Feature | Treatment Planning | July 08, 2020 | By Melinda Taschetta-Millane

Integrating Artificial Intelligence in Treatment Planning

This article ran as an introduction to ITN's Treatment Planning Systems comparison chart in the July/August issue. You can view the chart here.

At the American Association of Physicists in Medicine (AAPM) 2019 meeting, new artificial intelligence (AI) software to assist with radiotherapy treatment planning systems was highlighted. The goal of the AI-based systems is to save staff time, while still allowing clinicians to do the final patient review. 

At the American Association of Physicists in Medicine (AAPM) 2019 meeting, new artificial intelligence (AI) software to assist with radiotherapy treatment planning systems was highlighted. The goal of the AI-based systems is to save staff time, while still allowing clinicians to do the final patient review. 

Machine Learning Treatment Planning

RaySearch demonstrated a new U.S. Food and Drug Administration (FDA)-cleared machine learning treatment planning system. The RaySearch RayStation machine learning algorithm is being used clinically by University Health Network, Princess Margaret Cancer Center, Toronto, Canada, where it was rolled out over several months in late-2019. Medical physicist Leigh Conroy, Ph.D., was involved in this rollout and helped conduct a study, showing the automated plans and traditionally made plans to radiation oncologists to get valuable feedback. She spoke at the AAPM 2019 meeting on this topic.

In an interview with itnTV, Conroy explained that she worked with an algorithm that uses machine learning to create automated treatment plans. “With this, we train the algorithm using a curated set of high-quality, previously delivered plans. Then, it is able to detect the patient that is most similar to a novel patient and create a new treatment plan with no user interaction beyond pressing the play button,” she explained.

Conrad’s study directly compared those automated treatment plans to traditionally generated manual plans. The radiation oncologist then compared those two plans head-to-head and decided whether each plan was acceptable, and chose the favored plan. The plans are not modified as they are performing them, however they are modifiable. “The automated plan is modifiable, but for the purposes of this study, we are not going to be modifying them so we can directly compare the output of the machine learning algorithm to the output of the planners,” she said.

The system was trained using a model that was developed at Princess Margaret, and the model is being used clinically. “The way that we are doing our study is if the doctor does choose the automated plan, then it’s underlined that the physicist knows when they are doing their plan QA that it is an automated plan, and it goes to the same QA that it normally would, and the patient-specific QA is a fully deliverable plan, and that is the plan that the patient is treated with.”

Depending on how things go with the study, it is predicted that AI should see a regular implementation. “That is the point of the study, to make sure we can do this and work it into our regular process and eventually provide it if the doctors continue to like the automated plans,” she said.

One of the main ideas in implementing AI is to save time to get more patient throughput. “We are not measuring the end to end timing of a planner vs. the machine learning. But one of the major advantages is that it takes about 20 minutes for a new patient, however there is no user interacting during those 20 minutes, so the planner can go and do other work or other plans during that time, and come back so there is a fully done plan.”

There is a different process depending on what plan is being created. However, the AI would help to free them up to be able to do other duties. “From a planner expertise perspective, some of the planning techniques such as head and neck might be more complicated so it might take longer for the planner to do it. It’s more reliant on their level of expertise,” she explained. 

Other Automated Systems

Varian also has developed AI-driven automated treatment planning software that is currently being used by several hospitals.

Recently, Varian released the newest version of its treatment planning system, Eclipse v16. This new release includes intelligent features such as RapidPlan PT, a clinical application of machine learning in proton treatment planning, and RT Peer Review, a collaborative workspace designed to streamline and accelerate the peer review process for radiotherapy treatment plans.

Previously only available for photon-based radiotherapy treatment planning, RapidPlan is knowledge-based treatment planning software that enables clinicians to leverage knowledge and data from similar prior treatment plans to quickly develop high-quality personalized plans for patients. This knowledge-based planning software is now available for proton treatment planning with RapidPlan PT. The software also allows dose prediction with machine learning models that can be used as a decision support tool to determine which patients would be appropriate for proton or photon therapy. 

“With the number of operational proton treatment rooms continuing to increase, there is a need for experienced proton therapy clinicians,” said Kolleen Kennedy, chief growth officer, president, Proton Solutions, Varian, in a written statement. “RapidPlan PT helps bridge the learning curve, allowing established centers to share their models and clinical experience. The machine learning in RapidPlan PT has the potential to reduce proton treatment plan optimization from a one- to eight-hour process, as reported by clinical proton centers, to less than 10 minutes, while also potentially improving plan quality.”

Eclipse v16 has received the CE mark and is 510(k) pending.

