Heart Valve Disease, Mitraclip, M-TEER, TAVI, TMVI, TriClip, T-TEER, TTVI
Conditions
Keywords
Medical imaging, CT, Artificial intelligence, M-TEER, T-TEER, TAVI, TMVI, valvulopathy, predictive algorithms
Brief summary
This non-interventional study aims to use artificial intelligence to improve the prediction of transcatheter heart valve interventions and optimize patient outcomes. It is based on the analysis of retrospective data from various specialized centers worldwide.
Detailed description
The ENVISAGE study is a non-interventional, retrospective research study designed to validate an artificial intelligence (AI)-based framework for the automated analysis of cardiac imaging data, including multi-slice cardiac computed tomography (CT) and transesophageal echocardiography (TEE). The primary objective is to predict the success of transcatheter heart valve interventions, including aortic, mitral, and tricuspid valve interventions (TAVI, TMVI, M-TEER, T-TEER). The AI framework developed in this study will rely on deep learning algorithms, particularly convolutional neural networks (CNNs) and other advanced models, to automatically segment critical anatomical structures and perform accurate measurements of these structures from CT and TEE images. These measurements will then be combined with pre-interventional clinical data to optimize patient selection and intervention planning, as well as to predict surgical outcomes with high accuracy. AI will also aim to reduce human error and inter-observer variability in the interpretation of cardiac images, which could significantly improve clinical outcomes.
Interventions
DIAGNOSTIC_TESTMedical imaging analysis via artificial intelligence algorithms
Development of AI algorithms based on pre-procedural imaging annotations and clinical informations to predict the transcatheter procedural outcomes
Sponsors
Centre Hospitalier Universitaire de Bordeaux, FRANCE
Materialise
Pie Medical Imaging
Clinique Pasteur Toulouse
Centre Cardiologique du Nord
CHU Rennes,France
San Raffaele University Hospital, Italy
Istituto clinico Città di Brescia
Unity Health Toronto
University Hospital, Marseille
Lille University Hospital
Hospitaux Universitaires Paris Sud
Montefiore Medical Center
University Medical Center Mainz
Universitätsklinikum Hamburg-Eppendorf
Vancouver Hospital
Montreal Heart Institute
Study design
Observational model
COHORT
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
Patients who have reached the age of legal majority under local laws. * For TAVI group: All patients who have had TAVI with a third generation transcatheter heart valve (THV), with an available pre-procedural optimal quality CT scan as defined by an ECG- gating CT with: 1. five to ten image volumes at cardiac phases from 5% to 95% R-R 2. 0.625 mm slice thickness 3. 0.625 mm spacing between slices 4. 0.88 mm in-plane pixel spacing * For TMVI group: Patients who have had a TMVI with a dedicated device and screen failures, with an available optimal quality CT scan. * For TTVI group: Patients who have had a TTVI with a dedicated device and screen failures, with an available optimal quality CT scan. * For M-TEER: All patient who have had a M-TEER with 1) G4 or newer iteration of MitraClip or 2) G2 or newer iteration of Pascal, with available pre-procedural TEE videos images from one of two vendors: Phillips or GE, with clear identifiable views of the Mitral valve, frame per second equal or higher than 40 frames per second, acceptable 3D reconstructions. * For T-TEER: All patient who have had a T-TEER with G4 or newer iteration of TriClip or 2) G2 or newer iteration of Pascal, with available pre-procedural TEE videos images from one of two vendors: Phillips or GE, with clear identifiable views of the Tricuspid valve, frame per second equal or higher than 40 frames per second, acceptable transgastric image with acceptable 3D reconstructions.
Exclusion criteria
* For TAVI group: Valve-in-valve procedures * For TMVI group: Valve-in-valve and valve-in-ring procedures * For TTVI: Valve-in-valve and valve-in-ring procedures * For M-TEER: G3 or older MitraClip, G1 Pascal * For T-TEER: G3 Triclip, G1 Pascal
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Accuracy of transcatheter AI predictions | Preoperative phase: automated segmentation and measurements compared with manual assessments; Postoperative phase at day 30: comparison of predicted results with actual clinical patient outcomes. | Validation of artificial intelligence algorithms for automatic segmentation of anatomic structures and imaging measurements, and prediction of the success of transcatheter interventions. Output of AI algorithm: * Sizes, types, and number of devices to be implanted * Device success * Percentage risk of permanent pacemaker implantation (for TAVI and TTVI) * Percentage risk of 30-day (para)valvular regurgitation for TAVI, and residual regurgitation for M-TEER and T-TEER * Single leaflet detachment for M-TEER and T-TEER * Left ventricular outflow tract obstruction for TMVI. Key success indicators: * First, independent retrospective validation dataset AI algorithms predict procedural outcome with \>90% accuracy and low inter-reader observer variability when compared to measured procedural outcome. * Second independent retrospective dataset, perform a study to validate AI algorithms with \>90% accuracy and low inter-reader observer variability when compared to measured procedural outcome. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Performance of AI algorithms in CT and TEE image analysis | Through study completion, an average of 2 years (retrospective analysis and validation of algorithms). | Development and evaluation of AI algorithm training platform for data analysis of patients undergoing transcatheter valve procedures. Comparison of AI model performance with existing benchmarks and manual analyses |
Other
| Measure | Time frame | Description |
|---|---|---|
| AI-based discovery of clinical knowledge for patient selection | Baseline (pre-procedural) and post-procedural (day 90) analysis | Comparison of AI's ability to standardize patient selection, treatment planning and clinical outcomes with traditional methods. Assessment of accuracy in predicting patient outcomes and reducing procedure-specific complications. |
Countries
Canada, France, Germany, Italy, United States
Contacts
Primary ContactThomas Modine, MD, PhD
Backup ContactWalid Ben Ali, MD, PhD
Outcome results
None listed