AI Science Tracker
EN

Learned MRI analysis distinguishes postanoxic coma from healthy controls in a small cohort

Toulouse III Paul Sabatier University; Inserm; CNRS; Toulouse University Hospital · 25 Jul 2022 · biology

Discovery

Learned MRI analysis distinguishes postanoxic coma from healthy controls in a small cohort

Source check: Source checked. Not independent scientific confirmation.

Finding: A three-dimensional convolutional network distinguished 29 postanoxic-coma patients from 34 healthy volunteers using structural and functional MRI measurements. Posterior-cingulate functional connectivity provided the strongest reported internal discrimination, outperforming the tested structural inputs. The result is a small clinical imaging feasibility study, without independently validated prediction of recovery or demonstrated bedside benefit.

Technology: Custom three-dimensional convolutional neural network for multimodal brain MRI classification

Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
The trained 3D CNN learned spatial features from patient and control MRI volumes, enabling the represented clinical-status discrimination and comparison of the informative content of different MRI inputs.
Test and result
A prospectively collected single-center cohort included 29 postanoxic-coma patients and 34 healthy volunteers. Separate structural and functional MRI inputs were evaluated with a custom 3D CNN using ten-times repeated tenfold cross-validation. Input intensity normalization used training-set values. Posterior-cingulate functional-connectivity inputs achieved reported mean accuracy 0.96, versus 0.82 for T1 structural input.
Limitation / still unknown
Small groups, a healthy-control contrast, one scanner/center and repeated internal cross-validation limit clinical specificity and generalization. Prospective collection is not an external locked-model evaluation. Convolutional activation maps were derived from correctly classified training samples and do not establish causal consciousness mechanisms. Post-hoc favorable outcomes among a few misclassified patients are not independent recovery prediction. No clinical deployment, prognosis-guided intervention or outcome benefit was demonstrated.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC9343298/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 21 Nov 2026. Review scope: Read actual primary patient/control and clinical reference selection, imaging input, 3D CNN architecture/training/normalization/cross-validation Methods, modality-comparison Results and interpretive limitations. Post-hoc recovery hints and training activation maps are not admitted as validated prognosis or causal mechanisms. Exact original article front DOI/title and fresh PubMed primary relationship metadata checked..

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 1/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open in the full catalog for filters, charts and related discoveries. Browse biology discoveries.