Radiology: Cardiothoracic Imaging Highlights 2023.

Radiology: Cardiothoracic Imaging publishes novel research and technical developments in cardiac, thoracic, and vascular imaging. The journal published many innovative studies during 2023 and achieved an impact factor for the first time since its inaugural issue in 2019, with an impact factor of 7.0. The current review article, led by the Radiology: Cardiothoracic Imaging trainee editorial board, highlights the most impactful articles published in the journal between November 2022 and October 2023. The review encompasses various aspects of coronary CT, photon-counting detector CT, PET/MRI, cardiac MRI, congenital heart disease, vascular imaging, thoracic imaging, artificial intelligence, and health services research. Key highlights include the potential for photon-counting detector CT to reduce contrast media volumes, utility of combined PET/MRI in the evaluation of cardiac sarcoidosis, the prognostic value of left atrial late gadolinium enhancement at MRI in predicting incident atrial fibrillation, the utility of an artificial intelligence tool to optimize detection of incidental pulmonary embolism, and standardization of medical terminology for cardiac CT. Ongoing research and future directions include evaluation of novel PET tracers for assessment of myocardial fibrosis, deployment of AI tools in clinical cardiovascular imaging workflows, and growing awareness of the need to improve environmental sustainability in imaging. Keywords: Coronary CT, Photon-counting Detector CT, PET/MRI, Cardiac MRI, Congenital Heart Disease, Vascular Imaging, Thoracic Imaging, Artificial Intelligence, Health Services Research © RSNA, 2024.

Deep Learning-Based Automated Labeling of Coronary Segments for Structured Reporting of Coronary Computed Tomography Angiography in Accordance With Society of Cardiovascular Computed Tomography Guidelines.

PURPOSE: To evaluate a novel deep learning (DL)-based automated coronary labeling approach for structured reporting of coronary artery disease according to the guidelines of the Society of Cardiovascular Computed Tomography (CT) on coronary CT angiography (CCTA). PATIENTS AND METHODS: A retrospective cohort of 104 patients (60.3 ± 10.7 y, 61% males) who had undergone prospectively electrocardiogram-synchronized CCTA were included. Coronary centerlines were automatically extracted, labeled, and validated by 2 expert readers according to Society of Cardiovascular CT guidelines. The DL algorithm was trained on 706 radiologist-annotated cases for the task of automatically labeling coronary artery centerlines. The architecture leverages tree-structured long short-term memory recurrent neural networks to capture the full topological information of the coronary trees by using a two-step approach: a bottom-up encoding step, followed by a top-down decoding step. The first module encodes each sub-tree into fixed-sized vector representations. The decoding module then selectively attends to the aggregated global context to perform the local assignation of labels. To assess the performance of the software, percentage overlap was calculated between the labels of the algorithm and the expert readers. RESULTS: A total number of 1491 segments were identified. The artificial intelligence-based software approach yielded an average overlap of 94.4% compared with the expert readers' labels ranging from 87.1% for the posterior descending artery of the right coronary artery to 100% for the proximal segment of the right coronary artery. The average computational time was 0.5 seconds per case. The interreader overlap was 96.6%. CONCLUSIONS: The presented fully automated DL-based coronary artery labeling algorithm provides fast and precise labeling of the coronary artery segments bearing the potential to improve automated structured reporting for CCTA.

Intra-individual comparison of image quality of the coronary arteries between photon-counting detector and energy-integrating detector CT systems.

PURPOSE: To intra-individually compare the objective and subjective image quality of coronary computed tomography angiography (CCTA) between photon-counting detector CT (PCD-CT) and energy-integrating detector CT (EID-CT). METHOD: Consecutive patients undergoing clinically indicated CCTA on an EID-CT system were prospectively enrolled for a research CCTA performed on a PCD-CT system within 30 days. Polychromatic images were reconstructed for both EID- and PCD-CT, while virtual monoenergetic images (VMI) were generated at 40, 45, 50, 55, 60 and 70 keV for PCD-CT. Two blinded readers calculated contrast-to-noise ratio (CNR) for each major coronary artery and rated image noise, vessel attenuation, vessel sharpness, and overall quality on a 1-5 Likert scale. Patients were then stratified by body mass index (BMI) [high (>30 kg/m2) vs low (<30 kg/m2)] for subgroup analysis. RESULTS: A total of 20 patients (67.5 ± 9.0 years, 75% male) were included in the study. Compared with EID-CT, coronary artery CNR values from PCD-CT monoenergetic and polychromatic reconstructions were all significantly higher than CNR values from EID-CT, with incrementally greater differences in obese subjects (all p < 0.008). Subjective image noise and sharpness were also significantly higher for all VMI reconstructions compared to EID-CT (all p < 0.008). All subjective scores were significantly higher for 55, 60, and 70 keV PCD-CT than EID-CT values (all p < 0.05). CONCLUSIONS: The improved objective and subjective image quality of PCD-CT compared to EID-CT may provide better visualization of the coronary arteries for a wide array of patients, especially those with a high BMI.

