Heart failure & cardiomyopathiesDiagnostic or prognostic study
Leveraging a Vision Transformer Model to Improve Diagnostic Accuracy of Cardiac Amyloidosis With Cardiac Magnetic Resonance
- Diagnostic accuracy study
- retrospective cohort
- external validation
Authors' conclusion
A ViT model using only cine and late gadolinium enhancement CMR images can achieve high accuracy in differentiating CA from other underlying etiologies of suspected cardiomyopathy, especially in cases when reported human diagnostic confidence was uncertain in both a large single state health system and in an external CA cohort.
Study context
- Population
- Patients referred for cardiac magnetic resonance (CMR) for suspicion of infiltrative disease or hypertrophic cardiomyopathyexternal cohort n = 157
- Participants
- 807 patients
- Comparison
- Vision transformer (ViT) deep learning model for discrimination of cardiac amyloidosis on CMR
- Outcome
- Diagnostic accuracy and AUC for CA
Performance
As reported in the abstract
Discrimination of CA
| Test or model | Accuracy (internal test set) | AUC (internal test set) | Accuracy (external test set) | AUC (external test set) |
|---|---|---|---|---|
| ViT model | 84.1% | 0.954 | 82.8% | 0.957 |
Further findings
Accuracy by clinical report confidence of cardiac amyloidosis (CA) diagnosis (internal cohort)
- Moderate/high confidence diagnosis of CA: accuracy (55 of 61)
- 90%
- Uncertain, missing or incorrect diagnosis of CA: accuracy (22 of 36)
- 61.1%
Accuracy after removing studies with poor image quality, dual pathologies or ambiguous CA diagnosis
- Deep learning (DL) accuracy of this cohort increased to
- 79.1%
Original publication
Leveraging a Vision Transformer Model to Improve Diagnostic Accuracy of Cardiac Amyloidosis With Cardiac Magnetic Resonance. JACC Cardiovasc Imaging. Published 4 December 2024. doi:10.1016/j.jcmg.2024.09.010
Summarised from the published abstract; every number is checked against it