A multi-model deep learning approach for the identification of coronary artery calcifications within 2D coronary angiography images DOI Creative Commons
Edoardo De Rose, Ciro Benito Raggio, Ahmad Rasheed

et al.

International Journal of Computer Assisted Radiology and Surgery, Journal Year: 2025, Volume and Issue: unknown

Published: May 8, 2025

Abstract Purpose Identifying and quantifying coronary artery calcification (CAC) is crucial for preoperative planning, as it helps to estimate both the complexity of 2D angiography (2DCA) procedure risk developing intraoperative complications. Despite relevance, actual practice relies upon visual inspection 2DCA image frames by clinicians. This prone inaccuracies due poor contrast small size CAC; moreover, dependent on physician’s experience. To address this issue, we developed a workflow assist clinicians in identifying CAC within using data from 44 acquisitions across 14 patients. Methods Our consists three stages. In first stage, classification backbone based ResNet-18 applied guide identification extracting relevant features frames. second U-Net decoder architecture, mirroring encoding structure ResNet-18, employed identify regions interest (ROI) CAC. Eventually, post-processing step refines results obtain final ROI. The was evaluated leave-out cross-validation. Results proposed method outperformed comparative methods achieving an F1-score 0.87 (0.77 $$-$$ - 0.94) (median ± quartiles), while intersection over minimum (IoM) 0.64 (0.46 0.86) quartiles). Conclusion attempt propose clinical decision support system 2DCA. holds potential improve accuracy efficiency quantification, with promising applications. As future work, concurrent use multiple auxiliary tasks could be explored further segmentation performance.

Language: Английский

A multi-model deep learning approach for the identification of coronary artery calcifications within 2D coronary angiography images DOI Creative Commons
Edoardo De Rose, Ciro Benito Raggio, Ahmad Rasheed

et al.

International Journal of Computer Assisted Radiology and Surgery, Journal Year: 2025, Volume and Issue: unknown

Published: May 8, 2025

Abstract Purpose Identifying and quantifying coronary artery calcification (CAC) is crucial for preoperative planning, as it helps to estimate both the complexity of 2D angiography (2DCA) procedure risk developing intraoperative complications. Despite relevance, actual practice relies upon visual inspection 2DCA image frames by clinicians. This prone inaccuracies due poor contrast small size CAC; moreover, dependent on physician’s experience. To address this issue, we developed a workflow assist clinicians in identifying CAC within using data from 44 acquisitions across 14 patients. Methods Our consists three stages. In first stage, classification backbone based ResNet-18 applied guide identification extracting relevant features frames. second U-Net decoder architecture, mirroring encoding structure ResNet-18, employed identify regions interest (ROI) CAC. Eventually, post-processing step refines results obtain final ROI. The was evaluated leave-out cross-validation. Results proposed method outperformed comparative methods achieving an F1-score 0.87 (0.77 $$-$$ - 0.94) (median ± quartiles), while intersection over minimum (IoM) 0.64 (0.46 0.86) quartiles). Conclusion attempt propose clinical decision support system 2DCA. holds potential improve accuracy efficiency quantification, with promising applications. As future work, concurrent use multiple auxiliary tasks could be explored further segmentation performance.

Language: Английский

Citations

0