Interobserver variability among expert readers quantifying plaque volume and plaque characteristics on coronary CT angiography: a CLARIFY trial sub-study

Intra- and inter-reader reproducibility in quantitative coronary plaque analysis on coronary computed tomography angiography

PURPOSE

Coronary artery plaque burden, low attenuation non-calcified plaque (LAP), and pericoronary adipose tissue (PCAT) on coronary CT angiography (CCTA), have been linked to future cardiac events. The purpose of this study was to evaluate intra- and inter reader reproducibility in the quantification of coronary plaque burden and its characteristics using an artificial intelligence-enhanced semi-automated software.

MATERIALS AND METHODS

A total of 10 women and 6 men, aged 52 (IQR 49-58) underwent CCTA using a Siemens Somatom Force, Somatom Definition AS and Somatom Definition Flash scanners. Two expert readers utilized dedicated semi-automatic software (vascuCAP, Elucid Bioimaging, Wenham, MA) to assess calcified plaque, low attenuation plaque and PCAT. Readers were blinded to all clinical information and repeated their analysis at 6 weeks in random order to minimize recall bias. Data analysis was performed on the right and left coronary arteries. Intra- and inter-reader reproducibility was compared using Pearson correlation coefficient, while absolute values between analyses and readers were compared with paired non-parametric tests. This is a sub-study of the Specialized Center of Research Excellence (SCORE) clinical trial (5U54AG062334).

RESULTS

A total of 64 vessels from 16 patients were analyzed. Intra-reader Pearson correlation coefficients for calcified plaque volume, LAP volume and PCAT volumes were 0.96, 0.99 and 0.92 for reader 1 and 0.94, 0.94 and 0.95 for reader 2, respectively, (all p < 0.0001). Inter-reader Pearson correlation coefficients for calcified plaque volume, LAP and PCAT volumes were 0.92, 0.96 and 0.78, and 0.99, 0.99 and 0.93 on the second analyses, all had a p value <0.0001. There was no significant bias on the corresponding Bland-Altman analyses.

CONCLUSION

Volume measurement of coronary plaque burden and PCAT volume can be performed with high intra- and inter-reader agreement.

Automatic Coronary Artery Plaque Quantification and CAD-RADS Prediction Using Mesh Priors

Coronary artery disease (CAD) remains the leading cause of death worldwide. Patients with suspected CAD undergo coronary CT angiography (CCTA) to evaluate the risk of cardiovascular events and determine the treatment. Clinical analysis of coronary arteries in CCTA comprises the identification of atherosclerotic plaque, as well as the grading of any coronary artery stenosis typically obtained through the CAD-Reporting and Data System (CAD-RADS). This requires analysis of the coronary lumen and plaque. While voxel-wise segmentation is a commonly used approach in various segmentation tasks, it does not guarantee topologically plausible shapes. To address this, in this work, we propose to directly infer surface meshes for coronary artery lumen and plaque based on a centerline prior and use it in the downstream task of CAD-RADS scoring. The method is developed and evaluated using a total of 2407 CCTA scans. Our method achieved lesion-wise volume intraclass correlation coefficients of 0.98, 0.79, and 0.85 for calcified, non-calcified, and total plaque volume respectively. Patient-level CAD-RADS categorization was evaluated on a representative hold-out test set of 300 scans, for which the achieved linearly weighted kappa ( κ ) was 0.75. CAD-RADS categorization on the set of 658 scans from another hospital and scanner led to a κ of 0.71. The results demonstrate that direct inference of coronary artery meshes for lumen and plaque is feasible, and allows for the automated prediction of routinely performed CAD-RADS categorization.

Roadmap on the use of artificial intelligence for imaging of vulnerable atherosclerotic plaque in coronary arteries

