MétaCan
Menu
← Back to cohort
Record W4312964864 · doi:10.1161/atvb.42.suppl_1.347

Abstract 347: Using Deep Convolutional Neural Networks To Automate Classification Of Carotid Plaques From Ultrasound Imaging

2022· article· en· W4312964864 on OpenAlexaff
Nitish Bhatt, Rashmi Nedadur, Blair E. Warren, Sebastian Mafeld, Sneha Raju, Jason E. Fish, Bo Wang, Kathryn L. Howe

Bibliographic record

VenueArteriosclerosis Thrombosis and Vascular Biology · 2022
Typearticle
Languageen
FieldMedicine
TopicCerebrovascular and Carotid Artery Diseases
Canadian institutionsUniversity Health NetworkUniversity of Toronto
Fundersnot available
KeywordsInterpretabilityMedicineConvolutional neural networkStroke (engine)Carotid arteriesArtificial intelligenceUltrasoundRadiologyRisk stratificationRisk assessmentPattern recognition (psychology)Internal medicineComputer science

Abstract

fetched live from OpenAlex

Background: Stroke is a devastating consequence of plaque rupture from the carotid arteries. Current management of carotid plaques involves waiting for symptoms (e.g., stroke or mini-stroke), as intervention itself has risk of stroke and not all plaques are vulnerable to rupture. There is a need to better risk-stratify plaque that causes stroke. Carotid ultrasound (US) is a non-invasive and inexpensive visualization of plaques but is limited by human interpretation. We hypothesize that convolutional neural networks (CNNs) will identify unique features of carotid plaques for automated risk stratification. Methods: Our workflow is illustrated in Fig. a. A total of 141 B-mode US images of carotid arteries were included; 64 high-risk with symptomatic carotid plaques and 75 low-risk with no significant plaque. Data was cropped and divided into training (70%) and holdout test (30%) subsets. During model training, an ensemble of ResNet-18 CNNs learned classification of low-risk and high-risk cases using five-fold stratified cross validation and was used to predict on the holdout test set. The model was evaluated using ROC-AUC and sensitivity. Saliency maps were used for model interpretability to highlight relevant pixels for model decisions. Results: The cross-validation AUC was 0.995 ± 0.010. The testing AUC was 0.909 and class-wise sensitivities were 0.88 (low-risk) and 0.79 (high-risk). The density plot (Fig. b) shows that the classifier correctly identifies both classes with confidence. Model interpretability using saliency maps (Fig. c) shows pixels corresponding to carotid artery vessel edges and in high-risk cases, carotid plaques. Conclusions: Using this proof-of-concept model, carotid US long axis images are sufficient to identify high-risk plaques in symptomatic patients - we now need to determine whether we can identify high-risk plaques before symptoms to prevent devastating stroke caused by carotid disease.

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 imitation

Not 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.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.002
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Observational · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.021
Threshold uncertainty score0.042

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0010.002
Meta-epidemiology (narrow)0.0020.001
Meta-epidemiology (broad)0.0010.001
Bibliometrics0.0010.001
Science and technology studies0.0000.000
Scholarly communication0.0010.001
Open science0.0010.001
Research integrity0.0010.001
Insufficient payload (model declined to judge)0.0030.002

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.

Opus teacher head0.032
GPT teacher head0.285
Teacher spread0.253 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designObservational
Domainnot available
GenreEmpirical

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".

Quick stats

Citations0
Published2022
Admission routes1
Has abstractyes

Explore more

Same venueArteriosclerosis Thrombosis and Vascular Biology→Same topicCerebrovascular and Carotid Artery Diseases→French-language works237,207→