Academic Courses > DOS 711 > Machine Learning for Thoracic IMRT Planning

DOS 711 - Week 9 Discussion

Writing Prompt

Using the library resources - specifically the radiation therapy journal collection - locate a peer reviewed research article of interest and identify the problem statement and purpose of the study. Provide the problem statement and purpose statement here and attach a copy of the article to your post.

Initial Post: Machine Learning for Thoracic IMRT Planning

In the article "Automatic Learning-Based Beam Angle Selection for Thoracic IMRT" by Amit et al,1 the authors note that effective intensity modulated radiation therapy (IMRT) planning requires careful selection of beam angles to avoid placing excess dose in critical organ at risk (OR) structures while still maintaining adequate coverage of the treatment volumes. The identified problem is that thoracic treatments do not have class solutions because of the multitude of possible tumor sizes and locations; and anatomy configurations. As such, beam angle choice is left to an experienced human who can balance the complex interplay of dose objectives and anatomic variations. The quality of this process is then based on the planner's clinical experience.

The purpose of the study is to determine whether it is possible to construct a computationally efficient machine learning system that can analyze a set of existing thoracic IMRT plans and learn how to predict sets of beam angles similar to what an experienced human would choose. Such a system would help reduce the burden of choosing beam angles for thoracic cases, and may enable less experienced planners to produce higher quality plans.

Grayden, Chicago

  1. Amit G, Purdie TG, Levinshtein A, Hope AJ, Lindsay P, Marshall A, Jaffray DA, Pekar V. Automatic learning-based beam angle selection for thoracic IMRT. Med Phys. 2015;42(4):1992. http://dx.doi.org/10.1118/1.4908000
Academic Courses > DOS 711 > Machine Learning for Thoracic IMRT Planning
Written April 8, 2015
Second Semester, 3 Months into Internship