Comparison of Planning Quality and Efficiency Between Conventional and Knowledge-Based Algorithms in Nasopharyngeal Cancer Patients Using Intensity Modulated Radiation Therapy

Abstract

Conclusion: Three percent of head and neck cancer patients who received radical RT developed grade 3 osteonecrosis. Tumor location and radiation dose are important factors that determine osteonecrosis, but modern RT technique may decrease the risk.

A method for a priori estimation of best feasible DVH for organs‐at‐risk: Validation for head and neck VMAT planning

“…To the best of our knowledge, the approach described here is unique as a whole and has never been implemented before. However, while the method is different, the results are largely determined by the same physics principles, and certain elements of the described algorithm by necessity overlap with the previous body of work on KBP . In particular, the geometry of the OAR in relation to the target is the major driver of the achievable OAR DVH.…”

“…Another approach is to strive toward more effective, computer‐aided automation. Plan automation is a popular topic today with several commercial strategies already available, such as (a) explicitly mimicking the steps taken by experienced users (Autoplanning, AP), (b) referencing the database of previous similar plans (knowledge‐based planning, KBP), and (c) developing plans based on multicriteria optimization (MCO) …”

“…As for KBP, although the details of the specific methods differ, a common theme is to build a database derived from previously designed plans for generally similar disease presentations, and from them predict the achievable results for any new patient dataset — a machine learning approach. This method was applied to quality control and, more recently, to the design of the plans. One challenge of KBP is that it requires careful selection of learning plans.…”

Heuristic knowledge‐based planning for single‐isocenter stereotactic radiosurgery to multiple brain metastases

“…The use of knowledge‐based planning to standardize, automate, and improve the quality of radiotherapy treatment planning has been well documented . Because knowledge‐based planning (KBP) relies on the correlation of anatomical features and dosimetric endpoints through multiple prior observations, the large degree of geometric variation in size and intracranial location of SRS targets presents a unique challenge for a KBP methodology .…”

“…Specifically, because all of the lesions and OARs are optimized together for single‐isocenter VMAT treatment planning, a multimet dose distribution can be accurately predicted by an appropriate combination of single‐target knowledge‐based SRS dose distribution predictions. As the knowledge base used in this approach does not actually come from a database of multimet cases, this approach upends standard KBP methods of setting aside training and validation sets from the same disease presentation . In this work, all available clinical multimet plans represent validation cases since none of them were explicitly used to train the knowledge‐based dose prediction system.…”