Density clustering links unusual radiotherapy QA signals to maintenance records
Source check: Source checked. Not independent scientific confirmation.
Finding: DBSCAN identified unusual changes in longitudinal imaging-detector QA measurements from nine radiotherapy accelerators. Several flagged output, beam-profile and apparent jaw-position changes coincided with documented maintenance or calibration. These are retrospective measurement associations, not proof that maintenance caused a physical fault or that the system detects clinically dangerous failures.
Technology: DBSCAN unsupervised density-based clustering
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- DBSCAN learned density structure in longitudinal QA changes and flagged sparse observations, enabling targeted linkage of unusual measured signals to service documentation rather than a fixed list of expected calibration effects.
- Test and result
- Retrospective 2018–2024 EPID QA measurements and service histories from nine TrueBeam accelerators across six institutions. DBSCAN fit dense/noise groups separately to one-dimensional changes in each machine/metric, with a dataset-specific neighborhood rule. In one machine, 15 jaw-position outliers included a 1.3 mm apparent shift near a documented calibration. Independently paired ion-chamber measurements supported selected output-change comparisons on only two machines/12 events, not all DBSCAN flags.
- Limitation / still unknown
- Temporal associations do not establish service causation or independently confirmed physical faults. Apparent jaw shifts after steering may be measurement artifacts. Acquisition timing and detector recalibration records were incomplete, and two machines lacked complete service histories. Clustering parameters were not optimized against prospective truth; noise-versus-cluster statistical comparisons are circular. No sensitivity/specificity, superiority to existing QA, safer treatment or patient benefit was established.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC13720238/
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 21 Nov 2026. Review scope: Read actual primary acquisition/service-data scope, DBSCAN operation and parameter rule, source-specific outlier/service Results, independent-comparison limits and Discussion. Original canonical primary publication date October 29 is distinguished from November collection/index date. No operational treatment instructions represented. Exact primary identity/date and fresh primary PubMed correction/retraction relationships checked..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 1/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.