the health strategist
institute for continuous health transformation
and digital health
Joaquim Cardoso MSc
Chief Researcher & Editor of the Site
April 28, 2023
Source: Mayo Clinic News Network
A study conducted by Mayo Clinic suggests artificial intelligence could potentially improve time efficiency and standardization for radiation therapy planning in patients with head and neck cancers.
The validation study, published in Frontiers in Oncology, evaluated the efficiency of an algorithm trained by Mayo Clinic and developed in collaboration with Google Health.
Radiation therapy planning for head and neck cancers requires a heavy time investment from multiple specialty areas due to the many organs and structures in close proximity to each other and the anatomical variations of each patient. Experts from the Department of Radiation Oncology at Mayo Clinic investigated whether the algorithm could reduce the time it takes to plan treatment for head and neck cancers and improve the quality of radiation plans and patient outcomes.
In the study, radiation oncologists revised manually drawn head and neck contours, which outline organs at risk of radiation exposure, and contours automatically segmented by the deep-learning model.
According to survey data collected after each review case,
- the deep-learning model produced contours that were ready for clinical use with minor to no revisions 90% of the time,
- compared to 53% of the time for manual contours,
- and reduced the overall contouring and process time by 76%.
“Head and neck radiation treatment planning is resource-intensive and cumbersome,” says Samir Patel, M.D., a radiation oncologist at Mayo Clinic and co-principal investigator of the study.
“This auto-contouring algorithm saved time in the cases we analyzed.”
The model was trained using 445 de-identified CT scan contours from previously treated Mayo Clinic patients with head and neck cancer.
To further refine the dataset, each contour was re-created by two Mayo Clinic head and neck radiation oncologists to meet the gold-standard criteria for radiation therapy.
This initiative is the first research collaboration in a broader initiative between Mayo Clinic and Google Health, announced in 2019, which aims to improve the delivery of care for serious and complex conditions.
This strategic partnership harnesses knowledge and data to create digital tools available on innovative health care platforms that may reach patients worldwide.
“By leveraging the expertise of Mayo Clinic and Google Health, this algorithm has the potential to increase the reach of Mayo Clinic expert radiation oncologists to the global radiation oncology community,” says Dr. Patel.
“This joint research demonstrates the power of Mayo Clinic and Google’s collaboration, bringing together world-class radiation oncologists and medical physicists at Mayo Clinic with software engineers and researchers from Google to tackle an important issue for patients with cancer,” says Cían Hughes, M.B., Ch.B., informatics lead at Google Health. “This is early-stage work, and further clinical research will be required to see how this can be safely integrated into practice.”
The next phase of research will investigate the efficiency of the algorithm outside of Mayo Clinic.
“Mayo Clinic is planning future collaborations with other institutions to demonstrate that the model performs efficiently in their environments,” says Joseph (John) Lucido III, Ph.D., co-principal investigator of the study.
Originally published at https://newsnetwork.mayoclinic.org on April 27, 2023.
REFERENCE PUBLICATION
Validation of clinical acceptability of deep-learning-based automated segmentation of organs-at-risk for head-and-neck radiotherapy treatment planning
J. John Lucido1*,Todd A. DeWees2,Todd R. Leavitt2,Aman Anand3,Chris J. Beltran4,Mark D. Brooke5,Justine R. Buroker6,Robert L. Foote1,Olivia R. Foss7,Angela M. Gleason7,Teresa L. Hodge1,Cían O. Hughes5,Ashley E. Hunzeker1,Nadia N. Laack1,Tamra K. Lenz1,Michelle Livne5,Megumi Morigami5,Douglas J. Moseley1,Lisa M. Undahl1,Yojan Patel5,Erik J. Tryggestad1,Megan Z. Walker5,Alexei Zverovitch5 and Samir H. Patel3
- 1Department of Radiation Oncology, Mayo Clinic, Rochester, MN, United States
- 2Department of Health Sciences Research, Mayo Clinic, Phoenix, AZ, United States
- 3Department of Radiation Oncology, Mayo Clinic, Phoenix, AZ, United States
- 4Department of Radiation Oncology, Mayo Clinic, Jacksonville, FL, United States
- 5Google Health, Mountain View, CA, United States
- 6Research Services, Comprehensive Cancer Center, Mayo Clinic, Rochester, MN, United States
- 7Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery, Mayo Clinic, Rochester, MN, United States
Introduction:
Organ-at-risk segmentation for head and neck cancer radiation therapy is a complex and time-consuming process (requiring up to 42 individual structure, and may delay start of treatment or even limit access to function-preserving care. Feasibility of using a deep learning (DL) based autosegmentation model to reduce contouring time without compromising contour accuracy is assessed through a blinded randomized trial of radiation oncologists (ROs) using retrospective, de-identified patient data.
Methods:
Two head and neck expert ROs used dedicated time to create gold standard (GS) contours on computed tomography (CT) images. 445 CTs were used to train a custom 3D U-Net DL model covering 42 organs-at-risk, with an additional 20 CTs were held out for the randomized trial. For each held-out patient dataset, one of the eight participant ROs was randomly allocated to review and revise the contours produced by the DL model, while another reviewed contours produced by a medical dosimetry assistant (MDA), both blinded to their origin. Time required for MDAs and ROs to contour was recorded, and the unrevised DL contours, as well as the RO-revised contours by the MDAs and DL model were compared to the GS for that patient.
Results:
- Mean time for initial MDA contouring was 2.3 hours (range 1.6–3.8 hours) and
- RO-revision took 1.1 hours (range, 0.4–4.4 hours),
- compared to 0.7 hours (range 0.1–2.0 hours) for the RO-revisions to DL contours.
- Total time reduced by 76% (95%-Confidence Interval: 65%-88%) and
- RO-revision time reduced by 35% (95%-CI,-39%-91%).
- All geometric and dosimetric metrics computed, agreement with GS was equivalent or significantly greater (p<0.05) for RO-revised DL contours compared to the RO-revised MDA contours, including volumetric Dice similarity coefficient (VDSC), surface DSC, added path length, and the 95%-Hausdorff distance.
- 32 OARs (76%) had mean VDSC greater than 0.8 for the RO-revised DL contours, compared to 20 (48%) for RO-revised MDA contours, and 34 (81%) for the unrevised DL OARs.
Conclusion:
- DL autosegmentation demonstrated significant time-savings for organ-at-risk contouring while improving agreement with the institutional GS, indicating comparable accuracy of DL model.
- Integration into the clinical practice with a prospective evaluation is currently underway.
https://www.frontiersin.org/articles/10.3389/fonc.2023.1137803/full
