VALIDATION OF AUTOMATED ASPECTS SOFTWARE FOR DETECTION OF EARLY ISCHEMIC BRAIN CHANGES ON NON-CONTRAST CT SCANS
Bibliographic record
Abstract
Background and Purpose: In the Alberta Stroke Program Early Computed Tomography Score (ASPECTS), 10 brain regions are dichotomously scored on presence of ischemic stroke damage. However, considerable inter- and intra-reader variability exists, even for expert readers. We evaluated computed ASPECTS (ASPECTS 2.0.1, Frontier, Siemens Healthineers, Forchheim, Germany) in comparison to expert readers.Methods: Each included baseline non-contrast CT-scan (5 mm slice thickness) from the MR CLEAN trial (n=463) was evaluated by four expert readers for manual ASPECTS.Two observers provided a two-observer consensus for ASPECTS-regions (normal/abnormal) as ground truth for training and testing (0.2/0.8 division). The remaining two observers where used to provide individual ASPECTS-region scores.A Frontier ASPECTS region score specificity of u226590% was used to determine the software threshold (relative density difference between affected and contralateral region). Sensitivity, specificity and receiver-operating characteristic curves were calculated. Thereafter, we calculated ICC[1,1], agreement per region and trichotomized ASPECTS (0-4, 5-7, 8-10) for Frontier ASPECTS and expert readers, and between the expert readers in the test set.Results: A subset (n=459/463) was included. In the training set (n=104), a threshold of 4.7-5.6% was found for a specificity of u226590%, resulting in a sensitivity and area under the curve of 33-49% and 0.741-0.785. In the test set (n=355) the corresponding results were 89-89%, 41-57% and 0.750-0.795, respectively. Comparison of ground truth with other observers resulted in an ICC of 0.383-0.464, ASPECTS region agreement of 0.77-0.81 and trichotomized ASPECTS agreement of 0.57-0.60 Comparison of ground truth with Frontier ASPECTS resulted in an ICC of 0.537, ASPECTS region agreement of 0.78 and trichotomized ASPECTS agreement of 0.60.Conclusions: The performance of Frontier ASPECTS is comparable to expert readers.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.000 |
| Bibliometrics | 0.027 | 0.014 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.002 | 0.008 |
| Open science | 0.007 | 0.003 |
| 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; both teacher heads agree on what is shown here.
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".