Unified risk analysis in radiation therapy

Abstract

Purpose: The increasing complexity of new treatment methods as well as the Information Technology (IT) infrastructure within radiotherapy require new methods for risk analysis. This work presents a methodology on how to model the treatment process of radiotherapy in different levels. This subdivision makes it possible to perform workflow-specific risk analysis and to assess the impact of IT risks on the overall treatment workflow.

Methods: A Unified Modeling Language (UML) activity diagram is used to model the workflows. The subdivision of the workflows into different levels is done with the help of swim lanes. The model created in this way is exported in an xml-compatible format and stored in a database with the help of a Python program.

Results: Based on an existing risk analysis, the workflows CT Appointment, Glioblastoma Multiforme, and Deep Inspiration Breath Hold (DIBH) were modeled in detail. Part of the analysis are automatically generated workflow-specific risk matrices including risks of medical devices incorporated into a specific workflow. In addition, SQL queries allow to quickly retrieve e.g., the details of the medical device network installed in a department.

Conclusion: Activity diagrams of UML can be used to model workflows in radiotherapy. Through this, a connection between the different levels of the entire workflow can be established and workflow-specific risk analysis is possible.

Keywords: Activity diagram; IEC 80001-1; Risk analysis; UML.

Figure 1

Workflows are represented by arrows and labeled with their name above. The swim lanes on the left represent the workflow levels. The yellow boxes are activities that can take different forms. The green boxes are requirements that take the form of risks. Each workflow has a start and an end point. ${\mathit{MWS}}_i$, $P_i$, etc. are unique identifiers that e.g. allow using an activity or requirement in multiple workflows. Note, that once the workflow reaches the deepest level of the nested diagram it recursively returns to the next higher activity.

Figure 2

Overview of the main therapy level of the workflow. All steps of the workflow are shown in the top level.

Figure 3

Software architecture to read out the XMI from the model. (a) Enterprise Architect: Software to model the workflows as activity diagrams, (b) xmi: Format in which the modeled diagram is being exported, (c) JIRA: Platform in which all risks are being recorded, (d) Database: MySQL database which stores the data from the xmi in a structured format.

Figure 4

Section of the overall workflow for the imaging for treatment planning step with the workflow default and two sample workflows. The different levels of the swim lane are shown on the left. The figure reads from top to bottom and from left to right. The yellow boxes are workflow steps, the green boxes are risks that belong to the workflow steps. Note, that once the workflow reaches the deepest level of the nested diagram it recursively returns to the next higher activity.

Figure 5

Step of medical treatment planning for workflows default and glioblastoma multiforme. Notation is adopted from .

Figure 6

Workflow and equipment used for the workflows DIBH and default in the step of irradiating a patient. All steps on the Procedure level are performed in inspiration in case of workflow DIBH. Note, that once the workflow reaches the deepest level of the nested diagram it recursively returns to the next higher activity.

Figure 7

Risk matrix, for the workflow default (a) and workflow glioblastoma multiforme (b). Shown is the Occurrence and Severity for the net risk assessment (i.e. after implementation of measures – see for details).