Radiology: Cardiothoracic Imaging Highlights 2022.

Since its inaugural issue in 2019, Radiology: Cardiothoracic Imaging has disseminated the latest scientific advances and technical developments in cardiac, vascular, and thoracic imaging. In this review, we highlight select articles published in this journal between October 2021 and October 2022. The scope of the review encompasses various aspects of coronary artery and congenital heart diseases, vascular diseases, thoracic imaging, and health services research. Key highlights include changes in the revised Coronary Artery Disease Reporting and Data System 2.0, the value of coronary CT angiography in informing prognosis and guiding treatment decisions, cardiac MRI findings after COVID-19 vaccination or infection, high-risk features at CT angiography to identify patients with aortic dissection at risk for late adverse events, and CT-guided fiducial marker placement for preoperative planning for pulmonary nodules. Ongoing research and future directions include photon-counting CT and artificial intelligence applications in cardiovascular imaging. Keywords: Pediatrics, CT Angiography, CT-Perfusion, CT-Spectral Imaging, MR Angiography, PET/CT, Transcatheter Aortic Valve Implantation/Replacement (TAVI/TAVR), Cardiac, Pulmonary, Vascular, Aorta, Coronary Arteries © RSNA, 2023.

Structural basis of antibacterial photodynamic action of curcumin against S. aureus.

Antimicrobial photodynamic therapy (aPDT) is an alternative tool to commercial antibiotics for the inactivation of pathogenic bacteria (e.g., S. aureus). However, there is still a lack of understanding of the molecular modeling of the photosensitizers and their mechanism of action through oxidative pathways. Herein, a combined experimental and computational evaluation of curcumin as a photosensitizer against S. aureus was performed. The radical forms of keto-enol tautomers and the energies of curcumin's frontier molecular orbitals were evaluated by density functional theory (DFT) to point out the photodynamic action as well as the photobleaching process. Furthermore, the electronic transitions of curcumin keto-enol tautomers were undertaken to predict the transitions as a photosensitizer during the antibacterial photodynamic process. Moreover, molecular docking was used to evaluate the binding affinity with the S. aureus tyrosyl-tRNA synthetase as the proposed a target for curcumin. In this regard, the molecular orbital energies show that the curcumin enol form has a character of 4.5% more basic than the keto form - the enol form is a more promising electron donor than its tautomer. Curcumin is a strong electrophile, with the enol form being 4.6% more electrophilic than its keto form. In addition, the regions susceptible to nucleophilic attack and photobleaching were evaluated by the Fukui function. Regarding the docking analysis, the model suggested that four hydrogen bonds contribute to the binding energy of curcumin's interaction with the ligand binding site of S. aureus tyrosyl-tRNA synthetase. Finally, residues Tyr36, Asp40, and Asp177 contact curcumin and may contribute to orienting the curcumin in the active area. Moreover, curcumin presented a photoinactivation of 4.5 log unit corroborating the necessity of the combined action of curcumin, light, and O2 to promote the photooxidation damage of S. aureus. These computational and experimental data suggest insights regarding the mechanism of action of curcumin as a photosensitizer to inactivate S. aureus bacteria.

Quantitative myocardial T2 mapping adds value to Japanese circulation society diagnostic criteria for active cardiac sarcoidosis.

Noninvasive identification of active myocardial inflammation in patients with cardiac sarcoidosis plays a key role in management but remains elusive. T2 mapping is a proposed solution, but the added value of quantitative myocardial T2 mapping for active cardiac sarcoidosis is unknown. Retrospective cohort analysis of 56 sequential patients with biopsy-confirmed extracardiac sarcoidosis who underwent cardiac MRI for myocardial T2 mapping. The presence or absence of active myocardial inflammation in patients with CS was defined using a modified Japanese circulation society criteria within one month of MRI. Myocardial T2 values were obtained for the 16 standard American Heart Association left ventricular segments. The best model was selected using logistic regression. Receiver operating characteristic curves and dominance analysis were used to evaluate the diagnostic performance and variable importance. Of the 56 sarcoidosis patients included, 14 met criteria for active myocardial inflammation. Mean basal T2 value was the best performing model for the diagnosis of active myocardial inflammation in CS patients (pR2 = 0.493, AUC = 0.918, 95% CI 0.835-1). Mean basal T2 value > 50.8 ms was the most accurate threshold (accuracy = 0.911). Mean basal T2 value + JCS criteria was significantly more accurate than JCS criteria alone (AUC = 0.981 vs. 0.887, p = 0.017). Quantitative regional T2 values are independent predictors of active myocardial inflammation in CS and may add additional discriminatory capability to JCS criteria for active disease.

Myocardial Characterization with Extracellular Volume Mapping with a First-Generation Photon-counting Detector CT with MRI Reference.

