Automatic prediction of ASPECTS for patients with posterior circulation infarcts
Bibliographic record
Abstract
Background: The Alberta Stroke Program Early CT Score (ASPECTS) visually assesses the extent and location of ischemic core in Middle Cerebral Artery (MCA) ischemic strokes. Due to its simplicity, ASPECTS gained popularity and was adapted for diffusion-weighted MRIs (DWI). However, variability in human evaluation affects ASPECTS‘ effectiveness in treatment selection. Automatic classification methods that can replicate expert opinion might add an extra level of reproducibility and aid urgent medical decisions like thrombolysis or prognostic prediction. Objective: We previously developed a fully automatic system to calculate MCA-ASPECTS, matching consensus expert readings. This study extends the system to calculate ASPECTS in posterior circulation strokes (PCA-ASPECTS). Methods: The study included MRIs of patients admitted to our stroke center with ischemic stroke between 2009 and 2019. This subset of the public "Annotated Clinical MRIs and Linked Metadata of Patients with Acute Stroke" (https://www.icpsr.umich.edu/web/ICPSR/studies/38464). Stroke cores were manually traced. Models identifying ischemic regions were trained on 1,414 MRIs with ischemic evidence in DWI and tested on an independent dataset of 160 patients with posterior circulation infarcts. The models‘ efficiency was compared to consensus expert evaluations and inter-evaluator comparisons. Results: The Random Forest model was the most efficient for classifying stroke locations in the posterior circulation. Balanced accuracy (0-1 scale, with 1 being perfect) was: PCA 0.87; cerebellum 0.92; midbrain 0.84; pons 0.84; thalamus 0.87. This performance rivaled inter-evaluator classification. The model‘s accuracy in predicting unilateral or bilateral lesions in the PCA, cerebellum, and midbrain averaged 0.9. Accuracy for predicting PCA-ASPECTS was 0.65, and 0.85 for "flexible" ASPECTS (±1 point tolerance). We developed a fully automated system for calculating PCA-ASPECTS with accuracy comparable to inter-evaluator agreement. While accuracy for total score calculation was slightly lower due to multiclass issues, accuracy for predicting individual affected regions and flexible ASPECTS rivaled or exceeded inter-evaluator agreement. This system is part of our Acute Stroke Detection and Segmentation tool, ADS (https://www.nitrc.org/projects/ads), which outputs digital infarct masks, proportions of brain regions injured, automatic radiology reports, and MCA-ASPECTS, all with explanatory features. ADS is public, free, accessible to non-experts, and runs in real time on local CPUs with minimal computational requirements. Conclusion: We developed an automated and accessible system to accurately predict PCA-ASPECTS, potentially enhancing acute stroke assessment and supporting urgent medical decisions and reproducible research.
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 | 0.008 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 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 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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".