Medical Physics Seminar – Monday, November 14, 2022
Real Time Motion-adapted Prostate SBRT using the Radixact Synchrony System: Commissioning, Testing and Clinical Implementation
Wolfgang A. Tome, Ph.D.
Professor, Department of Radiation Oncology, Professor, The Saul R. Korey Department of Neurology Director of Medical Physics, Institute for Onco-Physics at the Albert Einstein College of Medicine
This seminar will discuss some of the scientific and engineering aspects of motion management. We will briefly look at models to predict a patient's breathing ahead that are based on state space based local linear prediction such as the model free Linear Prediction algorithm (Prediction based on analogues in the past) and models based on Functional Approximation using tangent space (Prediction based on the Ordinary Least square Estimator, Partial Least Squares Regression (PLS) estimator, and the Principal Component Regression (PCR) estimator. While these models can be used to predict a patient's breathing ahead, they are not useful for classifying a patient's breathing pattern. For this Recurrence Quantification Analysis (RQA) and RQA measures of complexity are more adequate tools.
While on the one hand science shows why things work, the solution it proposes might be very sensitive to perturbations, on the other hand an engineering solution to a problem needs to be very robust against perturbation. So, while not always as elegant the scientific solution, the engineering solution needs to solve the problem in a way that allows for predictable and consistent clinical use. The engineering approach to motion management implemented in Accuracy's Radixact Realtime Motion Management Solution (Synchrony) uses two motion tracking models a Quasi-Static Model and a respiratory model. We discuss the commissioning and quality assurance of this system and its use for Prostate SBRT.