Background Photon-counting detector (PCD) CT provides comprehensive spectral data with every acquisition, but studies evaluating myocardial extracellular volume (ECV) quantification with use of PCD CT compared with an MRI reference remain lacking. Purpose To compare ECV quantification for myocardial tissue characterization between a first-generation PCD CT system and cardiac MRI. Materials and Methods In this single-center prospective study, adults without contraindication to iodine-based contrast media underwent same-day cardiac PCD CT and MRI with native and postcontrast T1 mapping and late gadolinium enhancement for various clinical indications for cardiac MRI (the reference standard) between July 2021 and January 2022. Global and midventricular ECV were assessed with use of three methods: single-energy PCD CT, dual-energy PCD CT, and MRI T1 mapping. Quantitative comparisons among all techniques were performed. Correlation and reliability between different methods of ECV quantification were assessed with use of the Pearson correlation coefficient (r) and the intraclass correlation coefficient. Results The final sample included 29 study participants (mean age ± SD, 54 years ± 17; 15 men). There was a strong correlation of ECV between dual- and single-energy PCD CT (r = 0.91, P < .001). Radiation dose was 40% lower with dual-energy versus single-energy PCD CT (volume CT dose index, 10.1 mGy vs 16.8 mGy, respectively; P < .001). In comparison with MRI, dual-energy PCD CT showed strong correlation (r = 0.82 and 0.91, both P < .001) and good to excellent reliability (intraclass correlation coefficients, 0.81 and 0.90) for midventricular and global ECV quantification, but it overestimated ECV by approximately 2%. Single-energy PCD CT showed similar relationship with MRI but underestimated ECV by 3%. Conclusion Myocardial tissue characterization with photon-counting detector CT-based quantitative extracellular volume analysis showed a strong correlation to MRI. © RSNA, 2023 Supplemental material is available for this article.

Highly Accelerated Compressed-Sensing 4D Flow for Intracardiac Flow Assessment.

BACKGROUND: Four-dimensional (4D) flow MRI allows for the quantification of complex flow patterns; however, its clinical use is limited by its inherently long acquisition time. Compressed sensing (CS) is an acceleration technique that provides substantial reduction in acquisition time. PURPOSE: To compare intracardiac flow measurements between conventional and CS-based highly accelerated 4D flow acquisitions. STUDY TYPE: Prospective. SUBJECTS: Fifty healthy volunteers (28.0 ± 7.1 years, 24 males). FIELD STRENGTH/SEQUENCE: Whole heart time-resolved 3D gradient echo with three-directional velocity encoding (4D flow) with conventional parallel imaging (factor 3) as well as CS (factor 7.7) acceleration at 3 T. ASSESSMENT: 4D flow MRI data were postprocessed by applying a valve tracking algorithm. Acquisition times, flow volumes (mL/cycle) and diastolic function parameters (ratio of early to late diastolic left ventricular peak velocities [E/A] and ratio of early mitral inflow velocity to mitral annular early diastolic velocity [E/e']) were quantified by two readers. STATISTICAL TESTS: Paired-samples t-test and Wilcoxon rank sum test to compare measurements. Pearson correlation coefficient (r), Bland-Altman-analysis (BA) and intraclass correlation coefficient (ICC) to evaluate agreement between techniques and readers. A P value < 0.05 was considered statistically significant. RESULTS: A significant improvement in acquisition time was observed using CS vs. conventional accelerated acquisition (6.7 ± 1.3 vs. 12.0 ± 1.3 min). Net forward flow measurements for all valves showed good correlation (r > 0.81) and agreement (ICCs > 0.89) between conventional and CS acceleration, with 3.3%-8.3% underestimation by the CS technique. Evaluation of diastolic function showed 3.2%-17.6% error: E/A 2.2 [1.9-2.4] (conventional) vs. 2.3 [2.0-2.6] (CS), BA bias 0.08 [-0.81-0.96], ICC 0.82; and E/e' 4.6 [3.9-5.4] (conventional) vs. 3.8 [3.4-4.3] (CS), BA bias -0.90 [-2.31-0.50], ICC 0.89. DATA CONCLUSION: Analysis of intracardiac flow patterns and evaluation of diastolic function using a highly accelerated 4D flow sequence prototype is feasible, but it shows underestimation of flow measurements by approximately 10%. EVIDENCE LEVEL: 2 TECHNICAL EFFICACY: Stage 1.

Stent imaging on a clinical dual-source photon-counting detector CT system-impact of luminal attenuation and sharp kernels on lumen visibility.

