Diagnostic accuracy and performance of artificial intelligence in measuring left atrial volumes and function on multiphasic CT in patients with atrial fibrillation

Aquino, Gilberto; Chamberlin, Jordan; Yacoub, Basel; Kocher, Madison; Kabakus, Ismail; Akkaya, Selçuk; Mercer, Megan; Waltz, Jeffrey; Fiegel, Matthew; Leaphart, Nathan; Jacob, Athira; Gülsün, Mehmet A.; Gilkes, James; Stephenson, Joe; Sharma, Puneet; Sahbaee, Pooyan; Schoepf, Joseph; Zimmerman, Stefan L.; Field, Michael E.; Agha, Ali; Burt, Jeremy R.

doi:10.1007/s00330-022-08657-y

Cited by 10 publications

(7 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…However, our systematic review of DL-based cardiac CT segmentations identified only 18 studies. Due to the availability and speed of CT compared to other modalities, and its increased mention in cardiac disease management guidelines, automatic and accurate CT segmentation models for cardiac structures are likely to increasingly play a role in routine clinical use ( 2 , 19 ). This systematic review identified 18 original research studies published since 2013 that presented DL tools for automated segmentation of cardiac structures on CT.…”

Section: Discussionmentioning

confidence: 99%

“…Defining the inner and outer borders of the LV myocardium has been achieved using five different segmentation models. A combination of image-to-image (I2I) segmentation network with conditional variational autoencoder (cVAE) had the highest DSC score ( 19 ), followed by a 3D segmentation U-Net model using deep attention network (Attention Gates) to selectively emphasise the myocardium boundary structures ( 1 ). In contrast, only one study focused on segmenting the more complex and thinner RV myocardium.…”

Section: Discussionmentioning

confidence: 99%

“…Different limitations have been mentioned in the previous studies including that DL models can mitigate small mistakes in segmentation but cannot avoid observers’ mistakes in the model training step ( 1 , 2 ). Most of the studies developed their models and tested them in single-centre or single-vendor ( 3 , 6 , 8 , 19 ). In addition, it was reported that without using contrast media, contouring the borders between the structures can be difficult ( 11 ).…”

Section: Discussionmentioning

confidence: 99%

“…Elevated heart rate or atrial fibrillation patients were reported to negatively affect the images’ resolution, producing lower DSC scores ( 10 ). DL models using hinge points to find the structures’ borders may negatively affect the estimation, resulting in overestimation in segmentation ( 19 ).…”

Section: Discussionmentioning

confidence: 99%

See 3 more Smart Citations

Advancements in cardiac structures segmentation: a comprehensive systematic review of deep learning in CT imaging

Alnasser,

Abdulaal,

Maiter

et al. 2024

Front. Cardiovasc. Med.

View full text Add to dashboard Cite

BackgroundSegmentation of cardiac structures is an important step in evaluation of the heart on imaging. There has been growing interest in how artificial intelligence (AI) methods—particularly deep learning (DL)—can be used to automate this process. Existing AI approaches to cardiac segmentation have mostly focused on cardiac MRI. This systematic review aimed to appraise the performance and quality of supervised DL tools for the segmentation of cardiac structures on CT.MethodsEmbase and Medline databases were searched to identify related studies from January 1, 2013 to December 4, 2023. Original research studies published in peer-reviewed journals after January 1, 2013 were eligible for inclusion if they presented supervised DL-based tools for the segmentation of cardiac structures and non-coronary great vessels on CT. The data extracted from eligible studies included information about cardiac structure(s) being segmented, study location, DL architectures and reported performance metrics such as the Dice similarity coefficient (DSC). The quality of the included studies was assessed using the Checklist for Artificial Intelligence in Medical Imaging (CLAIM).Results18 studies published after 2020 were included. The DSC scores median achieved for the most commonly segmented structures were left atrium (0.88, IQR 0.83–0.91), left ventricle (0.91, IQR 0.89–0.94), left ventricle myocardium (0.83, IQR 0.82–0.92), right atrium (0.88, IQR 0.83–0.90), right ventricle (0.91, IQR 0.85–0.92), and pulmonary artery (0.92, IQR 0.87–0.93). Compliance of studies with CLAIM was variable. In particular, only 58% of studies showed compliance with dataset description criteria and most of the studies did not test or validate their models on external data (81%).ConclusionSupervised DL has been applied to the segmentation of various cardiac structures on CT. Most showed similar performance as measured by DSC values. Existing studies have been limited by the size and nature of the training datasets, inconsistent descriptions of ground truth annotations and lack of testing in external data or clinical settings.Systematic Review Registration[www.crd.york.ac.uk/prospero/], PROSPERO [CRD42023431113].

