TAVI-PREP: A Deep Learning-Based Tool for Automated Measurements Extraction in TAVI Planning.
automatic preoperative planning
deep neural networks
transcatheter aortic valve implantation (TAVI)
Journal
Diagnostics (Basel, Switzerland)
ISSN: 2075-4418
Titre abrégé: Diagnostics (Basel)
Pays: Switzerland
ID NLM: 101658402
Informations de publication
Date de publication:
11 Oct 2023
11 Oct 2023
Historique:
received:
01
09
2023
revised:
29
09
2023
accepted:
02
10
2023
medline:
28
10
2023
pubmed:
28
10
2023
entrez:
28
10
2023
Statut:
epublish
Résumé
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. 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. 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. TAVI-PREP provides reliable and time-efficient measurements of the aortic valvular complex that could aid clinicians in the preprocedural planning of TAVI procedures.
Sections du résumé
BACKGROUND
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
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
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
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.
Identifiants
pubmed: 37892002
pii: diagnostics13203181
doi: 10.3390/diagnostics13203181
pmc: PMC10606167
pii:
doi:
Types de publication
Journal Article
Langues
eng
Références
Med Image Anal. 2021 Dec;74:102222
pubmed: 34543913
JAMA. 2016 Dec 13;316(22):2402-2410
pubmed: 27898976
N Engl J Med. 2016 Apr 28;374(17):1609-20
pubmed: 27040324
N Engl J Med. 2019 May 2;380(18):1695-1705
pubmed: 30883058
Eur Radiol. 2014 Aug;24(8):1878-88
pubmed: 24845112
Diagn Interv Imaging. 2018 May;99(5):279-289
pubmed: 29307461
J Clin Med. 2023 Jul 19;12(14):
pubmed: 37510889
Trends Cardiovasc Med. 2022 Jan;32(1):34-41
pubmed: 33242635
Curr Cardiol Rep. 2018 Feb 12;20(2):9
pubmed: 29435741
J Interv Cardiol. 2020 May 28;2020:9843275
pubmed: 32549802
Comput Med Imaging Graph. 2023 Oct;109:102287
pubmed: 37634975
N Engl J Med. 2019 May 2;380(18):1706-1715
pubmed: 30883053
Eur J Radiol. 2014 Sep;83(9):1672-8
pubmed: 25022977
JACC Cardiovasc Imaging. 2019 Jan;12(1):1-24
pubmed: 30621986
Int J Cardiovasc Imaging. 2016 Mar;32(3):501-11
pubmed: 26498339
Ann Thorac Surg. 2011 Mar;91(3):716-23
pubmed: 21352985
J Clin Med. 2022 May 05;11(9):
pubmed: 35566708
Cardiovasc Revasc Med. 2018 Mar;19(2):229-236
pubmed: 29102344
N Engl J Med. 2020 Feb 27;382(9):799-809
pubmed: 31995682
N Engl J Med. 2017 Apr 6;376(14):1321-1331
pubmed: 28304219
Int J Cardiovasc Imaging. 2018 Nov;34(11):1819-1829
pubmed: 29909590
JACC Cardiovasc Interv. 2019 Jul 22;12(14):1293-1303
pubmed: 31320024
Int J Comput Assist Radiol Surg. 2022 Mar;17(3):507-519
pubmed: 35066774
JAMA Cardiol. 2021 Sep 1;6(9):1088-1089
pubmed: 34287627
J Cardiovasc Comput Tomogr. 2022 Sep-Oct;16(5):404-411
pubmed: 35337770
Comput Biol Med. 2023 Sep;163:107147
pubmed: 37329622
EuroIntervention. 2022 Feb 04;17(14):e1126-e1196
pubmed: 34931612