Explainable deep-learning-based ischemia detection using hybrid O-15 H2O perfusion positron emission tomography and computed tomography imaging with clinical data.

BACKGROUND: We developed an explainable deep-learning (DL)-based classifier to identify flow-limiting coronary artery disease (CAD) by O-15 H2O perfusion positron emission tomography computed tomography (PET/CT) and coronary CT angiography (CTA) imaging. The classifier uses polar map images with numerical data and visualizes data findings. METHODS: A DLmodel was implemented and evaluated on 138 individuals, consisting of a combined image-and data-based classifier considering 35 clinical, CTA, and PET variables. Data from invasive coronary angiography were used as reference. Performance was evaluated with clinical classification using accuracy (ACC), area under the receiver operating characteristic curve (AUC), F1 score (F1S), sensitivity (SEN), specificity (SPE), precision (PRE), net benefit, and Cohen's Kappa. Statistical testing was conducted using McNemar's test. RESULTS: The DL model had a median ACC = 0.8478, AUC = 0.8481, F1S = 0.8293, SEN = 0.8500, SPE = 0.8846, and PRE = 0.8500. Improved detection of true-positive and false-negative cases, increased net benefit in thresholds up to 34%, and comparable Cohen's kappa was seen, reaching similar performance to clinical reading. Statistical testing revealed no significant differences between DL model and clinical reading. CONCLUSIONS: The combined DL model is a feasible and an effective method in detection of CAD, allowing to highlight important data findings individually in interpretable manner.

Incorporating coronary artery calcium scoring in the prediction of obstructive coronary artery disease with myocardial ischemia: a study with sequential use of coronary computed tomography angiography and positron emission tomography imaging.

BACKGROUND: Additional strategies are needed to refine the referral for diagnostic testing of symptomatic patients with suspected coronary artery disease (CAD). We aimed to compare various models to predict hemodynamically obstructive CAD. METHODS AND RESULTS: Symptomatic patients with suspected CAD who underwent coronary artery calcium scoring (CACS) and sequential coronary computed tomography angiography (CCTA) and [15O]H2O positron emission tomography (PET) myocardial perfusion imaging were analyzed. Obstructive CAD was defined as a suspected coronary artery stenosis on CCTA with myocardial ischemia on PET (absolute stress myocardial perfusion ≤ 2.4 mL/g/min in ≥ 1 segment). Three models were developed to predict obstructive CAD-induced myocardial ischemia using logistic regression analysis: (1) basic model: including age, sex and cardiac symptoms, (2) risk factor model: adding number of risk factors to the basic model, and (3) CACS model: adding CACS to the risk factor model. Model performance was evaluated using discriminatory ability with area under the receiver-operating characteristic curves (AUC). A total of 647 patients (mean age 62 ± 9 years, 45% men) underwent CACS and sequential CCTA and PET myocardial perfusion imaging. Obstructive CAD with myocardial ischemia on PET was present in 151 (23%) patients. CACS was independently associated with myocardial ischemia (P < .001). AUC for the discrimination of ischemia for the CACS model was superior over the basic model and risk factor model (P < .001). CONCLUSIONS: Adding CACS to the model including age, sex, cardiac symptoms and number of risk factors increases the accuracy to predict obstructive CAD with myocardial ischemia on PET in symptomatic patients with suspected CAD.

Effect of respiratory motion correction and CT-based attenuation correction on dual-gated cardiac PET image quality and quantification.

BACKGROUND: Dual-gating reduces respiratory and cardiac motion effects but increases noise. With motion correction, motion is minimized and image quality preserved. We applied motion correction to create end-diastolic respiratory motion corrected images from dual-gated images. METHODS: [18F]-fluorodeoxyglucose ([18F]-FDG) PET images of 13 subjects were reconstructed with 4 methods: non-gated, dual-gated, motion corrected, and motion corrected with 4D-CT (MoCo-4D). Image quality was evaluated using standardized uptake values, contrast ratio, signal-to-noise ratio, coefficient of variation, and contrast-to-noise ratio. Motion minimization was evaluated using myocardial wall thickness. RESULTS: MoCo-4D showed improvement for contrast ratio (2.83 vs 2.76), signal-to-noise ratio (27.5 vs 20.3) and contrast-to-noise ratio (14.5 vs 11.1) compared to dual-gating. The uptake difference between MoCo-4D and non-gated images was non-significant (P > .05) for the myocardium (2.06 vs 2.15 g/mL), but significant (P < .05) for the blood pool (.80 vs .86 g/mL). Non-gated images had the lowest coefficient of variation (27.3%), with significant increase for all other methods (31.6-32.5%). MoCo-4D showed smallest myocardial wall thickness (16.6 mm) with significant decrease compared to non-gated images (20.9 mm). CONCLUSIONS: End-diastolic respiratory motion correction and 4D-CT resulted in improved motion minimization and image quality over standard dual-gating.

