Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Jan;52(2):485-500.
doi: 10.1007/s00259-024-06922-4. Epub 2024 Sep 23.

Multi-modality artificial intelligence-based transthyretin amyloid cardiomyopathy detection in patients with severe aortic stenosis

Isaac Shiri  1 , Sebastian Balzer  1 , Giovanni Baj  1   2 , Benedikt Bernhard  1 , Moritz Hundertmark  1 , Adam Bakula  1 , Masaaki Nakase  1 , Daijiro Tomii  1 , Giulia Barbati  2 , Stephan Dobner  1 , Waldo Valenzuela  3 , Axel Rominger  4 , Federico Caobelli  4 , George C M Siontis  1 , Jonas Lanz  1 , Thomas Pilgrim  1 , Stephan Windecker  1 , Stefan Stortecky  1 , Christoph Gräni  5
Affiliations

Multi-modality artificial intelligence-based transthyretin amyloid cardiomyopathy detection in patients with severe aortic stenosis

Isaac Shiri et al. Eur J Nucl Med Mol Imaging. 2025 Jan.

Abstract

Purpose: Transthyretin amyloid cardiomyopathy (ATTR-CM) is a frequent concomitant condition in patients with severe aortic stenosis (AS), yet it often remains undetected. This study aims to comprehensively evaluate artificial intelligence-based models developed based on preprocedural and routinely collected data to detect ATTR-CM in patients with severe AS planned for transcatheter aortic valve implantation (TAVI).

Methods: In this prospective, single-center study, consecutive patients with AS were screened with [99mTc]-3,3-diphosphono-1,2-propanodicarboxylic acid ([99mTc]-DPD) for the presence of ATTR-CM. Clinical, laboratory, electrocardiogram, echocardiography, invasive measurements, 4-dimensional cardiac CT (4D-CCT) strain data, and CT-radiomic features were used for machine learning modeling of ATTR-CM detection and for outcome prediction. Feature selection and classifier algorithms were applied in single- and multi-modality classification scenarios. We split the dataset into training (70%) and testing (30%) samples. Performance was assessed using various metrics across 100 random seeds.

Results: Out of 263 patients with severe AS (57% males, age 83 ± 4.6years) enrolled, ATTR-CM was confirmed in 27 (10.3%). The lowest performances for detection of concomitant ATTR-CM were observed in invasive measurements and ECG data with area under the curve (AUC) < 0.68. Individual clinical, laboratory, interventional imaging, and CT-radiomics-based features showed moderate performances (AUC 0.70-0.76, sensitivity 0.79-0.82, specificity 0.63-0.72), echocardiography demonstrated good performance (AUC 0.79, sensitivity 0.80, specificity 0.78), and 4D-CT-strain showed the highest performance (AUC 0.85, sensitivity 0.90, specificity 0.74). The multi-modality model (AUC 0.84, sensitivity 0.87, specificity 0.76) did not outperform the model performance based on 4D-CT-strain only data (p-value > 0.05). The multi-modality model adequately discriminated low and high-risk individuals for all-cause mortality at a mean follow-up of 13 months.

Conclusion: Artificial intelligence-based models using collected pre-TAVI evaluation data can effectively detect ATTR-CM in patients with severe AS, offering an alternative diagnostic strategy to scintigraphy and myocardial biopsy.

Keywords: Aortic stenosis; Artificial intelligence; Radiomics; TAVI; Transthyretin amyloid cardiomyopathy.