OBJECTIVES: To assess the impact of scan modes and reconstruction kernels using a novel dual-source photon-counting detector CT (PCD-CT) on lumen visibility and sharpness of different stent sizes. METHODS: A phantom containing six balloon-expandable stents (2.5 to 9 mm diameter) in silicone tubing was scanned on a PCD-CT with standard (0.6 mm and 0.4 mm thicknesses) and ultra-high-resolution (0.2 mm thickness) modes. With the use of increasing contrast medium concentrations, densities of 0, 200, 400, and 600 HU were achieved. Standard-resolution scans were reconstructed using increasing sharpness kernels, using both polyenergetic quantitative soft tissue "conventional" ((Qr40c(0.6 mm), Qr40c(0.4 mm), Qr72c(0.2 mm)) and vascular (Bv) virtual monoenergetic reconstructions (Bv44m(0.4 mm), Bv60m(0.4 mm)) at 70 keV. In-stent lumen visibility, sharpness (max. ΔHU of the stent measured in profile plots), and in-stent noise (standard deviation of HU) were measured. RESULTS: In-stent lumen visibility was highest for Qr72c(0.2 mm) (86.5 ± 2.8% to 88.3 ± 2.6%) and in Bv60m(0.4 mm) reconstructions (77.3 ± 2.9 to 82.7 ± 2.5%). Lumen visibility was lowest in the smallest stent (2.5 mm) ranging from 54.1% in Qr40c(0.6 mm) to 74.1% in Qr72c(0.2 mm) and highest in the largest stent (9 mm) ranging from 93.8% in Qr40c(0.6 mm) to 99.1% in the Qr72c(0.2 mm) series. Lumen visibility decreased by 2.1% for every 200-HU increase in lumen attenuation. Max. ΔHU between stents and stent lumen was highest in Qr72c(0.2 mm) (ΔHU 892 ± 504 to 1526 ± 517) and Bv60m(0.4 mm) series (ΔHU 480 ± 357 to 1030 ± 344). Improvement of lumen visibility and sharpness in UHR and Bv60m(0.4 mm) series was strongest in smaller stent sizes. CONCLUSION: UHR acquisition mode and sharp reconstruction kernels on a novel PCD-CT system significantly improve in-stent lumen visibility and sharpness-especially for smaller stent sizes. KEY POINTS: • In-stent lumen visibility and sharpness of stents significantly improve using sharp reconstruction kernels (Bv60) and ultra-high-resolution mode in photon-counting detector computed tomography. • The observed improvement of stent-lumen visibility was highest in smaller stent sizes.

Reduced Iodinated Contrast Media Administration in Coronary CT Angiography on a Clinical Photon-Counting Detector CT System: A Phantom Study Using a Dynamic Circulation Model.

PURPOSE: The aim of this study was to evaluate strategies to reduce contrast media volumes for coronary computed tomography (CT) angiography on a clinical first-generation dual-source photon-counting detector (PCD)-CT system using a dynamic circulation phantom. MATERIALS AND METHODS: Coronary CT angiograph is an established method for the assessment of coronary artery disease that relies on the administration of iodinated contrast media. Reduction of contrast media volumes while maintaining diagnostic image quality is desirable. In this study, a dynamic phantom containing a 3-dimensional-printed model of the thoracic aorta and coronary arteries was evaluated using a clinical contrast injection protocol with stepwise reduced contrast agent concentrations (100%, 75%, 50%, 40%, 30%, and 20% contrast media content of the same 50 mL bolus, resulting in iodine delivery rates of 1.5, 1.1, 0.7, 0.6, 0.4 and 0.3 gl/s) on a first-generation, dual-source PCD-CT. Polychromatic images (T3D) and virtual monoenergetic images were reconstructed in the range of 40 to 70 keV in 5-keV steps. Attenuation and noise were measured in the coronary arteries and background material and the contrast-to-noise ratio (CNR) were calculated. Attenuation of 350 HU and a CNR of the reference protocol at 70 keV were regarded as sufficient for simulation of diagnostic purposes. Vessel sharpness and noise power spectra were analyzed for the aforementioned reconstructions. RESULTS: The standard clinical contrast protocol (bolus with 100% contrast) yielded diagnostic coronary artery attenuation for all tested reconstructions (>398 HU). A 50% reduction in contrast media concentration demonstrated sufficient attenuation of the coronary arteries at 40 to 55 keV (>366 HU). Virtual monoenergetic image reconstructions of 40 to 45 and 40 keV allowed satisfactory attenuation of the coronary arteries for contrast concentrations of 40% and 30% of the original protocol. A reduction of contrast agent concentration to 20% of the initial concentration provided insufficient attenuation in the target vessels for all reconstructions. The highest CNR was found for virtual monoenergetic reconstructions at 40 keV for all contrast media injection protocols, yielding a sufficient CNR at a 50% reduction of contrast agent concentration. CONCLUSIONS: Using virtual monoenergetic image reconstructions at 40 keV on a dual-source PCD-CT system, contrast media concentration could be reduced by 50% to obtain diagnostic attenuation and objective image quality for coronary CT angiography in a dynamic vessel phantom. These initial feasibility study results have to be validated in clinical studies.

Computed Tomography Assessment of Gastric Band Slippage.

