AI digital imaging and communications in medicine (DICOM) standardisation may help improve radiology workflow efficiency, with a retrospective study finding that implementation of an AI-aided metadata standardisation tool was associated with shorter reading times for several CT examinations.
The study focused on inconsistencies in DICOM metadata, the standardised information attached to medical images that supports image acquisition, storage, interpretation and transfer. Inconsistent metadata can hinder workflow efficiency, complicate artificial intelligence (AI) integration and create challenges for sharing imaging data between healthcare organisations.
Growing Imaging Demand Puts Pressure on Workflows
Radiology departments continue to face rising imaging volumes, increasing the need for tools that can streamline daily operations. While many AI applications in radiology have centred on pathology detection and triage, workflow-focused technologies have received comparatively less research attention.
Researchers evaluated a commercially available AI-aided hybrid software tool designed to standardise DICOM metadata. The system automatically classifies image characteristics including body part, imaging plane, laterality and contrast protocol, aiming to create consistent labels across imaging studies.
Accuracy Assessment Suggests Consistent Labelling
The retrospective cohort analysis assessed the tool’s accuracy using 422 computed radiography images and 1,503 CT series. Standardised labels were manually reviewed against a reference standard.
Across the evaluated categories, accuracy ranged from 83% to 100% for body-part classification, 91% to 100% for imaging plane classification and 88% to 100% for protocol classification. Based on these findings, the researchers reported that the tool adequately standardised DICOM labels.
Shorter Reading Times Observed After Implementation
Investigators also examined reading times before and after deployment of the software at a teleradiology provider serving multiple institutions, including around 40 NHS trusts.
The analysis included 10,966 examinations before implementation and 10,342 afterwards. Average reading times decreased significantly for CT abdomen, total body, head and temporal bone examinations, with reductions ranging from 0.73 to 2.9 minutes per case. Relative efficiency gains ranged from 8% to 22%.
Based on the observed reductions, the researchers estimated annual time savings of approximately 270 hours.
Results Do Not Establish Cause and Effect
No significant changes were observed for CT chest or sinus and orbit examinations.
The authors noted that the retrospective before-and-after design cannot determine the exact mechanism responsible for the reduction in reading times. While AI DICOM standardisation was linked to improved efficiency, the study did not directly assess whether metadata standardisation itself drove the observed gains.
The findings nevertheless suggest that standardised DICOM metadata could help streamline radiology workflows and improve image accessibility, potentially supporting wider use of AI tools in clinical imaging.
Reference
Tariq B et al. Evaluation of standardized DICOM labels assigned by a hybrid AI tool and its impact on radiologists’ reading times. Eur Radiol. 2026;DOI:10.1007/s00330-026-12670-w.
Featured image: Pangoasis on Adobe Stock
- Author:
