Clinical Performance Evaluation of RDX-Aneurysm for Detecting Intracranial Aneurysms on TOF-MRA

N/AActive Not RecruitingNCT07655635

Taipei Medical University Shuang Ho Hospital550 enrolled

Overview

This retrospective, multi-center, observational case-control study evaluates the standalone performance of RDX-Aneurysm, a computer-assisted detection software, for detecting and marking suspected saccular intracranial aneurysm locations on adult head time-of-flight magnetic resonance angiography (TOF-MRA). Approximately 550 evaluable examinations will be included, consisting of approximately 250 positive examinations with at least one reference-standard-confirmed saccular intracranial aneurysm measuring 3 mm or greater and approximately 300 negative examinations with no reference-standard-confirmed intracranial aneurysm. The primary outcome is lesion-level sensitivity for detecting reference-standard-confirmed saccular intracranial aneurysms measuring 3 mm or greater. The secondary outcome is false positives per scan. This study uses existing retrospective imaging data only; study outputs are not returned to clinical care and do not affect patient diagnosis, treatment, or follow-up.

Existing adult head TOF-MRA examinations from participating medical institutions will be selected according to pre-specified inclusion and exclusion criteria. A reference standard for aneurysm presence and lesion location will be established by qualified radiologists according to the study protocol. RDX-Aneurysm outputs will be generated on the locked retrospective image dataset and compared with the reference standard. The study evaluates only the standalone lesion-detection and marking function of the software. It does not evaluate triage, worklist prioritization, urgent notification, autonomous diagnosis, autonomous measurement or classification, rupture-risk prediction, treatment planning, or reader-performance improvement.

Study Type

OBSERVATIONAL

Enrollment

550

Conditions

Intracranial Aneurysm

Eligibility

Medical Language ↔ Plain English

Inclusion Criteria: 1. Adult subjects aged 18 years or older at the time of imaging. 2. Underwent head MRA, including TOF-MRA, as part of clinical care or voluntary health screening. 3. TOF-MRA images are available in DICOM format and meet the predefined study image-quality and technical requirements for retrospective analysis. Exclusion Criteria: 1. MRI-related contraindications documented in the available record or imaging context, including implanted pacemaker or defibrillator, cochlear implant, neurostimulator, intraocular or intracranial metallic foreign body, or other physician-determined MRI-unsuitable condition. 2. Clinically diagnosed intracranial hemorrhage, including intracerebral hemorrhage, subarachnoid hemorrhage, subdural hemorrhage, or epidural hemorrhage. 3. Clinically diagnosed arteriovenous malformation or arteriovenous fistula. 4. Prior intracranial aneurysm treatment, including surgical clipping or endovascular treatment, intracranial arterial stent placement, or craniotomy, where post-treatment structural changes may affect software interpretation. 5. Severe imaging artifact, incomplete image data, or image quality insufficient for reliable retrospective analysis. 6. Image data format or technical parameters not compatible with study analysis requirements.

Outcomes

Primary Outcomes

Lesion-level sensitivity for saccular intracranial aneurysms measuring 3 mm or greater

Lesion-level sensitivity is defined as the proportion of reference-standard-confirmed saccular intracranial aneurysm lesions measuring 3 mm or greater that are correctly detected and marked by RDX-Aneurysm. Sensitivity = TP / (TP + FN), where TP is a true-positive lesion-level detection and FN is a false-negative reference-standard-confirmed lesion not detected by the software.

Time frame: Through study completion, an average of 3 years

Secondary Outcomes

False positives per scan

False positives per scan is defined as the average number of software-marked suspected aneurysm locations per TOF-MRA examination that do not correspond to a reference-standard-confirmed intracranial aneurysm lesion.