A Computational Framework for Data Fusion in MEMS-Based Cardiac and Respiratory Gating.

Dual cardiac and respiratory gating is a well-known technique for motion compensation in nuclear medicine imaging. In this study, we present a new data fusion framework for dual cardiac and respiratory gating based on multidimensional microelectromechanical (MEMS) motion sensors. Our approach aims at robust estimation of the chest vibrations, that is, high-frequency precordial vibrations and low-frequency respiratory movements for prospective gating in positron emission tomography (PET), computed tomography (CT), and radiotherapy. Our sensing modality in the context of this paper is a single dual sensor unit, including accelerometer and gyroscope sensors to measure chest movements in three different orientations. Since accelerometer- and gyroscope-derived respiration signals represent the inclination of the chest, they are similar in morphology and have the same units. Therefore, we use principal component analysis (PCA) to combine them into a single signal. In contrast to this, the accelerometer- and gyroscope-derived cardiac signals correspond to the translational and rotational motions of the chest, and have different waveform characteristics and units. To combine these signals, we use independent component analysis (ICA) in order to obtain the underlying cardiac motion. From this cardiac motion signal, we obtain the systolic and diastolic phases of cardiac cycles by using an adaptive multi-scale peak detector and a short-time autocorrelation function. Three groups of subjects, including healthy controls (n = 7), healthy volunteers (n = 12), and patients with a history of coronary artery disease (n = 19) were studied to establish a quantitative framework for assessing the performance of the presented work in prospective imaging applications. The results of this investigation showed a fairly strong positive correlation (average r = 0.73 to 0.87) between the MEMS-derived (including corresponding PCA fusion) respiration curves and the reference optical camera and respiration belt sensors. Additionally, the mean time offset of MEMS-driven triggers from camera-driven triggers was 0.23 to 0.3 ± 0.15 to 0.17 s. For each cardiac cycle, the feature of the MEMS signals indicating a systolic time interval was identified, and its relation to the total cardiac cycle length was also reported. The findings of this study suggest that the combination of chest angular velocity and accelerations using ICA and PCA can help to develop a robust dual cardiac and respiratory gating solution using only MEMS sensors. Therefore, the methods presented in this paper should help improve predictions of the cardiac and respiratory quiescent phases, particularly with the clinical patients. This study lays the groundwork for future research into clinical PET/CT imaging based on dual inertial sensors.

Verifying the Quantumness of Bipartite Correlations.

Entanglement is at the heart of most quantum information tasks, and therefore considerable effort has been made to find methods of deciding the entanglement content of a given bipartite quantum state. Here, we prove a fundamental limitation to deciding if an unknown state is entangled or not: we show that any quantum measurement which can answer this question for an arbitrary state necessarily gives enough information to identify the state completely. We also extend our treatment to other classes of correlated states by considering the problem of deciding if a state has negative partial transpose, is discordant, or is fully classically correlated. Remarkably, only the question related to quantum discord can be answered without resorting to full state tomography.

Importance of plaque volume and composition for the prediction of myocardial ischaemia using sequential coronary computed tomography angiography/positron emission tomography imaging.

