Back to the list
Congress: ECR25
Poster Number: C-22624
Type: Poster: EPOS Radiologist (scientific)
Authorblock: F. Pisu1, A. Colombo1, D. Vignale1, A. Palmisano1, A. Bartoli2, M. Liberotti1, V. Morrone1, L. Saba3, A. Esposito1; 1Milan/IT, 2Marseille/FR, 3Cagliari/IT
Disclosures:
Francesco Pisu: Nothing to disclose
Alberto Colombo: Nothing to disclose
Davide Vignale: Nothing to disclose
Anna Palmisano: Nothing to disclose
Axel Bartoli: Nothing to disclose
Marta Liberotti: Nothing to disclose
Vittorio Morrone: Nothing to disclose
Luca Saba: Nothing to disclose
Antonio Esposito: Nothing to disclose
Keywords: Cardiovascular system, Vascular, CT-Angiography, Computer Applications-Detection, diagnosis, Arteriosclerosis
Purpose Coronary Computed Tomography angiography (CCTA) represents the non-invasive method of choice for the assessment of coronary artery disease. Accurate multi-planar reconstructions (MPR) using centerlines are vital for detecting stenosis and extracting radiomic features. However, manual delineation centerline delineation is prohibitively time-consuming in clinical practice. Current automated tools present significant limitations, including extensive user interactions requirements, inaccurate centerline tracing (particularly at vessel bifurcations), and the inability to process multiple cases simultaneously.This study aimed to develop a fully automated deep-learning (DL) method...
Read more Methods and materials All study methods received approval from the ethics committee of IRCCS San Raffaele Scientific Institute (study registration number CE 158/INT/2022) and adhered to relevant guidelines.The proposed approach extracts topology-preserving centerlines from coronary lumen meshes extracted from DL-derived lumen segmentations (Fig. 1). [fig 1]  AI-enabled Coronary Lumen SegmentationA 2D encoder-decoder architecture was employed to segment the coronary lumen from individual CCTA slices (Fig. 2). The network utilized a pre-trained EfficientNet as the encoder’s backbone for hierarchical feature extraction, with skip connections to preserve...
Read more Results Coronary segmentation performanceThe method showed high accuracy in segmentation with IoU of 76%±3 and DSC of 0.87±0.02 (Fig. 3). [fig 3] Centerline extraction performanceQuantitative evaluation against reference centerlines yielded an OV of 92%±0.09 and AD of 0.48±0.06 mm, indicating high extraction accuracy. Representative segmentation and centerline results are shown in Fig. 4, with straightened MPR views comparing automated and reference centerlines along major coronary vessels shown in Fig. 5. [fig 4] [fig 5] Computational efficiencyThe pipeline extracts coronary tree centerlines automatically in under a minute....
Read more Conclusion ConclusionThe proposed pipeline enables rapid and accurate coronary centerline extraction, achieving high agreement with expert annotations while requiring minimal computational resources. The fully automated nature of our approach eliminates the need for manual interaction, making it ideal for large-scale studies. This automation accelerates stenosis and plaque characterization workflows, supporting timely and effective patient care in cardiology.Our approach has limitations, notably the reduced resolution from downsampling images due to computational constraints, the relatively small training dataset size and the dependency on...
Read more References [1] Wasserthal J et al., TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images. Radiol Artif Intell. 2023 Jul 5;5(5):e230024. doi: 10.1148/ryai.230024[2] R. Gharleghi, D. Adikari, K. Ellenberger, S. Y. Ooi, C. Ellis, C. M. Chen et al., “Automated segmentation of normal and diseased coronary arteries—The ASOCA challenge,” Comput. Med. Imaging Graphics 97, 102049 (2022)[3] W. Schroeder, R. Maynard and B. Geveci, "Flying edges: A high-performance scalable isocontouring algorithm," 2015 IEEE 5th Symposium on Large Data Analysis and Visualization (LDAV),...
Read more
GALLERY