Background: The purpose of this study was to develop and validate reliable computed tomography (CT) imaging criteria for the diagnosis of gastric band slippage. Material and Methods: We retrospectively evaluated 67 patients for gastric band slippage using CT. Of these, 14 had surgically proven gastric band slippage (study group), 22 had their gastric bands removed for reasons other than slippage (control group 1), and 31 did not require removal (control group 2). All of the studies were read independently by two radiologists in a blinded fashion. The "O" sign, phi angle, amount of inferior displacement from the esophageal hiatus, and gastric pouch size were used to create CT diagnostic criteria. Standard statistical methods were used. Results: There was good overall interobserver agreement for diagnosis of gastric band slippage using CT diagnostic criteria (kappa = 0.83). Agreement was excellent for the "O" sign (kappa = 0.93) and phi angle (intraclass correlation coefficient = 0.976). The "O" sign, inferior displacement from the hiatus >3.5 cm, and gastric pouch volume >55 cm3 each had 100% positive predictive value. A phi angle <20° or >60° had the highest negative predictive value (NPV) (98%). Of all CT diagnostic criteria, enlarged gastric pouch size was most correlated with band slippage with an AUC of 0.991. Conclusion: All four imaging parameters were useful in evaluating for gastric band slippage on CT, with good interobserver agreement. Of these parameters, enlarged gastric pouch size was most correlated with slippage and abnormal phi angle had the highest NPV.

Feasibility of Coronary CT Angiography-derived Left Ventricular Long-Axis Shortening as an Early Marker of Ventricular Dysfunction in Transcatheter Aortic Valve Replacement.

Purpose: To evaluate the value of using left ventricular (LV) long-axis shortening (LAS) derived from coronary CT angiography (CCTA) to predict mortality in patients with severe aortic stenosis (AS) undergoing transcatheter aortic valve replacement (TAVR). Materials and Methods: Patients with severe AS who underwent CCTA for preprocedural TAVR planning between September 2014 and December 2019 were included in this retrospective study. CCTA covered the whole cardiac cycle in 10% increments. Image series reconstructed at end systole and end diastole were used to measure LV-LAS. All-cause mortality within 24 months of follow-up after TAVR was recorded. Cox regression analysis was performed, and hazard ratios (HRs) are presented with 95% CIs. The C index was used to evaluate model performance, and the likelihood ratio χ2 test was performed to compare nested models. Results: The study included 175 patients (median age, 79 years [IQR, 73-85 years]; 92 men). The mortality rate was 22% (38 of 175). When adjusting for predictive clinical confounders, it was found that LV-LAS could be used independently to predict mortality (adjusted HR, 2.83 [95% CI: 1.13, 7.07]; P = .03). In another model using the Society of Thoracic Surgeons Predicted Risk of Mortality (STS-PROM), LV-LAS remained significant (adjusted HR, 3.38 [95 CI: 1.48, 7.72]; P = .004), and its use improved the predictive value of the STS-PROM, increasing the STS-PROM C index from 0.64 to 0.71 (χ2 = 29.9 vs 19.7, P = .001). In a subanalysis of patients with a normal LV ejection fraction (LVEF), the significance of LV-LAS persisted (adjusted HR, 3.98 [95 CI: 1.56, 10.17]; P = .004). Conclusion: LV-LAS can be used independently to predict mortality in patients undergoing TAVR, including those with a normal LVEF.Keywords: CT Angiography, Transcatheter Aortic Valve Implantation/Replacement (TAVI/TAVR), Cardiac, Outcomes Analysis, Cardiomyopathies, Left Ventricle, Aortic Valve Supplemental material is available for this article. © RSNA, 2022See also the commentary by Everett and Leipsic in this issue.

Automated diagnosis and prognosis of COVID-19 pneumonia from initial ER chest X-rays using deep learning.

BACKGROUND: Airspace disease as seen on chest X-rays is an important point in triage for patients initially presenting to the emergency department with suspected COVID-19 infection. The purpose of this study is to evaluate a previously trained interpretable deep learning algorithm for the diagnosis and prognosis of COVID-19 pneumonia from chest X-rays obtained in the ED. METHODS: This retrospective study included 2456 (50% RT-PCR positive for COVID-19) adult patients who received both a chest X-ray and SARS-CoV-2 RT-PCR test from January 2020 to March of 2021 in the emergency department at a single U.S. INSTITUTION: A total of 2000 patients were included as an additional training cohort and 456 patients in the randomized internal holdout testing cohort for a previously trained Siemens AI-Radiology Companion deep learning convolutional neural network algorithm. Three cardiothoracic fellowship-trained radiologists systematically evaluated each chest X-ray and generated an airspace disease area-based severity score which was compared against the same score produced by artificial intelligence. The interobserver agreement, diagnostic accuracy, and predictive capability for inpatient outcomes were assessed. Principal statistical tests used in this study include both univariate and multivariate logistic regression. RESULTS: Overall ICC was 0.820 (95% CI 0.790-0.840). The diagnostic AUC for SARS-CoV-2 RT-PCR positivity was 0.890 (95% CI 0.861-0.920) for the neural network and 0.936 (95% CI 0.918-0.960) for radiologists. Airspace opacities score by AI alone predicted ICU admission (AUC = 0.870) and mortality (0.829) in all patients. Addition of age and BMI into a multivariate log model improved mortality prediction (AUC = 0.906). CONCLUSION: The deep learning algorithm provides an accurate and interpretable assessment of the disease burden in COVID-19 pneumonia on chest radiographs. The reported severity scores correlate with expert assessment and accurately predicts important clinical outcomes. The algorithm contributes additional prognostic information not currently incorporated into patient management.