Artificial intelligence (AI) is likely to revolutionize the way medical images are analysed and has the potential to improve the identification and analysis of vulnerable or high-risk atherosclerotic plaques in coronary arteries, leading to advances in the treatment of coronary artery disease. However, coronary plaque analysis is challenging owing to cardiac and respiratory motion, as well as the small size of cardiovascular structures. Moreover, the analysis of coronary imaging data is time-consuming, can be performed only by clinicians with dedicated cardiovascular imaging training, and is subject to considerable interreader and intrareader variability. AI has the potential to improve the assessment of images of vulnerable plaque in coronary arteries, but requires robust development, testing and validation. Combining human expertise with AI might facilitate the reliable and valid interpretation of images obtained using CT, MRI, PET, intravascular ultrasonography and optical coherence tomography. In this Roadmap, we review existing evidence on the application of AI to the imaging of vulnerable plaque in coronary arteries and provide consensus recommendations developed by an interdisciplinary group of experts on AI and non-invasive and invasive coronary imaging. We also outline future requirements of AI technology to address bias, uncertainty, explainability and generalizability, which are all essential for the acceptance of AI and its clinical utility in handling the anticipated growing volume of coronary imaging procedures.

Determination of lipid-rich plaques by artificial intelligence-enabled quantitative computed tomography using near-infrared spectroscopy as reference

BACKGROUND AND AIMS

Artificial intelligence quantitative CT (AI-QCT) determines coronary plaque morphology with high efficiency and accuracy. Yet, its performance to quantify lipid-rich plaque remains unclear. This study investigated the performance of AI-QCT for the detection of low-density noncalcified plaque (LD-NCP) using near-infrared spectroscopy-intravascular ultrasound (NIRS-IVUS).

METHODS

The INVICTUS Registry is a multi-center registry enrolling patients undergoing clinically indicated coronary CT angiography and IVUS, NIRS-IVUS, or optical coherence tomography. We assessed the performance of various Hounsfield unit (HU) and volume thresholds of LD-NCP using maxLCBI4mm ≥ 400 as the reference standard and the correlation of the vessel area, lumen area, plaque burden, and lesion length between AI-QCT and IVUS.

RESULTS

This study included 133 atherosclerotic plaques from 47 patients who underwent coronary CT angiography and NIRS-IVUS The area under the curve of LD-NCP<30HU was 0.97 (95% confidence interval [CI]: 0.93-1.00] with an optimal volume threshold of 2.30 mm3. Accuracy, sensitivity, and specificity were 94% (95% CI: 88-96%], 93% (95% CI: 76-98%), and 94% (95% CI: 88-98%), respectively, using <30 HU and 2.3 mm3, versus 42%, 100%, and 27% using <30 HU and >0 mm3 volume of LD-NCP (p < 0.001 for accuracy and specificity). AI-QCT strongly correlated with IVUS measurements; vessel area (r2 = 0.87), lumen area (r2 = 0.87), plaque burden (r2 = 0.78) and lesion length (r2 = 0.88), respectively.

CONCLUSIONS

AI-QCT demonstrated excellent diagnostic performance in detecting significant LD-NCP using maxLCBI4mm ≥ 400 as the reference standard. Additionally, vessel area, lumen area, plaque burden, and lesion length derived from AI-QCT strongly correlated with respective IVUS measurements.

Machine Learning in Invasive and Noninvasive Coronary Angiography

PURPOSE OF REVIEW

The objective of this review is to shed light on the transformative potential of machine learning (ML) in coronary angiography. We aim to understand existing developments in using ML for coronary angiography and discuss broader implications for the future of coronary angiography and cardiovascular medicine.

RECENT FINDINGS

The developments in invasive and noninvasive imaging have revolutionized diagnosis and treatment of coronary artery disease (CAD). However, CAD remains underdiagnosed and undertreated. ML has emerged as a powerful tool to further improve image analysis, hemodynamic assessment, lesion detection, and predictive modeling. These advancements have enabled more accurate identification of CAD, streamlined workflows, reduced the need for invasive diagnostic procedures, and improved the diagnostic value of invasive procedures when they are needed. Further integration of ML with coronary angiography will advance the prevention, diagnosis, and treatment of CAD. The integration of ML with coronary angiography is ushering in a new era in cardiovascular medicine. We highlight five use cases to leverage ML in coronary angiography: (1) improvement of quality and efficacy, (2) characterization of plaque, (3) hemodynamic assessment, (4) prediction of future outcomes, and (5) diagnosis of non-atherosclerotic coronary disease.

How early can atherosclerosis be detected by coronary CT angiography? Insights from quantitative CT analysis of serial scans in the PARADIGM trial

BACKGROUND

Non-obstructing small coronary plaques may not be well recognized by expert readers during coronary computed tomography angiography (CCTA) evaluation. Recent developments in atherosclerosis imaging quantitative computed tomography (AI-QCT) enabled by machine learning allow for whole-heart coronary phenotyping of atherosclerosis, but its diagnostic role for detection of small plaques on CCTA is unknown.