PubMed Disclaimer

Conflict of interest statement

Declarations. Informed consent: Informed consent was obtained from all individual participants included in the study. Consent to participate: All procedures performed in studies involving human participants were in accordance with the ethical standard of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The study design was approved by the Bern cantonal ethics committee (ClinicalTrials.gov: NCT04061213), conducted in accordance with the Declaration of Helsinki, and study participants provided written informed consent before any data collection. Competing interest: Dr. Bernhard reports a career development grant from the Swiss National Science Foundation. Dr. Pilgrim reports research grants to the institution from Biotronik, Boston Scientific and Edwards Lifesciences; speaker fees from Biotronik, Boston Scientific, Abbott, and Metronic; Clinical event committee for study sponsored by HighLifeSAS. Dr. Federico Caobelli reports ongoing Grants supports from Siemens Healthineers and from the University of Bern, as well as speaker honoraria from Bracco AG, Siemens AG and Pfizer AG, all for matters not related to the present study. Dr. Dobner reports a research grant for the Bern amyloidosis registry (B-CARE) (NCT04776824) and the ATTR Amyloidosis in Elderly Patients With Aortic Stenosis study (NCT04061213) on behalf of Inselspital Bern from Pfizer, and acknowledges speaker fees and travel grants unrelated to the submitted work from Boehringer Ingelheim, Alnylam and Pfizer. Dr. Windecker reports research, travel or educational grants to the institution from Abbott, Abiomed, Amgen, Astra Zeneca, Bayer, Biotronik, Boehringer Ingelheim, Boston Scientific, Bristol Myers Squibb, Cardinal Health, CardioValve, Corflow Therapeutics, CSL Behring, Daiichi Sankyo, Edwards Lifesciences, Guerbet, InfraRedx, Janssen-Cilag, Johnson & Johnson, Medicure, Medtronic, Merck Sharp & Dohm, Miracor Medical, Novartis, Novo Nordisk, Organon, OrPha Suisse, Pfizer, Polares, Regeneron, Sanofi-Aventis, Servier, Sinomed, Terumo, Vifor, V-Wave. Dr. Windecker serves as advisory board member and/or member of the steering/executive group of trials funded by Abbott, Abiomed, Amgen, Astra Zeneca, Bayer, Boston Scientific, Biotronik, Bristol Myers Squibb, Edwards Lifesciences, Janssen, MedAlliance, Medtronic, Novartis, Polares, Recardio, Sinomed, Terumo, V-Wave and Xeltis with payments to the institution but no personal payments. He is also member of the steering/executive committee group of several investigator-initiated trials that receive funding by industry without impact on his personal remuneration. Dr. Stortecky reports research grants to the institution from Edwards Lifesciences, Medtronic, Boston Scientific and Abbott, as well as personal fees from Boston Scientific, Teleflex and BTG. Dr. Gräni received research funding from the GAMBIT foundation for this work. Dr. Stortecky reports research grants to the institution from Edwards Lifesciences, Medtronic, Boston Scientific and Abbott, as well as personal fees from Boston Scientific, Teleflex and BTG. Dr. Gräni further received funding from the Swiss National Science Foundation and Innosuisse, from the Center for Artificial Intelligence in Medicine Research Project Fund University Bern, outside of the submitted work. Dr. Bakula reports speaker fees and travel grants from Pfizer. Dr. Shiri reports speaker fees and travel grants from Alnylam Pharmaceuticals. Dr. Rominger and Dr. Caobelli are editors of European Journal of Nuclear Medicine and Molecular Imaging. All other authors report no conflicts. The remaining authors have nothing to disclose.