AIMS: Coronary atherosclerosis with a large necrotic core has been postulated to reduce the vasodilatory capacity of vascular tissue. In the present analysis, we explored whether total plaque volume and necrotic core volume on coronary computed tomography angiography (CCTA) are independently associated with myocardial ischaemia on positron emission tomography (PET). METHODS AND RESULTS: From a registry of symptomatic patients with suspected coronary artery disease and clinically indicated CCTA with sequential [15O]H2O PET myocardial perfusion imaging, we quantitatively measured diameter stenosis, total and compositional plaque volumes on CCTA. Primary endpoint was myocardial ischaemia on PET, defined as an absolute stress myocardial blood flow ≤2.4 mL/g/min in ≥1 segment. Multivariable prediction models for myocardial ischaemia were consecutively created using logistic regression analysis (stenosis model: diameter stenosis ≥50%; plaque volume model: +total plaque volume; plaque composition model: +necrotic core volume). A total of 493 patients (mean age 63 ± 8 years, 54% men) underwent sequential CCTA/PET imaging. In 153 (31%) patients, myocardial ischaemia was detected on PET. Diameter stenosis ≥50% (P < 0.001) and necrotic core volume (P = 0.029) were independently associated with myocardial ischaemia, while total plaque volume showed borderline significance (P = 0.052). The plaque composition model (χ2 = 169) provided incremental value for the prediction of ischaemia when compared with the stenosis model (χ2 = 138, P < 0.001) and plaque volume model (χ2 = 164, P = 0.021). CONCLUSION: The volume of necrotic core on CCTA independently and incrementally predicts myocardial ischaemia on PET, beyond diameter stenosis alone.

Plaque volume, composition, and fraction versus ischemia and outcomes in patients with coronary artery disease.

BACKGROUND: The various plaque components have been associated with ischemia and outcomes in patients with coronary artery disease (CAD). The main goal of this analysis was to test the hypothesis that, at patient level, the fraction of non-calcified plaque volume (PV) of total PV is associated with ischemia and outcomes in patients with CAD. This ratio could be a simple and clinically useful parameter, if predicting outcomes. METHODS: Consecutive patients with suspected CAD undergoing coronary computed tomography angiography with selective positron emission tomography perfusion imaging were selected. Plaque components were quantitatively analyzed at patient level. The fraction of various plaque components were expressed as percentage of total PV and examined among patients with non-obstructive CAD, suspected stenosis with normal perfusion, and those with reduced myocardial perfusion. Clinical outcomes included all-cause mortality and myocardial infarction. RESULTS: In total, 494 patients (age 63 â€‹± â€‹9 years, 55% male) were included. Total PV and all plaque components were significantly larger in patients with reduced myocardial perfusion compared to patients with normal perfusion and those with non-obstructive CAD. During follow-up 35 events occurred. Patients with any plaque component â€‹≥ â€‹median showed worse outcomes (log-rank p â€‹< â€‹0.001 for all). In addition, low-attenuation plaque â€‹≥ â€‹median was associated with worse outcomes independent of total PV (adjusted HR: 2.754, 95% CI: 1.022-7.0419, p â€‹= â€‹0.045). The fractions of the various plaque components were not associated with outcomes. CONCLUSION: Larger total PV or any plaque component at patient level are associated with abnormal myocardial perfusion and adverse events. The various plaque components as fraction of total PV lack additional prognostic value.

Coronary computed tomography angiography-based endothelial wall shear stress in normal coronary arteries.

Endothelial wall shear stress (ESS) is a biomechanical force which plays a role in the formation and evolution of atherosclerotic lesions. The purpose of this study is to evaluate coronary computed tomography angiography (CCTA)-based ESS in coronary arteries without atherosclerosis, and to assess factors affecting ESS values. CCTA images from patients with suspected coronary artery disease were analyzed to identify coronary arteries without atherosclerosis. Minimal and maximal ESS values were calculated for 3-mm segments. Factors potentially affecting ESS values were examined, including sex, lumen diameter and distance from the ostium. Segments were categorized according to lumen diameter tertiles into small (< 2.6 mm), intermediate (2.6-3.2 mm) or large (≥ 3.2 mm) segments. A total of 349 normal vessels from 168 patients (mean age 59 ± 9 years, 39% men) were included. ESS was highest in the left anterior descending artery compared to the left circumflex artery and right coronary artery (minimal ESS 2.3 Pa vs. 1.9 Pa vs. 1.6 Pa, p < 0.001 and maximal ESS 3.7 Pa vs. 3.0 Pa vs. 2.5 Pa, p < 0.001). Men had lower ESS values than women, also after adjusting for lumen diameter (p < 0.001). ESS values were highest in small segments compared to intermediate or large segments (minimal ESS 3.8 Pa vs. 1.7 Pa vs. 1.2 Pa, p < 0.001 and maximal ESS 6.0 Pa vs. 2.6 Pa vs. 2.0 Pa, p < 0.001). A weak to strong correlation was found between ESS and distance from the ostium (ρ = 0.22-0.62, p < 0.001). CCTA-based ESS values increase rapidly and become widely scattered with decreasing lumen diameter. This needs to be taken into account when assessing the added value of ESS beyond lumen diameter in highly stenotic lesions.