METHODS

We performed AI-QCT in patients who underwent serial CCTA in the multinational PARADIGM study. AI-QCT results were verified by a level III experienced reader, who was blinded to baseline and follow-up status of CCTA. This retrospective analysis aimed to characterize small plaques on baseline CCTA and evaluate their serial changes on follow-up imaging. Small plaques were defined as a total plaque volume <50 ​mm3.

RESULTS

A total of 99 patients with 502 small plaques were included. The median total plaque volume was 6.8 ​mm3 (IQR 3.5-13.9 ​mm3), most of which was non-calcified (median 6.2 ​mm3; 2.9-12.3 ​mm3). The median age at the time of baseline CCTA was 61 years old and 63% were male. The mean interscan period was 3.8 ​± ​1.6 years. On follow-up CCTA, 437 (87%) plaques were present at the same location as small plaques on baseline CCTA; 72% were larger and 15% decreased in volume. The median total plaque volume and non-calcified plaque volume increased to 18.9 ​mm3 (IQR 8.3-45.2 ​mm3) and 13.8 ​mm3 (IQR 5.7-33.4 ​mm3), respectively, among plaques that persisted on follow-up CCTA. Small plaques no longer visualized on follow-up CCTA were significantly more likely to be of lower volume, shorter in length, non-calcified, and more distal in the coronary artery, as compared with plaques that persisted at follow-up.

CONCLUSION

In this retrospective analysis from the PARADIGM study, small plaques (<50 ​mm3) identified by AI-QCT persisted at the same location and were often larger on follow-up CCTA.

Deep Learning-based Post Hoc CT Denoising for the Coronary Perivascular Fat Attenuation Index

RATIONALE AND OBJECTIVES

Coronary inflammation related to high-risk hemorrhagic plaques can be captured by the perivascular fat attenuation index (FAI) using coronary computed tomography angiography (CCTA). Since the FAI is susceptible to image noise, we believe deep learning (DL)-based post hoc noise reduction can improve diagnostic capability. We aimed to assess the diagnostic performance of the FAI in DL-based denoised high-fidelity CCTA images compared with coronary plaque magnetic resonance imaging (MRI) delivered high-intensity hemorrhagic plaques (HIPs).

MATERIALS AND METHODS

We retrospectively reviewed 43 patients who underwent CCTA and coronary plaque MRI. We generated high-fidelity CCTA images by denoising the standard CCTA images using a residual dense network that supervised the denoising task by averaging three cardiac phases with nonrigid registration. We measured the FAIs as the mean CT value of all voxels (range of -190 to -30 HU) located within a radial distance from the outer proximal right coronary artery wall. The diagnostic reference standard was defined as HIPs (high-risk hemorrhagic plaques) using MRI. The diagnostic performance of the FAI in the original and denoised images was assessed using receiver operating characteristic curves.

RESULTS

Of 43 patients, 13 had HIPs. The denoised CCTA improved the area under the curve (0.89 [95% confidence interval (CI) 0.78-0.99]) of the FAI compared with that in the original image (0.77 [95% CI, 0.62-0.91], p = 0.008). The optimal cutoff value for predicting HIPs in denoised CCTA was -69 HU with 0.85 (11/13) sensitivity, 0.79 (25/30) specificity, and 0.80 (36/43) accuracy.

CONCLUSION

DL-based denoised high-fidelity CCTA improved the AUC and specificity of the FAI for predicting HIPs.

Artificial Intelligence in Cardiovascular CT and MR Imaging

The technological development of Artificial Intelligence (AI) has grown rapidly in recent years. The applications of AI to cardiovascular imaging are various and could improve the radiologists' workflow, speeding up acquisition and post-processing time, increasing image quality and diagnostic accuracy. Several studies have already proved AI applications in Coronary Computed Tomography Angiography and Cardiac Magnetic Resonance, including automatic evaluation of calcium score, quantification of coronary stenosis and plaque analysis, or the automatic quantification of heart volumes and myocardial tissue characterization. The aim of this review is to summarize the latest advances in the field of AI applied to cardiovascular CT and MR imaging.