ms. The position reconstruction is compared to earlier methods.