Implementation of a Knowledge-Based Treatment Planning System in Tays University Hospital
Pakarinen, Tomppa (2018)
Pakarinen, Tomppa
2018
Sähkötekniikka
Tieto- ja sähkötekniikan tiedekunta - Faculty of Computing and Electrical Engineering
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2018-06-06
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tty-201805221736
https://urn.fi/URN:NBN:fi:tty-201805221736
Tiivistelmä
External beam radiotherapy is the most often used radiation therapy method in curative and palliative cancer treatment. Since the discovery of X-rays, radiotherapy has developed to highly sophisticated treatment system consisting of multiple phases and challenges. Successful cancer treatment requires expertise and continuous co-operation across different professions. Today’s radiotherapy methods aim for optimal dose delivery with dynamically conformed field shapes, minimizing the harmful dose effects in surrounding normal tissue.
In this master of science thesis the radiotherapy plans were constructed for intensity-modulated radiotherapy (IMRT) and volumetric arc therapy (VMAT) using Rapidplan (RP), a knowledge-based treatment planning (KBTP) system. Without KBTP, the planner must interactively guide the plan optimization. This is time consuming and may produce lower plan coherence between different planners. Previous studies have shown that RP generated plans shorten the planning time, increase planning coherence within hospitals and can generate clinically acceptable plans with proper organs at risk (OAR) sparing.
In this thesis two head and neck cancer- (HNC) and a prostate model were built in RP. In addition, a previously built robust prostate model was modified for further validation. Prostate models were trained using 126 and 38 plans and HNC models were trained with 156 plans. Model evaluation statistics were used as guiding indicators and most OAR structures yielded good model fit statistics (R^2>0.7,X^2<1.1). Only the robust prostate model had large deviations (∆R^2>0.1) from the guidelines.
The model validation against clinical plans showed similar results to previous research. All RP models could create individual plans meeting the clinical dose-volume constraints and were mainly comparable with the clinical validation plans with no statistically significant deviations (p<0.05). Differences were found in higher PTV doses for prostate and for those OAR structures, which have high sparing priority in clinical planning. This thesis shows that RP models can produce clinically acceptable plans with proper OAR sparing and conformal PTV dose distributions. As a conclusion, RP-generated plans can be used in treatment planning directly or as a starting point for manual optimization.
In this master of science thesis the radiotherapy plans were constructed for intensity-modulated radiotherapy (IMRT) and volumetric arc therapy (VMAT) using Rapidplan (RP), a knowledge-based treatment planning (KBTP) system. Without KBTP, the planner must interactively guide the plan optimization. This is time consuming and may produce lower plan coherence between different planners. Previous studies have shown that RP generated plans shorten the planning time, increase planning coherence within hospitals and can generate clinically acceptable plans with proper organs at risk (OAR) sparing.
In this thesis two head and neck cancer- (HNC) and a prostate model were built in RP. In addition, a previously built robust prostate model was modified for further validation. Prostate models were trained using 126 and 38 plans and HNC models were trained with 156 plans. Model evaluation statistics were used as guiding indicators and most OAR structures yielded good model fit statistics (R^2>0.7,X^2<1.1). Only the robust prostate model had large deviations (∆R^2>0.1) from the guidelines.
The model validation against clinical plans showed similar results to previous research. All RP models could create individual plans meeting the clinical dose-volume constraints and were mainly comparable with the clinical validation plans with no statistically significant deviations (p<0.05). Differences were found in higher PTV doses for prostate and for those OAR structures, which have high sparing priority in clinical planning. This thesis shows that RP models can produce clinically acceptable plans with proper OAR sparing and conformal PTV dose distributions. As a conclusion, RP-generated plans can be used in treatment planning directly or as a starting point for manual optimization.