1.
Motion-corrected coronary calcium scores by a convolutional neural network: a robotic simulating study.
Zhang, Y, van der Werf, NR, Jiang, B, van Hamersvelt, R, Greuter, MJW, Xie, X
European radiology. 2020;(2):1285-1294
Abstract
OBJECTIVE To classify motion-induced blurred images of calcified coronary plaques so as to correct coronary calcium scores on nontriggered chest CT, using a deep convolutional neural network (CNN) trained by images of motion artifacts. METHODS Three artificial coronary arteries containing nine calcified plaques of different densities (high, medium, and low) and sizes (large, medium, and small) were attached to a moving robotic arm. The artificial arteries moving at 0-90 mm/s were scanned to generate nine categories (each from one calcified plaque) of images with motion artifacts. An inception v3 CNN was fine-tuned and validated. Agatston scores of the predicted classification by CNN were considered as corrected scores. Variation of Agatston scores on moving plaque and by CNN correction was calculated using the scores at rest as reference. RESULTS The overall accuracy of CNN classification was 79.2 ± 6.1% for nine categories. The accuracy was 88.3 ± 4.9%, 75.9 ± 6.4%, and 73.5 ± 5.0% for the high-, medium-, and low-density plaques, respectively. Compared with the Agatston score at rest, the overall median score variation was 37.8% (1st and 3rd quartile, 10.5% and 68.8%) in moving plaques. CNN correction largely decreased the variation to 3.7% (1.9%, 9.1%) (p < 0.001, Mann-Whitney U test) and improved the sensitivity (percentage of non-zero scores among all the scores) from 65 to 85% for detection of coronary calcifications. CONCLUSIONS In this experimental study, CNN showed the ability to classify motion-induced blurred images and correct calcium scores derived from nontriggered chest CT. CNN correction largely reduces the overall Agatston score variation and increases the sensitivity to detect calcifications. KEY POINTS • A deep CNN architecture trained by CT images of motion artifacts showed the ability to correct coronary calcium scores from blurred images. • A correction algorithm based on deep CNN can be used for a tenfold reduction in Agatston score variations from 38 to 3.7% of moving coronary calcified plaques and to improve the sensitivity from 65 to 85% for the detection of calcifications. • This experimental study provides a method to improve its accuracy for coronary calcium scores that is a fundamental step towards a real clinical scenario.
2.
Epicardial Fat Volume Improves the Prediction of Obstructive Coronary Artery Disease Above Traditional Risk Factors and Coronary Calcium Score.
Zhou, J, Chen, Y, Zhang, Y, Wang, H, Tan, Y, Liu, Y, Huang, L, Zhang, H, Ma, Y, Cong, H
Circulation. Cardiovascular imaging. 2019;(1):e008002
Abstract
BACKGROUND Recent studies have demonstrated the tremendous potential of epicardial fat volume (EFV) to predict obstructive coronary artery disease. We aimed to develop a new model to estimate pretest probability of obstructive coronary artery disease using traditional risk factors with coronary calcium score and EFV and compare it with proposed models in Chinese patients who underwent coronary computed tomography angiography. METHODS The new models were derived from 5743 consecutive patients using multivariate logistic regression and validated in an internal cohort using invasive coronary angiography as the outcome and an external cohort with clinical outcome data. Hosmer-Lemeshow goodness-of-fit test, area under the receiver operating characteristic curve, integrated discrimination improvement and net reclassification improvement were calculated to validate and compare the performance of models. RESULTS EFV improved prediction above conventional risk factors and coronary calcium score (area under the receiver operating characteristic curve increased from 0.856 to 0.874, integrated discrimination improvement 0.0487, net reclassification improvement 0.1181, P<0.0001 for all). The final model included 5 predictors: age, sex, symptom, coronary calcium score, and EFV. Good internal validation and external validation of the new model were achieved, with positive net reclassification improvement and integrated discrimination improvement, excellent area under the receiver operating characteristic curve and favorable calibration. Further, the new model demonstrated a better prediction of clinical outcome, resulting in a more cost-effective risk stratification to optimize decision-making of downstream diagnosis and treatment. CONCLUSIONS Addition of EFV to conventional risk factors and coronary calcium score offered a more accurate and effective estimation for pretest probability of obstructive coronary artery disease, which may help to improve initial management of stable chest pain.
3.
Validation and comparison of four models to calculate pretest probability of obstructive coronary artery disease in a Chinese population: A coronary computed tomographic angiography study.
Zhou, J, Liu, Y, Huang, L, Tan, Y, Li, X, Zhang, H, Ma, Y, Zhang, Y
Journal of cardiovascular computed tomography. 2017;(4):317-323
Abstract
OBJECTIVE We sought to compare the performance of the updated Diamond-Forrester method (UDFM), Duke clinical score (DCS), Genders clinical model (GCM) and Genders extended model (GEM) in a Chinese population referred to coronary computed tomography angiography (coronary CTA). BACKGROUND The reliability of existing models to calculate the pretest proability (PTP) of obstructive coronary artery disease (CAD) have not been fully investigated, especially in a Chinese population. METHODS We identified 5743 consecutive patients with suspected stable CAD who underwent coronary calcium scoring (CCS) and coronary CCTA. Obstructive CAD was defined as with the presence of ≥50% diameter stenosis in coronary CTA or unassessable segments due to severe calcification. Area under the receiver operating characteristic curve (AUC), integrated discrimination improvement (IDI), net reclassification improvement (NRI) and Hosmer-Lemeshow goodness-of-fit statistic (H-L χ2) were assessed to validate and compare these models. RESULTS Overall, 1872 (32%) patients had obstructive CAD and 2467 (43%) had a CCS of 0. GEM demonstrated improved discrimination over the other models through the largest AUC (0.816 for GEM, 0.774 for GCM, 0.772 for DCS and 0.765 for UDFM). UDFM (-0.3255, p < 0.0001), DCS (-0.3149, p < 0.0001) and GCM (-0.2264, p < 0.0001) showed negative IDI compared to GEM. The NRI was significantly higher for GEM than the other models (0.7152, p < 0.0001, 0.5595, p < 0.0001 and 0.3195, p < 0.0001, respectively). All of the four models overestimated the prevalence of obstructive CAD, with unsatisfactory (p < 0.01 for all) calibration for UDFM (H-L χ2 = 137.82), DCS (H-L χ2 = 156.70), GCM (H-L χ2 = 51.17) and GEM (H-L χ2 = 29.67), respectively. CONCLUSION Although GEM was superior for calculating PTP in a Chinese population referred for coronary CTA, developing new models allowing for more accurate and operational estimation are warranted.