Heart failure & cardiomyopathiesDiagnostic or prognostic study
Deep Learning on Bone Scintigraphy to Detect Abnormal Cardiac Uptake at Risk of Cardiac Amyloidosis
- Diagnostic accuracy study
- 5-fold cross-validation
- external validation
Authors' conclusion
The authors' detection model is effective at identifying patients with cardiac uptake Perugini grade ≥2 on WBS and may help in the diagnosis of patients with cardiac amyloidosis.
Study context
- Population
- Technetium-99m whole-body scintigraphies (whole-body scintigraphy (WBS)) from hospital databases; detection of significant cardiac uptake (Perugini grade ≥2) to retrieve patients at risk of cardiac amyloidosisimages
- Participants
- 3,048 vs 1,633
- Comparison
- 5-fold cross-validation vs External validation
- 5-fold cross-validationn = 3,048
- External validationn = 1,633
- Outcome
- Performance (C-statistics: sensitivity, specificity, AUC) of a deep learning model (convolutional neural network) detecting cardiac uptake Perugini grade ≥2 on WBS
Performance
As reported in the abstract
Detection of cardiac uptake Perugini grade ≥2
| Test or model | Sensitivity | Specificity | AUC |
|---|---|---|---|
| 5-fold cross-validation | 98.9% (± 1.0) | 99.5% (± 0.4) | 0.999 (SD = 0.000) |
| External validation | 96.1% | 99.5% | 0.999 |
Further findings
Performance by sex, age <90 years, BMI, injection-acquisition delay, radionuclides and WBS indication
- Only slightly affected performances
Original publication
Deep Learning on Bone Scintigraphy to Detect Abnormal Cardiac Uptake at Risk of Cardiac Amyloidosis. JACC Cardiovasc Imaging. Published 12 April 2023. doi:10.1016/j.jcmg.2023.01.014
Summarised from the published abstract; every number is checked against it