Figures

Fig. 1
Fig. 1
The flowchart of the current study represents the design of the study, starting the phases of data collection, preprocessing, model training, validation, and reporting phases utilized in the current study
Fig. 2
Fig. 2
Comparative analysis of different metrics, including Accuracy, AUC, Sensitivity, and Specificity for the best-performing models in each modality, evaluated across 100 iterations. Clinical: RFE + LR, Laboratory: UniVa + LR, ECG: RFE + AdaBoost, Echo: UniVa + SVM, Invasive Cath: MRMR + LR, Interventional Imaging: UniVa + LR, CT Non-Contrast Radiomics: RFE + LR, CT Diastolic Radiomics: UniVa + LR, CT Systolic Radiomics: UniVa + LR, CT Delta Radiomics: UniVa + LR, CT All Radiomics: UniVa + LR, CT Strain: RFE + LR, Multi-Modality: RFE + LR. Classifiers include Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), AdaBoost (AdaBo), and eXtreme Gradient Boosting (XGB). Feature selection methods featured are Recursive Feature Elimination (RFE), Univariate Analysis (UniVa), and Minimum Redundancy Maximum Relevance (MRMR)
Fig. 3
Fig. 3
ROC curve of best-performing models in each modality. Clinical: Strat. 1 (RFE + LR), Strat. 2 (UniVa + LR), Strat. 3 (MRMR + LR); Laboratory: Strat. 1 (RFE + LR), Strat. 2 (UniVa + LR), Strat. 3 (MRMR + LR); ECG: Strat. 1 (RFE + AdaBo), Strat. 2 (UniVa + AdaBo), Strat. 3 (MRMR + AdaBo); Echo: Strat. 1 (RFE + LR), Strat. 2 (UniVa + SVM), Strat. 3 (MRMR + LR); Invasive Cath: Strat. 1 (RFE + AdaBo), Strat. 2 (UniVa + LR), Strat. 3 (MRMR + LR); Interventional Imaging: Strat. 1 (RFE + LR), Strat. 2 (UniVa + LR), Strat. 3 (MRMR + LR); CT Non-Contrast Radiomics: Strat. 1 (RFE + LR), Strat. 2 (UniVa + LR), Strat. 3 (MRMR + LR); CT Diastolic Radiomics: Strat. 1 (RFE + LR), Strat. 2 (UniVa + LR), Strat. 3 (MRMR + LR); CT Systolic Radiomics: Strat. 1 (RFE + LR), Strat. 2 (UniVa + LR), Strat. 3 (MRMR + LR). Classifiers include Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), AdaBoost (AdaBo), and eXtreme Gradient Boosting (XGB). Feature selection methods featured are Recursive Feature Elimination (RFE), Univariate Analysis (UniVa), and Minimum Redundancy Maximum Relevance (MRMR)
Fig. 4
Fig. 4
CT Delta Radiomics: Strat. 1 (RFE + LR), Strat. 2 (UniVa + LR), Strat. 3 (MRMR + AdaBo); CT All Radiomics: Strat. 1 (RFE + LR), Strat. 2 (UniVa + LR), Strat. 3 (MRMR + LR); CT Strain: Strat. 1 (Manual + LR), Strat. 2 (RFE + LR), Strat. 3 (UniVa + AdaBo) Strat. 4 (MRMR + SVM); Multi-Modality: Strat. 1 (RFE + LR), Strat. 2 (UniVa + LR), Strat. 3 (MRMR + LR). Classifiers include Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), AdaBoost (AdaBo), and eXtreme Gradient Boosting (XGB). Feature selection methods featured are Recursive Feature Elimination (RFE), Univariate Analysis (UniVa), and Minimum Redundancy Maximum Relevance (MRMR)
Fig. 5
Fig. 5
Heat maps displaying various metrics for echocardiography (Echo), CT strain, and Multi-Modal data. Classifiers include Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), AdaBoost (AdaBo), and eXtreme Gradient Boosting (XGB). Feature selection methods featured are Recursive Feature Elimination (RFE), Univariate Analysis (UniVa), and Minimum Redundancy Maximum Relevance (MRMR)
Fig. 6
Fig. 6
SHAP (SHapley Additive exPlanations) summary plot displaying the impact of various features across Echo, CT Strain, and Multi-Modality models. This visualization highlights the contribution of individual features to each model’s predictive performance for ATTR-CM detection. LVMi: Left ventricular mass (end-diastolic) index to gr/m2, GLS apical/base-mid: Left ventricular global longitudinal strain apical /base and mid, LV-GLS: Left ventricular global longitudinal strain (%), LA-GLS: Left atrial Global longitudinal strain (%), Mean gradient aortic valve: Mean gradient aortic valve [mmHg], Maximum Septal Wall Thickness: maximum septal wall thickness of the left ventricle [mm], Peak Gradient: peak pressure gradient of the aortic valve [mmHg], Mean Gradient: mean pressure gradient of the aortic valve [mmHg], Biplanar Left Atrial ES volume: Biplanar LAESVi [ml / BSA in m2], LV mass index: left ventricular mass index LVMi [g/m2], LVEDP: estimated left ventricular end-diastolic pressure [mmHg], Maximum lateral wall thickness: maximum lateral wall thickness of the left ventricle [mm], NT-proBNP: N-terminal pro B-type natriuretic peptide [pg/ml]
Fig. 7
Fig. 7
Kaplan-Meier curves for all-cause mortality and cardiovascular mortality, stratified based on the median value of the outputs of the diagnostic model for ATTR-CM and the ground truth of ATTR-CM. The stratification categories are above and below the median ML output values, illustrating survival probabilities over time for each group with the x-axis in months
See this image and copyright information in PMC

References

    1. Alwan L, Benz DC, Cuddy SAM, Dobner S, Shiri I, Caobelli F, et al. Current and evolving Multimodality Cardiac Imaging in managing transthyretin amyloid cardiomyopathy. JACC Cardiovasc Imaging. 2024;17:195–211. 10.1016/j.jcmg.2023.10.010. - PubMed
    1. Pereyra Pietri M, Farina JM, Mahmoud AK, Scalia IG, Galasso F, Killian ME, et al. The prognostic value of artificial intelligence to predict cardiac amyloidosis in patients with severe aortic stenosis undergoing transcatheter aortic valve replacement. Eur Heart J - Digit Health. 2024. 10.1093/ehjdh/ztae022. - PMC - PubMed
    1. Fabbri G, Serenelli M, Cantone A, Sanguettoli F, Rapezzi C. Transthyretin amyloidosis in aortic stenosis: clinical and therapeutic implications. Eur Heart J Supplements. 2021;23:E128–32. 10.1093/eurheartj/suab107. - PMC - PubMed
    1. Castaño A, Narotsky DL, Hamid N, Khalique OK, Morgenstern R, DeLuca A, et al. Unveiling transthyretin cardiac amyloidosis and its predictors among elderly patients with severe aortic stenosis undergoing transcatheter aortic valve replacement. Eur Heart J. 2017;38:2879–87. 10.1093/eurheartj/ehx350. - PMC - PubMed
    1. Nitsche C, Scully PR, Patel KP, Kammerlander AA, Koschutnik M, Dona C, et al. Prevalence and outcomes of concomitant aortic stenosis and Cardiac Amyloidosis. J Am Coll Cardiol. 2021;77:128–39. 10.1016/j.jacc.2020.11.006. - PMC - PubMed

Supplementary concepts

LinkOut - more resources