Automated Detection, Segmentation, and Classification of Pericardial Effusions on Chest CT Using a Deep Convolutional Neural Network.
AI (Artificial Intelligence)
DCNN (deep convolutional neural network)
X-ray
computed tomography
hemopericardium
pericardial effusion
Journal
Diagnostics (Basel, Switzerland)
ISSN: 2075-4418
Titre abrégé: Diagnostics (Basel)
Pays: Switzerland
ID NLM: 101658402
Informations de publication
Date de publication:
21 Apr 2022
21 Apr 2022
Historique:
received:
18
03
2022
revised:
09
04
2022
accepted:
19
04
2022
entrez:
28
5
2022
pubmed:
29
5
2022
medline:
29
5
2022
Statut:
epublish
Résumé
Pericardial effusions (PEFs) are often missed on Computed Tomography (CT), which particularly affects the outcome of patients presenting with hemodynamic compromise. An automatic PEF detection, segmentation, and classification tool would expedite and improve CT based PEF diagnosis; 258 CTs with (206 with simple PEF, 52 with hemopericardium) and without PEF (each 134 with contrast, 124 non-enhanced) were identified using the radiology report (01/2016−01/2021). PEF were manually 3D-segmented. A deep convolutional neural network (nnU-Net) was trained on 316 cases and separately tested on the remaining 200 and 22 external post-mortem CTs. Inter-reader variability was tested on 40 CTs. PEF classification utilized the median Hounsfield unit from each prediction. The sensitivity and specificity for PEF detection was 97% (95% CI 91.48−99.38%) and 100.00% (95% CI 96.38−100.00%) and 89.74% and 83.61% for diagnosing hemopericardium (AUC 0.944, 95% CI 0.904−0.984). Model performance (Dice coefficient: 0.75 ± 0.01) was non-inferior to inter-reader (0.69 ± 0.02) and was unaffected by contrast administration nor alternative chest pathology (p > 0.05). External dataset testing yielded similar results. Our model reliably detects, segments, and classifies PEF on CT in a complex dataset, potentially serving as an alert tool whilst enhancing report quality. The model and corresponding datasets are publicly available.
Identifiants
pubmed: 35626201
pii: diagnostics12051045
doi: 10.3390/diagnostics12051045
pmc: PMC9139725
pii:
doi:
Types de publication
Journal Article
Langues
eng
Références
Acta Radiol. 2013 May;54(4):396-400
pubmed: 23436832
J Forensic Leg Med. 2012 Apr;19(3):126-31
pubmed: 22390997
Contrast Media Mol Imaging. 2019 Jul 1;2019:1545747
pubmed: 31354393
Nat Med. 2020 Sep;26(9):1351-1363
pubmed: 32908284
IEEE Trans Biomed Eng. 2013 Jul;60(7):1834-40
pubmed: 23372069
Anatol J Cardiol. 2017 Apr;17(4):322-327
pubmed: 28100899
Radiol Clin North Am. 2020 Jan;58(1):133-150
pubmed: 31731897
Eur Radiol Exp. 2020 Aug 20;4(1):50
pubmed: 32814998
J Card Surg. 2019 Dec;34(12):1486-1491
pubmed: 31573114
Prog Cardiovasc Dis. 2017 Jan - Feb;59(4):380-388
pubmed: 28062268
Radiology. 2020 Apr;295(1):66-79
pubmed: 32043947
Forensic Sci Med Pathol. 2017 Dec;13(4):426-431
pubmed: 28819715
Radiology. 2020 Aug;296(2):E65-E71
pubmed: 32191588
Eur Heart J. 2015 Nov 7;36(42):2921-2964
pubmed: 26320112
J Am Coll Radiol. 2018 Aug;15(8):1087-1096
pubmed: 29941240
JAMA. 2019 Sep 3;322(9):843-856
pubmed: 31479136
Nat Methods. 2021 Feb;18(2):203-211
pubmed: 33288961
Cardiol Clin. 2017 Nov;35(4):515-524
pubmed: 29025543
Radiol Artif Intell. 2020 Mar 25;2(2):e200029
pubmed: 33937821
BMC Bioinformatics. 2011 Mar 17;12:77
pubmed: 21414208
Heart Fail Rev. 2013 May;18(3):289-306
pubmed: 22446985
Herz. 2012 Mar;37(2):183-7
pubmed: 21301790
J Am Soc Echocardiogr. 2013 Sep;26(9):965-1012.e15
pubmed: 23998693
Forensic Sci Med Pathol. 2014 Dec;10(4):583-606
pubmed: 24723662
Eur J Radiol. 2013 Jan;82(1):100-5
pubmed: 22749769