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

See 2 more Smart Citations

Advancements in cardiac structures segmentation: a comprehensive systematic review of deep learning in CT imaging

Alnasser,

Abdulaal,

Maiter

et al. 2024

Front. Cardiovasc. Med.

View full text Add to dashboard Cite

show abstract

“…Our study corroborates findings from the Heinz Nixdorf Recall Study and others, and further brings to light the value of non-coronary findings in coronary calcium scans for a comprehensive CVD risk assessment beyond coronary heart disease 12,13,14,15 . Although manual and automated LA volumetry in chest CT scans are relatively novel 37 ,38 , the pathophysiology of enlarged LA and its relationship with AF is well understood 39,40 .…”

Section: Discussionmentioning

confidence: 99%

AI-enabled Left Atrial Volumetry in Cardiac CT Scans Improves CHARGE-AF and Outperforms NT-ProBNP for Prediction of Atrial Fibrillation in Asymptomatic Individuals: Multi-Ethnic Study of Atherosclerosis

Naghavi,

Yankelevitz,

Reeves

et al. 2024

Preprint

View full text Add to dashboard Cite

Background: Coronary artery calcium (CAC) scans contain actionable information beyond CAC scores that is not currently reported. Methods: We have applied artificial intelligence-enabled automated cardiac chambers volumetry to CAC scans (AI-CAC) of 5535 asymptomatic individuals (52.2% women, ages 45-84) that were previously obtained for CAC scoring in the baseline examination (2000-2002) of the Multi-Ethnic Study of Atherosclerosis (MESA). We then used the 5-year outcomes data and compared the C-statistic of the AI-CAC LA volume with known predictors of atrial fibrillation (AF), the CHARGE-AF Risk Score and NT-proBNP (BNP) for prediction of incident AF. The AI-CAC automated cardiac chambers volumetry took on average 21 seconds per CAC scan. Results: At 1,2,3,4, and 5 years follow up 36, 77, 123, 182, and 236 cases of AF were identified respectively. The C-statistic AUC for AI-CAC LA volume was significantly higher than CHARGE-AF or BNP at year 1 (0.836, 0.742, 0.742), year 2 (0.842, 0.807,0.772), and year 3 (0.811, 0.785, 0.745) (p<0.02). For year 4 (0.785, 0.769, 0.725) and year 5 (0.781, 0.767, 0.734) respectively (p>0.05). AI-CAC LA volume significantly improved the continuous Net Reclassification Index for prediction of AF at year 1-5 when added to CAC score (0.74, 0.49, 0.53, 0.39, 0.44), CHARGE-AF Risk Score (0.60, 0.28, 0.32, 0.19, 0.24), and BNP (0.68, 0.44, 0.42, 0.30, 0.37) respectively (p<0.01). Conclusion: AI-CAC LA volume enabled prediction of AF as early as one year and significantly improved on risk classification of CHARGE-AF Risk Score and BNP.

show abstract

Current and Future Applications of Artificial Intelligence in Cardiac CT

Joshi

Melo²,

Ouyang

et al. 2023

Curr Cardiol Rep

View full text Add to dashboard Cite

Diagnostic accuracy and performance of artificial intelligence in measuring left atrial volumes and function on multiphasic CT in patients with atrial fibrillation

Cited by 10 publications

References 28 publications

Advancements in cardiac structures segmentation: a comprehensive systematic review of deep learning in CT imaging

Advancements in cardiac structures segmentation: a comprehensive systematic review of deep learning in CT imaging

AI-enabled Left Atrial Volumetry in Cardiac CT Scans Improves CHARGE-AF and Outperforms NT-ProBNP for Prediction of Atrial Fibrillation in Asymptomatic Individuals: Multi-Ethnic Study of Atherosclerosis

Current and Future Applications of Artificial Intelligence in Cardiac CT

Contact Info

Product

Resources

About