Learned artifact localization supports a bounded radiotherapy-dose comparison
Source check: Source checked. Not independent scientific confirmation.
Finding: An Inception-V3 classifier localized dental artifacts in CT images. In fifteen separate head-and-neck treatment plans, its outputs guided comparisons of artifact-management choices; calculated target-coverage differences stayed below 0.25 percentage points. This is a retrospective planning result, not measured delivered-dose equivalence or evidence of safer treatment.
Technology: Inception-V3 convolutional neural network
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- The learned slice classifier identifies locations used to define the compared artifact-management regions.
- Test and result
- Classifier development used 549 patients with patient-separated partitions. Fifteen other plans supplied comparisons of calculated dose after model-guided artifact localization.
- Limitation / still unknown
- One institution and planning system; no independent physical dosimetry, clinical outcomes or general treatment recommendation. Dose calculations are comparisons between modeled management scenarios.
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/PMC8180493/
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: Indexed primary full clinical/model Methods and dosimetric Results/discussion; exact front identity/date and current primary notice metadata..
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 2/5 · 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.