Related Content:

Artificial Intelligence Greatly Speeds Radiation Therapy Treatment Planning

VIDEO: Use of Machine Learning to Automate Radiotherapy Treatment Planning

 

Related Content

GE Healthcare introduced its artificial intelligence (AI) automation features on its Voluson Swift ultrasound platform at the 2020 Radiological Society of North America (RSNA) virtual meeting. Features of this system include semi-automated contouring, auto identification of fetal anatomy and positioning on imaging. AI is seeing increasing integration in ultrasound systems from numerous vendors.

GE Healthcare introduced its artificial intelligence (AI) automation features on its Voluson Swift ultrasound platform at the 2020 Radiological Society of North America (RSNA) virtual meeting. Features of this system include semi-automated contouring, auto identification of fetal anatomy and positioning on imaging. AI is seeing increasing integration in ultrasound systems from numerous vendors.

Feature | Ultrasound Imaging | February 18, 2021 | By Dave Fornell, Editor
Recent advances in ultrasound image sy...
Example MR images from paediatric brain tumour patients. This first column shows T1-weighted images following the injection of gadolinium contrast agent. The second column shows T2-weighted images and the final column shows apparent diffusion coefficient maps calculated from diffusion-weighted images. (a–c) are taken from a patient with a Pilocytic Astrocytoma, (d–f) are from a patient with an Ependymoma and (g–i) were acquired from a patient with a Medulloblastoma.

Example MR images from paediatric brain tumour patients. This first column shows T1-weighted images following the injection of gadolinium contrast agent. The second column shows T2-weighted images and the final column shows apparent diffusion coefficient maps calculated from diffusion-weighted images. (ac) are taken from a patient with a Pilocytic Astrocytoma, (df) are from a patient with an Ependymoma and (gi) were acquired from a patient with a Medulloblastoma. Image courtesy of Nature Research Journal

News | Pediatric Imaging | February 17, 2021
February 17, 2021 — Diffusio...
Immunotherapy-based precision medicine clinical trials being developed

Getty Images

News | Prostate Cancer | February 16, 2021
February 16, 2021 — Black men die more often of prostate cancer yet, paradoxically, have greater survival benefits fr
The research collaboration agreement covers a joint clinical retrospective study on liver fibrosis severity in Non-Alcoholic Steato-Hepatitis (NASH) patients
News | Artificial Intelligence | February 10, 2021
February 10, 2021 — Median Technologies announced the company has signed a research collaboration agreement with the
 Nuance Communications, Inc. announced the acquisition of Saykara, Inc., a like-minded startup focused on developing a mobile AI assistant to automate clinical documentation for physicians. The acquisition underscores Nuance's ongoing expansion of market and technical leadership in conversational artificial intelligence (AI) and ambient clinical intelligence (ACI) solutions that reduce clinician burnout, enhance patient experiences, and improve overall health system financial integrity.
News | Artificial Intelligence | February 08, 2021
February 8, 2021 — ...

Chart courtesy of the American Cancer Society

News | Breast Imaging | February 08, 2021
February 8, 2021 — Cancer ranks as a leading cause of death in every country in the world, and, for the first time,
Phase III clinical trial of men with a clinical suspicion of prostate cancer finds MRI with targeted biopsies to be more accurate at diagnosis and less intrusive than current standard
News | Magnetic Resonance Imaging (MRI) | February 08, 2021
February 8, 2021 — The results of a Phase III randomized clinical trial have shown that when it comes to detecting cl

Images in 69-year-old man with biopsy-confirmed Gleason score 7 (3+4) prostate cancer. (a) Pretreatment axial T2-weighted fast spin-echo MRI scan (repetition time msec/echo time msec, 3820/97) shows tumor in midline anterior transition zone (arrow). (b) Intraoperative MRI scan shows contoured rectal wall (red line), prostate margin (blue outline), and region of interest (orange outline). Because the urethra was included in planned treatment volume, a suprapubic catheter was placed for continuous bladder drainage during treatment. (c) Intraoperative MRI scan shows focused ultrasound beam path (blue) overlaid on treatment plan. Rectangles illustrate each sonication spot. (d) Thermal map image obtained during treatment with heat deposition color coded in red overlaid on sonication spot. (e) Axial gadopentetate dimeglumine-enhanced MRI scan (230/2.97) obtained immediately after treatment shows devascularized ablated volume (arrows). (f) Corresponding T2-weighted fast spin-echo MRI scan (3820/97) at 5 months after ablation shows complete involution of transition zone. All seven cores from treatment area margins were negative for cancer at biopsy. Image courtesy of the Radiological Society of North America

News | Prostate Cancer | February 05, 2021
February 5, 2021 — A technique that delivers...
Real Time Imaging Technologies LLC iannounced that it has received U.S. Patent No. 10,898,070 IMAGING APPARATUS AND METHODS which covers the company’s methods for incorporating microlens into X-ray imaging detectors that enables significantly lower radiation dose without compromising image resolution.

Getty Images

News | Radiation Dose Management | February 05, 2021
February 5, 2021 —