A CT-based deep learning system for automatic assessment of aortic root morphology for TAVI planning

A fully automated artificial intelligence-driven software for planning of transcatheter aortic valve replacement

BACKGROUND

Transcatheter aortic valve replacement (TAVR) is increasingly performed for the treatment of aortic stenosis. Computed tomography (CT) analysis is essential for pre-procedural planning. Currently available software packages for TAVR planning require substantial human interaction. We describe development and validation of an artificial intelligence (AI) powered software to automatically rend anatomical measurements and other information required for TAVR planning and implantation.

METHODS

Automated measurements from 100 CTs were compared to measurements from three expert clinicians and TAVR operators using commercially available software packages. Correlation coefficients and mean differences were calculated to assess precision and accuracy.

RESULTS

AI-generated annular measurements had excellent agreements with manual measurements by expert operators yielding correlation coefficients of 0.97 for both perimeter and area. There was no relevant bias with a mean difference of -0.07 mm and - 1.4 mm2 for perimeter and area, respectively. For the ascending aorta measured 5 cm above the annular plane, correlation coefficient was 0.95 and mean difference was 1.4 mm. Instruction for use-based sizing yielded agreement with the effective implant size in 87-88 % of patients for self-expanding valves (perimeter-based sizing) and in 88 % for balloon-expandable valves (area-based sizing).

CONCLUSIONS

A fully automated software enables accurate and precise anatomical segmentation and measurements required for TAVR planning without human interaction and with high reliability.

Angulation and curvature of aortic landing zone affect implantation depth in transcatheter aortic valve implantation

In transcatheter aortic valve implantation (TAVI), final device position may be affected by device interaction with the whole aortic landing zone (LZ) extending to ascending aorta. We investigated the impact of aortic LZ curvature and angulation on TAVI implantation depth, comparing short-frame balloon-expanding (BE) and long-frame self-expanding (SE) devices. Patients (n = 202) treated with BE or SE devices were matched based on one-to-one propensity score. Primary endpoint was the mismatch between the intended (HPre) and the final (HPost) implantation depth. LZ curvature and angulation were calculated based on the aortic centerline trajectory available from pre-TAVI computed tomography. Total LZ curvature ( k L Z , t o t ) and LZ angulation distal to aortic annulus ( α L Z , D i s t a l ) were greater in the SE compared to the BE group (P < 0.001 for both). In the BE group, HPost was significantly higher than HPre at both cusps (P < 0.001). In the SE group, HPost was significantly deeper than HPre only at the left coronary cusp (P = 0.013). At multivariate analysis, α L Z , D i s t a l was the only independent predictor (OR = 1.11, P = 0.002) of deeper final implantation depth with a cut-off value of 17.8°. Aortic LZ curvature and angulation significantly affected final TAVI implantation depth, especially in high stent-frame SE devices reporting, upon complete release, deeper implantation depth with respect to the intended one.

A fully automated deep learning approach for coronary artery segmentation and comprehensive characterization

Coronary computed tomography angiography (CCTA) allows detailed assessment of early markers associated with coronary artery disease (CAD), such as coronary artery calcium (CAC) and tortuosity (CorT). However, their analysis can be time-demanding and biased. We present a fully automated pipeline that performs (i) coronary artery segmentation and (ii) CAC and CorT objective analysis. Our method exploits supervised learning for the segmentation of the lumen, and then, CAC and CorT are automatically quantified. 281 manually annotated CCTA images were used to train a two-stage U-Net-based architecture. The first stage employed a 2.5D U-Net trained on axial, coronal, and sagittal slices for preliminary segmentation, while the second stage utilized a multichannel 3D U-Net for refinement. Then, a geometric post-processing was implemented: vessel centerlines were extracted, and tortuosity score was quantified as the count of branches with three or more bends with change in direction forming an angle >45°. CAC scoring relied on image attenuation. CAC was detected by setting a patient specific threshold, then a region growing algorithm was applied for refinement. The application of the complete pipeline required <5 min per patient. The model trained for coronary segmentation yielded a Dice score of 0.896 and a mean surface distance of 1.027 mm compared to the reference ground truth. Tracts that presented stenosis were correctly segmented. The vessel tortuosity significantly increased locally, moving from proximal, to distal regions (p < 0.001). Calcium volume score exhibited an opposite trend (p < 0.001), with larger plaques in the proximal regions. Volume score was lower in patients with a higher tortuosity score (p < 0.001). Our results suggest a linked negative correlation between tortuosity and calcific plaque formation. We implemented a fast and objective tool, suitable for population studies, that can help clinician in the quantification of CAC and various coronary morphological parameters, which is helpful for CAD risk assessment.

TAVI-PREP: A Deep Learning-Based Tool for Automated Measurements Extraction in TAVI Planning

BACKGROUND

Transcatheter aortic valve implantation (TAVI) is a less invasive alternative to open-heart surgery for treating severe aortic stenosis. Despite its benefits, the risk of procedural complications necessitates careful preoperative planning.

METHODS

This study proposes a fully automated deep learning-based method, TAVI-PREP, for pre-TAVI planning, focusing on measurements extracted from computed tomography (CT) scans. The algorithm was trained on the public MM-WHS dataset and a small subset of private data. It uses MeshDeformNet for 3D surface mesh generation and a 3D Residual U-Net for landmark detection. TAVI-PREP is designed to extract 22 different measurements from the aortic valvular complex. A total of 200 CT-scans were analyzed, and automatic measurements were compared to the ones made manually by an expert cardiologist. A second cardiologist analyzed 115 scans to evaluate inter-operator variability.

RESULTS

High Pearson correlation coefficients between the expert and the algorithm were obtained for most parameters (0.90-0.97), except for left and right coronary height (0.8 and 0.72, respectively). Similarly, the mean absolute relative error was within 5% for most measurements, except for left and right coronary height (11.6% and 16.5%, respectively). A greater consensus was observed among experts than when compared to the automatic approach, with TAVI-PREP showing no discernable bias towards either the lower or higher ends of the measurement spectrum.

CONCLUSIONS

TAVI-PREP provides reliable and time-efficient measurements of the aortic valvular complex that could aid clinicians in the preprocedural planning of TAVI procedures.

CardioVision: A fully automated deep learning package for medical image segmentation and reconstruction generating digital twins for patients with aortic stenosis

Aortic stenosis (AS) is the most prevalent heart valve disease in western countries that poses a significant public health challenge due to the lack of a medical treatment to prevent valve calcification. Given the aging population demographic, the prevalence of AS is projected to rise, resulting in a progressively significant healthcare and economic burden. While surgical aortic valve replacement (SAVR) has been the gold standard approach, the less invasive transcatheter aortic valve replacement (TAVR) is poised to become the dominant method for high- and medium-risk interventions. Computational simulations using patient-specific models, have opened new research avenues for optimizing emerging devices and predicting clinical outcomes. The traditional techniques of generating digital replicas of patients' aortic root, native valve, and calcification are time-consuming and labor-intensive processes requiring specialized tools and expertise in anatomy. Alternatively, deep learning models, such as the U-Net architecture, have emerged as reliable and fully automated methods for medical image segmentation. Two-dimensional U-Nets have been shown to produce comparable or more accurate results than trained clinicians' manual segmentation while significantly reducing computational costs. In this study, we have developed a fully automatic AI tool capable of reconstructing the digital twin geometry and analyzing the calcification distribution on the aortic valve. The developed automatic segmentation package enables the modeling of patient-specific anatomies, which can then be used to simulate virtual interventional procedures, optimize emerging prosthetic devices, and predict clinical outcomes.