Measurement of right and left ventricular volume ratio may be a superior approach ( 13). In acute PE the relative diameter of the right ventricle to left ventricle is used to predict mortality ( 12, 13). Cardiac features such as right ventricular (RV) dilatation ( 11), RV hypertrophy and septal flattening ( 5) add to pulmonary arterial dilatation as predictors of the presence of PH. This feature may be the clue to the diagnosis of pulmonary hypertension ( 7– 10). Pulmonary arterial dilatation is a salient feature radiologists observe on routine thoracic imaging. Current imaging approaches in pulmonary vascular disease rely on visual assessments or manual measurements of cardiac, pulmonary arterial and aortic size such measures are used to risk stratify patients with acute PE ( 5– 8) and diagnose PH ( 5, 6). Computed tomography pulmonary angiography (CTPA) is a crucial imaging investigation performed in patients with suspected pulmonary embolism (PE) and in the work up of patients with suspected pulmonary hypertension (PH) ( 1). Pulmonary vascular disease encompasses a range of conditions that are linked with a large disease burden worldwide and are associated with high mortality and morbidity ( 1– 4). DL volumetric biomarkers can potentially improve CTPA cardiac assessment and invasive haemodynamic prediction. The failure rates in mixed pulmonary vascular disease were low (<3.9%) indicating good generalisability of the model to different diseases.Ĭonclusion: Fully automated segmentation of the four cardiac chambers and great vessels has been achieved in CTPA with high accuracy and low rates of failure. The model demonstrated good generalisability to different vendors and hospitals with similar performance in the external test cohort. The volume of segmented cardiac structures by deep learning had higher or equivalent correlation with invasive haemodynamics than by manual segmentations. Right ventricle myocardial volume had strong correlation with mean pulmonary artery pressure (Spearman's correlation coefficient = 0.7). Interobserver comparison found that the left and right ventricle myocardium segmentations showed the most variation between observers: mean DSC (range) of 0.795 (0.785–0.801) and 0.520 (0.482–0.542) respectively. The left and right ventricle myocardium segmentations had lower DSC of 0.83 and 0.58 respectively while all other structures had DSC >0.89 in the internal test cohort and >0.87 in the external test cohort. Results: Dice similarity coefficients (DSC) for segmented structures were in the range 0.58–0.93 for both the internal and external test cohorts. Volumetric imaging biomarkers were correlated with invasive haemodynamics in the test cohort. Segmentation was achieved using deep learning via a convolutional neural network. Failure analysis was conducted in 1,333 patients with mixed pulmonary vascular disease. Ground truth segmentations were performed by consultant cardiothoracic radiologists. Methods: A nine structure semantic segmentation model of the heart and great vessels was developed using 200 patients (80/20/100 training/validation/internal testing) with testing in 20 external patients. The primary aim of this study was to develop an automated whole heart segmentation (four chamber and great vessels) model for CTPA. Cardiac and great vessel assessments on CTPA are based on visual assessment and manual measurements which are known to have poor reproducibility. Introduction: Computed tomography pulmonary angiography (CTPA) is an essential test in the work-up of suspected pulmonary vascular disease including pulmonary hypertension and pulmonary embolism. 9Department of Computer Science, University of Sheffield, Sheffield, United Kingdom.8Sheffield Pulmonary Vascular Disease Unit, Sheffield Teaching Hospitals NHS Trust, Sheffield, United Kingdom.7Department of Radiology, Leiden University Medical Center, Leiden, Netherlands.6MRC London Institute of Medical Sciences, Imperial College London, London, United Kingdom.5Norwich Medical School, University of East Anglia, Norwich, United Kingdom.4Radiology Department, Sheffield Teaching Hospitals NHSFT, Sheffield, United Kingdom.3Insigneo Institute for in Silico Medicine, University of Sheffield, Sheffield, United Kingdom.23D Imaging Lab, Sheffield Teaching Hospitals NHSFT, Sheffield, United Kingdom.1Department of Infection, Immunity and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom.Kiely 1,3,8, Michail Mamalakis 1,3,9 and Andrew J. van der Geest 7, Robin Condliffe 1,8, David G. Johns 4, Smitha Rajaram 4, Pankaj Garg 5, Dheyaa Alkhanfar 1, Peter Metherall 2, Declan P. Taylor 2, Samer Alabed 1, Krit Dwivedi 1,3, Kavitasagary Karunasaagarar 1,4, Christopher S.
0 Comments
Leave a Reply. |