Abstract TP171: Automated Large Vessel Occlusions Detection: Improved Diagnostic Accuracy in MCA M2 Segment Using Deep Learning
Bibliographic record
Abstract
Introduction: Integrating AI into clinical practice enables precise diagnosis and could potentially enhance outcomes for patients with large vessel occlusion (LVO). Although advancements in AI have shown promise in enhancing diagnostic accuracy, several studies have reported very low sensitivities when including MCA-M2 occlusions, resulting in unbalanced performance across different LVO segments. Therefore, continuous retraining and evaluation of these evolving technologies in clinical settings are essential. This study aims to analyze the real-world performance of the latest version of an FDA-approved and CE-marked AI software for LVO detection in computed tomography angiograms (CTAs), specifically designed to provide balanced performance across all LVO segments. Methods: This retrospective, multicenter, multinational, and blinded study analyzed anonymized CTA scans from 7 clinical sources across Europe and the USA, acquired using 40 scan models from 5 different CT vendors. The improved version of CINA-LVO (Avicenna.AI, La Ciotat, France) uses deep learning to accurately detect ICA, MCA-M1 and proximal MCA-M2 without impairment, specifically with MCA-M2 segments. Its diagnostic performance was assessed against the ground truth, which was determined through the consensus of three US board-certified senior neuroradiologists who analyzed each LVO segment. Results: A total of 557 scans (mean age 66.2 ± 20.5 [SD] years old, 50.3% female, 49.0% positive for LVO) were included. Overall sensitivity and specificity including M2 occlusions (all 557 cases) were 93.8% (95%CI: 90.2%-96.3%) and 91.2% (95%CI: 87.3%-94.2%), respectively. The PPV and NPV were 91.1% and 93.8%, respectively (see Table 1). Moreover, the detection accuracy for LVO exceeded 90% across all LVO segments, including the proximal and distal internal carotid artery and the M1 and proximal M2 segments of the middle cerebral artery (see Table 2). There were no statistically significant differences between the accuracy of each LVO segment (p-value>0.05). Conclusions: Unlike previous studies, our results demonstrate balanced performance across all LVO segments, with no impairment in diagnostic accuracy from the inclusion of M2 occlusions in the analysis. These promising findings suggest that the integration of continuously improved AI tools into clinical practice could substantially enhance clinical workflow and diagnostic accuracy for patients with LVO.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".