Impact of Artificial Intelligence Assistance on Chest CT Interpretation Times: A Prospective Randomized Study.

BACKGROUND. Deep learning-based convolutional neural networks have enabled major advances in development of artificial intelligence (AI) software applications. Modern AI applications offer comprehensive multiorgan evaluation. OBJECTIVE. The purpose of this article was to evaluate the impact of an automated AI platform integrated into clinical workflow for chest CT interpretation on radiologists' interpretation times when evaluated in a real-world clinical setting. METHODS. In this prospective single-center study, a commercial AI software solution was integrated into clinical workflow for chest CT interpretation. The software provided automated analysis of cardiac, pulmonary, and musculoskeletal findings, including labeling, segmenting, and measuring normal structures as well as detecting, labeling, and measuring abnormalities. AI-annotated images and autogenerated summary results were stored in the PACS and available to interpreting radiologists. A total of 390 patients (204 women, 186 men; mean age, 62.8 ± 13.3 [SD] years) who underwent out-patient chest CT between January 19, 2021, and January 28, 2021, were included. Scans were randomized using 1:1 allocation between AI-assisted and non-AI-assisted arms and were clinically interpreted by one of three cardiothoracic radiologists (65 scans per arm per radiologist; total of 195 scans per arm) who recorded interpretation times using a stopwatch. Findings were categorized according to review of report impressions. Interpretation times were compared between arms. RESULTS. Mean interpretation times were significantly shorter in the AI-assisted than in the non-AI-assisted arm for all three readers (289 ± 89 vs 344 ± 129 seconds, p < .001; 449 ± 110 vs 649 ± 82 seconds, p < .001; 281 ± 114 vs 348 ± 93 seconds, p = .01) and for readers combined (328 ± 122 vs 421 ± 175 seconds, p < .001). For readers combined, the mean difference was 93 seconds (95% CI, 63-123 seconds), corresponding with a 22.1% reduction in the AI-assisted arm. Mean interpretation time was also shorter in the AI-assisted arm compared with the non-AI-assisted arm for contrast-enhanced scans (83 seconds), noncontrast scans (104 seconds), negative scans (84 seconds), positive scans without significant new findings (117 seconds), and positive scans with significant new findings (92 seconds). CONCLUSION. Cardiothoracic radiologists exhibited a 22.1% reduction in chest CT interpretations times when they had access to results from an automated AI support platform during real-world clinical practice. CLINICAL IMPACT. Integration of the AI support platform into clinical workflow improved radiologist efficiency.

An Interpretable Chest CT Deep Learning Algorithm for Quantification of COVID-19 Lung Disease and Prediction of Inpatient Morbidity and Mortality.

RATIONALE AND OBJECTIVES: The burden of coronavirus disease 2019 (COVID-19) airspace opacities is time consuming and challenging to quantify on computed tomography. The purpose of this study was to evaluate the ability of a deep convolutional neural network (dCNN) to predict inpatient outcomes associated with COVID-19 pneumonia. MATERIALS AND METHODS: A previously trained dCNN was tested on an external validation cohort of 241 patients who presented to the emergency department and received a chest computed tomography scan, 93 with COVID-19 and 168 without. Airspace opacity scoring systems were defined by the extent of airspace opacity in each lobe, totaled across the entire lungs. Expert and dCNN scores were concurrently evaluated for interobserver agreement, while both dCNN identified airspace opacity scoring and raw opacity values were used in the prediction of COVID-19 diagnosis and inpatient outcomes. RESULTS: Interobserver agreement for airspace opacity scoring was 0.892 (95% CI 0.834-0.930). Probability of each outcome behaved as a logistic function of the opacity scoring (25% intensive care unit admission at score of 13/25, 25% intubation at 17/25, and 25% mortality at 20/25). Length of hospitalization, intensive care unit stay, and intubation were associated with larger airspace opacity score (p = 0.032, 0.039, 0.036, respectively). CONCLUSION: The tested dCNN was highly predictive of inpatient outcomes, performs at a near expert level, and provides added value for clinicians in terms of prognostication and disease severity.

Impact of machine-learning-based coronary computed tomography angiography-derived fractional flow reserve on decision-making in patients with severe aortic stenosis undergoing transcatheter aortic valve replacement.

