Purpose: Despite the capability to perform remote monitoring and preventive maintenance for radiotherapy machines, treatments are often interrupted by unforeseen machine issues. In contrast to the process-based FMEA for clinical workflow analysis (i.e. AAPM TG100), a machinery FMEA (MFMEA) was devised for the machine issues and treatment interruptions recorded in an in-house electronic reporting system. The impact of these events on treatment schedule, patient safety, and operators’ competency in resolving these issues is assessed.
Methods: Reported machine issues were classified under five sub-systems: beam-generating system, beam-limiting, kV-imaging, control software, and patient couch. MFMEA was implemented with four modules to analyze the impact of a component failure in each of the sub-systems. MFEMA modules include (1) functional analysis – identifies components of each sub-system (e.g. MLC for beam-limiting); (2) failure analysis – describes functional failures (e.g. stuck MLC), impacts (e.g. treatment terminations) and causes (e.g. faulty MLC-motor); (3) risk analysis – describes the ease of detecting the failure; and (4) risk management – risk prioritization and preventive measures. Scoring for occurrence, severity, and detectability for each component failure were guided by data such as the frequency of error reported, duration of downtime, and detection rate through routine QA and from a remote monitoring software.
Results: Based on a total record of 215 issues reported over one year on one machine, 40% (kV-imaging), 29% (beam-limiting), 16% (beam-generating), 8% (control software), and 7% (patient couch) failures correspond to 20 component failures generated. The top three component failures (with their risk-priority-number) are – CBCT (424), couch positioner (214), and MLC (176).
Conclusion: The risks encountered with machine issues and interruptions have been analyzed for the first time using an MFEMA model with data from an electronic reporting system. Trainings and preventive measures for improving treatment efficiency are identified based on the risk model