David P. Gierga, PhD
Massachusetts General Hospital, Boston

Automated treatment planning is on the way

The goal of treatment planning in external beam radiotherapy is to implement the radiation oncologist’s prescription in the most optimal way.  Treatment planning requires sophisticated tools but, for the sizable part, it is a manual, time-consuming process requiring multiple iterations to reach the desired goal.  While some methods, such as multi-criteria optimization (Craft et al 2007), have been introduced to provide an optimal treatment plan with fewer iterations, significant effort is still required to progress through the treatment planning workflow.

Automated treatment planning has been proposed as a method to increase efficiency, ensure consistent plan quality, or to simply enable treatment planning in resource limited settings.  One example of automated planning, knowledge-based planning, utilizes information from prior plans to predict the dose for future individual patients, based on their specific anatomy (Appenzoller et al 2012).  Recently, knowledge-based planning has been used as a quality control tool for clinical trials and resulted in improved normal tissue sparing (Li et al 2017).  Knowledge-based planning has also been used to automatically generate plans for stereotactic radiosurgery (Ziemer et al 2017).  More recently, Kisling et al have developed a Radiation Planning Assistant, specifically developed for limited resource settings, to automatically generate plans for post-mastectomy radiation therapy.

Machine learning will also likely be utilized for a higher degree of automated treatment planning in the near future, with recent examples published by Nguyen et al for dose prediction for prostate and head and neck cancer patients.  Finally, a recent publication by Zarepisheh et al successfully demonstrated that hierarchical constrained optimization can also be used for automatic intensity modulated radiation therapy (IMRT) treatment planning.  In summary, multiple techniques have been developed for automated treatment planning, and will likely become prevalent within the next few years.

References:

1. Craft, T. Halabi, H.A. Shih, T. Bortfeld. An approach for practical multi-objective IMRT treatment planning. Int J Radiat Oncol Biol Phys, 69(5), 2007.
2. M. Appenzoller, J.M. Michalski, W.L. Thorstad, S. Mutic, K.L. Moore. Predicting dose-volume histograms for organs-at-risk in IMRT planning.  Med Phys 39(12), 2012.
3. Li, R. Carmona, I. Sirak, L. Kasaova, D. Followill, J. Michalski, W. Bosch, W. Straube, L.K. Mell, K.L. Moore. Highly Efficient Training, Refinement, and Validation of a Knowledge-based Planning Quality-Control System for Radiation Therapy Clinical Trials. Int J Radiat Oncol Biol Phys 97(1), 2017.
4. Ziemer, S. Shiraishi, J.A. Hattangadi-Gluth, P. Sanghvi, K.L. Moore. Fully automated, comprehensive knowledge-based planning for stereotactic radiosurgery: Preclinical validation through blinded physician review. Pract Rad Onc 7(6), 2017.
5. Kisling, L. Zhang, S. F. Shaitelman, D. Anderson, T. Thebe, J. Yang, P.A. Balter, R.M. Howell, A. Jhingran, K. Schmeler, H. Simonds, M. du Toit, C. Trauernicht, H. Burger, K. Botha, N. Joubert, B.M. Beadle, L. Court. Automated treatment planning of postmastectomy radiotherapy.  Med Phys 46(9), 2019.
6. Nguyen, X. Jia, D. Sher, M.-H. Lin, Z. Iqbal, H. Liu, S. Jiang. 3D radiotherapy dose prediction on head and neck cancer patients with a hierarchically densely connected U-net deep learning architecture. Phys Med Biol, 64(6), 2019.
7. Nguyen, T. Long, X. Jia, W. Lu, X. Gu, Z. Iqbal, S. Jiang. A feasibility study for predicting optimal radiation therapy dose distributions of prostate cancer patients from patient anatomy using deep learning. Scientific reports, 9(1), 2019.
8. Zarepisheh, L. Hong, Y. Zhou, J.H. Oh, J.G. Mechalakos, M.A. Hunt, G.S. Mageras, J.O. Deasy. Automated intensity modulated treatment planning: The expedited constrained hierarchical optimization (ECHO) system. Med Phys 46(7), 2019.