Graph‐based risk assessment and error detection in radiation therapy

Abstract

Purpose The objective of this study was to formalize and automate quality assurance (QA) in radiation oncology. Quality assurance in radiation oncology entails a multistep verification of complex, personalized radiation plans to treat cancer involving an interdisciplinary team and high technology, multivendor software and hardware systems. We addressed the pretreatment physics chart review (TPCR) using methods from graph theory and constraint programming to study the effect of dependencies between variables and automatically identify logical inconsistencies and how they propagate. Materials and methods We used a modular approach to decompose the TPCR process into tractable units comprising subprocesses, modules and variables. Modules represented the main software entities comprised in the radiation treatment planning workflow and subprocesses grouped the checks to be performed by functionality. Module‐associated variables served as inputs to the subprocesses. Relationships between variables were modeled by means of a directed graph. The detection of errors, in the form of inconsistencies, was formalized as a constraint satisfaction problem whereby checks were encoded as logical formulae. The sequence in which subprocesses were visited was described in an activity diagram. Results The comprehensive model for the TPCR process comprised 5 modules, 19 subprocesses and 346 variables, 225 of which were distinct. Modules included “Treatment Planning System” and “Record and Verify System.” Subprocesses included “Dose Prescription,” “Documents,” “CT Integrity,” “Anatomical Contours,” “Beam Configuration,” “Dose Calculation,” “3D Dose Distribution Quality,” and “Treatment Approval.” Variable inconsistencies, and their source and propagation were determined by checking for constraint violation and through graph traversal. Impact scores, obtained through graph traversal, combined with severity scores associated with an inconsistency, allowed risk assessment. Conclusions Directed graphs combined with constraint programming hold promise for formalizing complex QA processes in radiation oncology, performing risk assessment and automating the TPCR process. Though complex, the process is tractable.

Unified risk analysis in radiation therapy

“…This is reasonable, since most errors occur during treatment planning [27] and physicists have a high workload in the entire therapy process [28] . For example Munbodh et al reported on graph-based risk assessment with respect to pretreatment physics chart review [17] . They used a directed graph to decompose the chart review into individual steps which were associated with variables.…”

“…There is also already a publication that describes graphical modeling of a workflow in the form of an activity diagram as a directed graph, which takes different process levels into account [17] . Munbodh et al describes the risks and workflow steps that occur during the pretreatment physics chart review (TPCR).…”

Automatic quality assurance of radiotherapy treatment plans using Bayesian networks: A multi-institutional study

“…Their ability to deal with missing values (which is a common phenomenon in RT datasets) in combination with the intuitiveness of their probabilistic reasoning, make BNs an ideal method for decision support in radiation oncology (18). Compared to rules-based algorithms and checklists that have been developed to assist treatment plan review both in-house and commercially (21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31), the BN has the advantage of mimicking human reasoning and adapting to changes in clinical practice by updating the network model with new data (11).…”