OBJECTIVES: To evaluate feasibility and diagnostic performance of coronary CT angiography (CCTA)-derived fractional flow reserve (CT-FFR) for detection of significant coronary artery disease (CAD) and decision-making in patients with severe aortic stenosis (AS) undergoing transcatheter aortic valve replacement (TAVR) to potentially avoid additional pre-TAVR invasive coronary angiography (ICA). METHODS: Consecutive patients with severe AS (n = 95, 78.6 ± 8.8 years, 53% female) undergoing pre-procedural TAVR-CT followed by ICA with quantitative coronary angiography were retrospectively analyzed. CCTA datasets were evaluated using CAD Reporting and Data System (CAD-RADS) classification. CT-FFR measurements were computed using an on-site machine-learning algorithm. A combined algorithm was developed for decision-making to determine if ICA is needed based on pre-TAVR CCTA: [1] all patients with CAD-RADS ≥ 4 are referred for ICA; [2] patients with CAD-RADS 2 and 3 are evaluated utilizing CT-FFR and sent to ICA if CT-FFR ≤ 0.80; [3] patients with CAD-RADS < 2 or CAD-RADS 2-3 and normal CT-FFR are not referred for ICA. RESULTS: Twelve patients (13%) had significant CAD (≥ 70% stenosis) on ICA and were treated with PCI. Twenty-eight patients (30%) showed CT-FFR ≤ 0.80 and 24 (86%) of those were reported to have a maximum stenosis ≥ 50% during ICA. Using the proposed algorithm, significant CAD could be identified with a sensitivity, specificity, and positive and negative predictive value of 100%, 78%, 40%, and 100%, respectively, potentially decreasing the number of necessary ICAs by 65 (68%). CONCLUSION: Combination of CT-FFR and CAD-RADS is able to identify significant CAD pre-TAVR and bears potential to significantly reduce the number of needed ICAs. KEY POINTS: • Coronary CT angiography-derived fractional flow reserve (CT-FFR) using machine learning together with the CAD Reporting and Data System (CAD-RADS) classification safely identifies significant coronary artery disease based on quantitative coronary angiography in patients prior to transcatheter aortic valve replacement. • The combination of CT-FFR and CAD-RADS enables decision-making and bears the potential to significantly reduce the number of needed invasive coronary angiographies.

Automated Dual-energy Computed Tomography-based Extracellular Volume Estimation for Myocardial Characterization in Patients With Ischemic and Nonischemic Cardiomyopathy.

OBJECTIVES: We aimed to validate and test a prototype algorithm for automated dual-energy computed tomography (DECT)-based myocardial extracellular volume (ECV) assessment in patients with various cardiomyopathies. METHODS: This retrospective study included healthy subjects (n=9; 61±10 y) and patients with cardiomyopathy (n=109, including a validation cohort n=60; 68±9 y; and a test cohort n=49; 69±11 y), who had previously undergone cardiac DECT. Myocardial ECV was calculated using a prototype-based fully automated algorithm and compared with manual assessment. Receiver-operating characteristic analysis was performed to test the algorithm's ability to distinguish healthy subjects and patients with cardiomyopathy. RESULTS: The fully automated method led to a significant reduction of postprocessing time compared with manual assessment (2.2±0.4 min and 9.4±0.7 min, respectively, P <0.001). There was no significant difference in ECV between the automated and manual methods ( P =0.088). The automated method showed moderate correlation and agreement with the manual technique ( r =0.68, intraclass correlation coefficient=0.66). ECV was significantly higher in patients with cardiomyopathy compared with healthy subjects, regardless of the method used ( P <0.001). In the test cohort, the automated method yielded an area under the curve of 0.98 for identifying patients with cardiomyopathies. CONCLUSION: Automated ECV estimation based on DECT showed moderate agreement with the manual method and matched with previously reported ECV values for healthy volunteers and patients with cardiomyopathy. The automatically derived ECV demonstrated an excellent diagnostic performance to discriminate between healthy and diseased myocardium, suggesting that it could be an effective initial screening tool while significantly reducing the time of assessment.

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

OBJECTIVES: To evaluate the effectiveness of a novel artificial intelligence (AI) algorithm for fully automated measurement of left atrial (LA) volumes and function using cardiac CT in patients with atrial fibrillation. METHODS: We included 79 patients (mean age 63 ± 12 years; 35 with atrial fibrillation (AF) and 44 controls) between 2017 and 2020 in this retrospective study. Images were analyzed by a trained AI algorithm and an expert radiologist. Left atrial volumes were obtained at cardiac end-systole, end-diastole, and pre-atrial contraction, which were then used to obtain LA function indices. Intraclass correlation coefficient (ICC) analysis of the LA volumes and function parameters was performed and receiver operating characteristic (ROC) curve analysis was used to compare the ability to detect AF patients. RESULTS: The AI was significantly faster than manual measurement of LA volumes (4 s vs 10.8 min, respectively). Agreement between the manual and automated methods was good to excellent overall, and there was stronger agreement in AF patients (all ICCs ≥ 0.877; p < 0.001) than controls (all ICCs ≥ 0.799; p < 0.001). The AI comparably estimated LA volumes in AF patients (all within 1.3 mL of the manual measurement), but overestimated volumes by clinically negligible amounts in controls (all by ≤ 4.2 mL). The AI's ability to distinguish AF patients from controls using the LA volume index was similar to the expert's (AUC 0.81 vs 0.82, respectively; p = 0.62). CONCLUSION: The novel AI algorithm efficiently performed fully automated multiphasic CT-based quantification of left atrial volume and function with similar accuracy as compared to manual quantification. Novel CT-based AI algorithm efficiently quantifies left atrial volumes and function with similar accuracy as manual quantification in controls and atrial fibrillation patients. KEY POINTS: • There was good-to-excellent agreement between manual and automated methods for left atrial volume quantification. • The AI comparably estimated LA volumes in AF patients, but overestimated volumes by clinically negligible amounts in controls. • The AI's ability to distinguish AF patients from controls was similar to the manual methods.