Classification of ischemia from myocardial polar maps in 15O-H2O cardiac perfusion imaging using a convolutional neural network.

We implemented a two-dimensional convolutional neural network (CNN) for classification of polar maps extracted from Carimas (Turku PET Centre, Finland) software used for myocardial perfusion analysis. 138 polar maps from 15O-H2O stress perfusion study in JPEG format from patients classified as ischemic or non-ischemic based on finding obstructive coronary artery disease (CAD) on invasive coronary artery angiography were used. The CNN was evaluated against the clinical interpretation. The classification accuracy was evaluated with: accuracy (ACC), area under the receiver operating characteristic curve (AUC), F1 score (F1S), sensitivity (SEN), specificity (SPE) and precision (PRE). The CNN had a median ACC of 0.8261, AUC of 0.8058, F1S of 0.7647, SEN of 0.6500, SPE of 0.9615 and PRE of 0.9286. In comparison, clinical interpretation had ACC of 0.8696, AUC of 0.8558, F1S of 0.8333, SEN of 0.7500, SPE of 0.9615 and PRE of 0.9375. The CNN classified only 2 cases differently than the clinical interpretation. The clinical interpretation and CNN had similar accuracy in classifying false positives and true negatives. Classification of ischemia is feasible in 15O-H2O stress perfusion imaging using JPEG polar maps alone with a custom CNN and may be useful for the detection of obstructive CAD.

Detailed behaviour of endothelial wall shear stress across coronary lesions from non-invasive imaging with coronary computed tomography angiography.

AIMS: Evolving evidence suggests that endothelial wall shear stress (ESS) plays a crucial role in the rupture and progression of coronary plaques by triggering biological signalling pathways. We aimed to investigate the patterns of ESS across coronary lesions from non-invasive imaging with coronary computed tomography angiography (CCTA), and to define plaque-associated ESS values in patients with coronary artery disease (CAD). METHODS AND RESULTS: Symptomatic patients with CAD who underwent a clinically indicated CCTA scan were identified. Separate core laboratories performed blinded analysis of CCTA for anatomical and ESS features of coronary atherosclerosis. ESS was assessed using dedicated software, providing minimal and maximal ESS values for each 3 mm segment. Each coronary lesion was divided into upstream, start, minimal luminal area (MLA), end and downstream segments. Also, ESS ratios were calculated using the upstream segment as a reference. From 122 patients (mean age 64 ± 7 years, 57% men), a total of 237 lesions were analyzed. Minimal and maximal ESS values varied across the lesions with the highest values at the MLA segment [minimal ESS 3.97 Pa (IQR 1.93-8.92 Pa) and maximal ESS 5.64 Pa (IQR 3.13-11.21 Pa), respectively]. Furthermore, minimal and maximal ESS values were positively associated with stenosis severity (P < 0.001), percent atheroma volume (P < 0.001), and lesion length (P ≤ 0.023) at the MLA segment. Using ESS ratios, similar associations were observed for stenosis severity and lesion length. CONCLUSIONS: Detailed behaviour of ESS across coronary lesions can be derived from routine non-invasive CCTA imaging. This may further improve risk stratification.

Probing quantum state space: does one have to learn everything to learn something?

Determining the state of a quantum system is a consuming procedure. For this reason, whenever one is interested only in some particular property of a state, it would be desirable to design a measurement set-up that reveals this property with as little effort as possible. Here, we investigate whether, in order to successfully complete a given task of this kind, one needs an informationally complete measurement, or if something less demanding would suffice. The first alternative means that in order to complete the task, one needs a measurement which fully determines the state. We formulate the task as a membership problem related to a partitioning of the quantum state space and, in doing so, connect it to the geometry of the state space. For a general membership problem, we prove various sufficient criteria that force informational completeness, and we explicitly treat several physically relevant examples. For the specific cases that do not require informational completeness, we also determine bounds on the minimal number of measurement outcomes needed to ensure success in the task.