Knowledge Based Planning for Head and Neck Radiotherapy

Abstract

Aims: Investigate the potential application, utilisation and clinical implementation of a simple knowledge based planning solution for head and neck radiotherapy within a clinical radiotherapy department. Methods: A knowledge base of 141 previously treated head and neck patients was created by extracting data using a Python data mining script and the existing scripting capabilities of the RayStation treatment planning system. This knowledge base was used to create three separate knowledge based models to predict the optimal and mandatory achievable doses for the spinal cord, brainstem, and parotids respectively. The models were validated using a range of methods. A graphical user interface was developed and validated to display the predicted model doses from within the planning system. Results: It was demonstrated that the three developed models could accurately identify treatment plans in which the doses to the brainstem, spinal cord and parotids could be reduced without adversely affecting any other aspects of treatment plan quality. Within the validation patient cohort, it was shown that the implementing models could potentially reduce the maximum spinal cord, maximum brainstem and mean parotid doses by 5.42Gy, 3.62Gy and 5.93Gy respectively. It was also demonstrated that the developed GUI was accurate and could feasibly be introduced into routine clinical use. Conclusions: Three simple knowledge based models have been developed and validated which could be clinically implemented and potentially significantly reduce organ at risk doses for head and neck patients within the clinical radiotherapy department. These models present a low cost, accessible, and simple alternative to commercially available knowledge based planning solutions.

Date of Award	1 Aug 2022
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Julia-Claire Handley (Supervisor)