Coronary Computed Tomography Angiography-Based Calcium Scoring: In Vitro and In Vivo Validation of a Novel Virtual Noniodine Reconstruction Algorithm on a Clinical, First-Generation Dual-Source Photon Counting-Detector System.

PURPOSE: The aim of this study was to evaluate coronary computed tomography angiography (CCTA)-based in vitro and in vivo coronary artery calcium scoring (CACS) using a novel virtual noniodine reconstruction (PureCalcium) on a clinical first-generation photon-counting detector-computed tomography system compared with virtual noncontrast (VNC) reconstructions and true noncontrast (TNC) acquisitions. MATERIALS AND METHODS: Although CACS and CCTA are well-established techniques for the assessment of coronary artery disease, they are complementary acquisitions, translating into increased scan time and patient radiation dose. Hence, accurate CACS derived from a single CCTA acquisition would be highly desirable. In this study, CACS based on PureCalcium, VNC, and TNC, reconstructions was evaluated in a CACS phantom and in 67 patients (70 [59/80] years, 58.2% male) undergoing CCTA on a first-generation photon counting detector-computed tomography system. Coronary artery calcium scores were quantified for the 3 reconstructions and compared using Wilcoxon test. Agreement was evaluated by Pearson and Spearman correlation and Bland-Altman analysis. Classification of coronary artery calcium score categories (0, 1-10, 11-100, 101-400, and >400) was compared using Cohen κ . RESULTS: Phantom studies demonstrated strong agreement between CACS PureCalcium and CACS TNC (60.7 ± 90.6 vs 67.3 ± 88.3, P = 0.01, r = 0.98, intraclass correlation [ICC] = 0.98; mean bias, 6.6; limits of agreement [LoA], -39.8/26.6), whereas CACS VNC showed a significant underestimation (42.4 ± 75.3 vs 67.3 ± 88.3, P < 0.001, r = 0.94, ICC = 0.89; mean bias, 24.9; LoA, -87.1/37.2). In vivo comparison confirmed a high correlation but revealed an underestimation of CACS PureCalcium (169.3 [0.7/969.4] vs 232.2 [26.5/1112.2], P < 0.001, r = 0.97, ICC = 0.98; mean bias, -113.5; LoA, -470.2/243.2). In comparison, CACS VNC showed a similarly high correlation, but a substantially larger underestimation (24.3 [0/272.3] vs 232.2 [26.5/1112.2], P < 0.001, r = 0.97, ICC = 0.54; mean bias, -551.6; LoA, -2037.5/934.4). CACS PureCalcium showed superior agreement of CACS classification ( κ = 0.88) than CACS VNC ( κ = 0.60). CONCLUSIONS: The accuracy of CACS quantification and classification based on PureCalcium reconstructions of CCTA outperforms CACS derived from VNC reconstructions.

Tumor burden of lung metastases at initial staging in breast cancer patients detected by artificial intelligence as a prognostic tool for precision medicine.

BACKGROUND: Determination of the total number and size of all pulmonary metastases on chest CT is time-consuming and as such has been understudied as an independent metric for disease assessment. A novel artificial intelligence (AI) model may allow for automated detection, size determination, and quantification of the number of pulmonary metastases on chest CT. OBJECTIVE: To investigate the utility of a novel AI program applied to initial staging chest CT in breast cancer patients in risk assessment of mortality and survival. METHODS: Retrospective imaging data from a cohort of 226 subjects with breast cancer was assessed by the novel AI program and the results validated by blinded readers. Mean clinical follow-up was 2.5 years for outcomes including cancer-related death and development of extrapulmonary metastatic disease. AI measurements including total number of pulmonary metastases and maximum nodule size were assessed by Cox-proportional hazard modeling and adjusted survival. RESULTS: 752 lung nodules were identified by the AI program, 689 of which were identified in 168 subjects having confirmed lung metastases (Lmet+) and 63 were identified in 58 subjects without confirmed lung metastases (Lmet-). When compared to the reader assessment, AI had a per-patient sensitivity, specificity, PPV and NPV of 0.952, 0.639, 0.878, and 0.830. Mortality in the Lmet + group was four times greater compared to the Lmet-group (p = 0.002). In a multivariate analysis, total lung nodule count by AI had a high correlation with overall mortality (OR 1.11 (range 1.07-1.15), p < 0.001) with an AUC of 0.811 (R2 = 0.226, p < 0.0001). When total lung nodule count and maximum nodule diameter were combined there was an AUC of 0.826 (R2 = 0.243, p < 0.001). CONCLUSION: Automated AI-based detection of lung metastases in breast cancer patients at initial staging chest CT performed well at identifying pulmonary metastases and demonstrated strong correlation between the total number and maximum size of lung metastases with future mortality. CLINICAL IMPACT: As a component of precision medicine, AI-based measurements at the time of initial staging may improve prediction of which breast cancer patients will have negative future outcomes.