Machine-learning Optimization for Prostate Brachytherapy Planning (MOPP): a Randomized-controlled Trial Evaluating Dosimetric Outcomes
Machine-learning Optimization for Prostate Brachytherapy Planning (MOPP): a Randomized-controlled Trial Evaluating Dosimetric Outcomes
The proposed, mono-institutional, randomized-controlled trial aims to determine whether the dosimetric outcomes following prostate Low-Dose-Rate (LDR) brachytherapy, planned using a novel machine learning (ML-LDR) algorithm, are equivalent to manual treatment planning techniques. Forty-two patients with low-to-intermediate-risk prostate cancer will be planned using ML-LDR and expert manual treatment planning over the course of the 12-month study. Expert radiation oncology (RO) physicians will then evaluate and modify blinded, randomized plans prior to implantation in patients. Planning time, pre-operative dosimetry, and plan modifications will be assessed before treatment, and post-operative dosimetry will be evaluated 1-month following the implant, respectively.
Study Outline:
Traditionally treatment planning for prostate Low-Dose-Rate (LDR) brachytherapy has relied on manual planning by an expert treatment planner. This process involves the planner selecting the location of 80-110 small, radioactive seeds within the prostate; the goal of this process is to maximize the amount of radiation delivered to the cancer while minimizing radiation to healthy tissues, all while making sure the seeds are implantable by the physician. Although this process is effective it is time-consuming (taking anywhere from 30 minutes to several hours to plan).
Machine learning (ML), a form of statistical computation that relies on historical training information to adapt and predict novel solutions, has significant potential for improving the efficiency and uniformity of prostate LDR brachytherapy. The ability of this algorithm to mimic several features demonstrated by expert treatment plans has been difficult to perform using conventional computer algorithms and is a significant advantage. It is expected that by implementing an ML program in the planning workflow for prostate LDR brachytherapy it is possible to significantly decrease the planning time, while improving the uniformity of plan outcomes, and maintaining comparable quality to human planners.
This study will evaluate whether a computer program based on machine learning (ML) can be used to maintain plan quality in prostate LDR brachytherapy that is not inferior to manual planning by a human expert. In addition, it is expected that planning time may decrease to only a few minutes using ML planning.
What Will Happen:
If you decide to participate in this study your first visit will involve an ultrasound study of your prostate to map out the treatment area. After your initial visit for ultrasound imaging nothing further is required on your part for the purposes of the study.
Your images and treatment information will then be used to create a brachytherapy treatment plan by both a human planner, and one by an ML program. Only one treatment plan from one of these groups (a process known as randomization) will be used, your treating physician will not know where your plan came from (a process known as blinding). Your physician will examine the plans, grade its acceptability, and make modifications to it if needed. This final plan will be used to deliver your brachytherapy.
Follow-Up Visits:
You will have a follow-up study approximately 1 month after your brachytherapy treatment. The purpose of this study is to gauge how well your brachytherapy was delivered.
For the follow-up study you will have a CT scan to show the area that was treated (the prostate gland). No further action is required on your part.
Length of Study Participation:
Your participation in this study will after your follow-up visit, approximately 1 month after your brachytherapy treatment.
A total of 42 patients will be enrolled in this study from the Odette Cancer Centre.
Inclusion Criteria:
Exclusion Criteria: