An observational study in Sinus Tarsi Abnormality, sponsored by The Third Affiliated Hospital of Southern Medical University. Completed at 1 site in China. Open to participants aged 18 Years to 50 Years, including healthy volunteers. Per ClinicalTrials.gov, last updated 2026-10-14.
Sponsored by The Third Affiliated Hospital of Southern Medical University · Observational
This study constructs a lightweight AI screening workflow based on sagittal fat-suppressed ankle MRI for automatic localization and normal-abnormal classification of sinus tarsi. With multi-slice feature fusion, it compares performances of different input strategies and lightweight networks. Multi-center datasets are used to validate model generalization. An MRMC (multi-reader multi-case) study is conducted to compare diagnostic performance between the model and radiologists. A web-based verification workflow is developed, and CPU-based runtime performance is evaluated to explore its clinical potential.
Based on sagittal fat-suppressed MRI of the ankle, this study builds a lightweight artificial intelligence-assisted screening workflow to realize automatic localization and normal-abnormal classification of the sinus tarsi region. It adopts automatic target localization to eliminate redundant background information and integrates local features of continuous MRI images via multi-slice feature fusion, while comparing the screening performance of different image input strategies and lightweight deep learning networks. Internal and external independent multi-center datasets are utilized to validate the stability and generalization ability of the model. Furthermore, a rigorous MRMC (multi-reader multi-case) study with eight radiology readers was performed on a total of 240 MRI examinations. All readers evaluated the full dataset in two independent reading sessions separated by a 28-day washout period. In the first session, each reader randomly interpreted 120 cases with AI assistance and the remaining 120 cases without AI assistance. In the second session, the AI assistance status of all cases was crossover-swapped, enabling each examination to be assessed by every reader under both AI-assisted and unassisted conditions. Meanwhile, a web-based visual verification workflow was established, and the operational efficiency and practical feasibility of the model under ordinary CPU environments were evaluated, aiming to explore the clinical application value of the proposed AI-assisted ankle MRI screening scheme.
The Third Affiliated Hospital of Southern Medical University is the lead sponsor of 11 studies on the registry; 5 are open to participants now.
Counted across the registry records on this site, refreshed daily.
Patients from multiple medical centers in China who underwent ankle MRI examinations due to ankle injury, pain or discomfort, representing patients with both normal and abnormal sinus tarsi findings.
Go through the criteria one by one — yes, no or not sure — and take the list to your doctor or the study team. Kept in this browser only.
Exclusion Criteria:
Retrospective observational validation cohorts were established using retrospective ankle MRI data collected from six independent centers after model locking for observational validation, with eligible cases included in the randomized crossover MRMC reader study.
Other: Standalone Radiologist Interpretation · Other: AI-assisted Radiologist Interpretation
Radiologists interpret the medical images independently without any assistance from the AI model to establish a baseline performance.
Radiologists adopt the proposed artificial intelligence model for auxiliary interpretation of the same set of medical images to quantitatively evaluate the improvement in diagnostic performance brought by the model.
Examination-level Sensitivity and Area Under the ROC Curve (AUC)
Evaluation of examination-level sensitivity (defined as the proportion of true-positive examinations among all examinations with positive reference-standard labels) and discrimination performance (measured by AUC) to compare unaided radiologist performance versus AI-assisted performance.
Time frame: Up to 1 week per evaluation period
Diagnostic Efficiency (Reading and Reporting Time)
Measurement of diagnostic efficiency recorded as the time (in seconds) taken by radiologists to complete the case review and generate findings, with and without AI assistance.
Time frame: Up to 1 week per evaluation period
Examination-level Specificity and Accuracy
Evaluation of examination-level specificity (defined as the proportion of true-negative examinations among all examinations with negative reference-standard labels) and accuracy (defined as the proportion of diagnostic decisions matching the reference-standard label) to compare unaided radiologist performance versus AI-assisted performance.
Time frame: Up to 1 week per evaluation period
Plan to share: No
No publications or documents are linked to this record.
From the registry record's own update history. This site started tracking changes on Sep 25, 2026; for anything earlier, see the record history on ClinicalTrials.gov ↗
This study is completed, as verified in Oct 2026. You cannot join it, but the record below documents what was studied.
Get an email when the registry record changes — status, dates, results — or when someone posts here.
Sign in to followQuestions and observations about this study, from anyone following it. Not medical advice, and not a channel to the study team — their contact details are on the registry record.
Sign in to join the discussion. Reading takes no account; posting does. You choose a display name, and a pseudonym is the default.
Nothing here yet. If you are running this trial, taking part in it, or weighing whether to, this is the place to say so.
The Third Affiliated Hospital of Southern Medical University