Abstract
Aims
To optimize per-vessel prediction of early coronary revascularization (ECR) within 90 days after fast single-photon emission computed tomography (SPECT) myocardial perfusion imaging ...(MPI) using machine learning (ML) and introduce a method for a patient-specific explanation of ML results in a clinical setting.
Methods and results
A total of 1980 patients with suspected coronary artery disease (CAD) underwent stress/rest 99mTc-sestamibi/tetrofosmin MPI with new-generation SPECT scanners were included. All patients had invasive coronary angiography within 6 months after SPECT MPI. ML utilized 18 clinical, 9 stress test, and 28 imaging variables to predict per-vessel and per-patient ECR with 10-fold cross-validation. Area under the receiver operator characteristics curve (AUC) of ML was compared with standard quantitative analysis total perfusion deficit (TPD) and expert interpretation. ECR was performed in 958 patients (48%). Per-vessel, the AUC of ECR prediction by ML (AUC 0.79, 95% confidence interval (CI) 0.77, 0.80) was higher than by regional stress TPD (0.71, 0.70, 0.73), combined-view stress TPD (AUC 0.71, 95% CI 0.69, 0.72), or ischaemic TPD (AUC 0.72, 95% CI 0.71, 0.74), all P < 0.001. Per-patient, the AUC of ECR prediction by ML (AUC 0.81, 95% CI 0.79, 0.83) was higher than that of stress TPD, combined-view TPD, and ischaemic TPD, all P < 0.001. ML also outperformed nuclear cardiologists’ expert interpretation of MPI for the prediction of early revascularization performance. A method to explain ML prediction for an individual patient was also developed.
Conclusion
In patients with suspected CAD, the prediction of ECR by ML outperformed automatic MPI quantitation by TPDs (per-vessel and per-patient) or nuclear cardiologists’ expert interpretation (per-patient).
Precise definition of the mitral valve plane (VP) during segmentation of the left ventricle for SPECT myocardial perfusion imaging (MPI) quantification often requires manual adjustment, which affects ...the quantification of perfusion. We developed a machine learning approach using support vector machines (SVM) for automatic VP placement.
A total of 392 consecutive patients undergoing
Tc-tetrofosmin stress (5 min; mean ± SD, 350 ± 54 MBq) and rest (5 min; 1,024 ± 153 MBq) fast SPECT MPI attenuation corrected (AC) by CT and same-day coronary CT angiography were studied; included in the 392 patients were 48 patients who underwent invasive coronary angiography and had no known coronary artery disease. The left ventricle was segmented with standard clinical software (quantitative perfusion SPECT) by 2 experts, adjusting the VP if needed. Two-class SVM models were computed from the expert placements with 10-fold cross validation to separate the patients used for training and those used for validation. SVM probability estimates were used to compute the best VP position. Automatic VP localizations on AC and non-AC images were compared with expert placement on coronary CT angiography. Stress and rest total perfusion deficits and detection of per-vessel obstructive stenosis by invasive coronary angiography were also compared.
Bland-Altman 95% confidence intervals (CIs) for VP localization by SVM and experts for AC stress images (bias, 1; 95% CI, -5 to 7 mm) and AC rest images (bias, 1; 95% CI, -7 to 10 mm) were narrower than interexpert 95% CIs for AC stress images (bias, 0; 95% CI, -8 to 8 mm) and AC rest images (bias, 0; 95% CI, -10 to 10 mm) (
< 0.01). Bland-Altman 95% CIs for VP localization by SVM and experts for non-AC stress images (bias, 1; 95% CI, -4 to 6 mm) and non-AC rest images (bias, 2; 95% CI, -7 to 10 mm) were similar to interexpert 95% CIs for non-AC stress images (bias, 0; 95% CI, -6 to 5 mm) and non-AC rest images (bias, -1; 95% CI, -9 to 7 mm) (
was not significant NS). For regional detection of obstructive stenosis, ischemic total perfusion deficit areas under the receiver operating characteristic curve for the 2 experts (AUC, 0.79 95% CI, 0.7-0.87; AUC, 0.81 95% CI, 0.73-0.89) and the SVM (0.82 0.74-0.9) for AC data were the same (
= NS) and were higher than those for the unadjusted VP (0.63 0.53-0.73) (
< 0.01). Similarly, for non-AC data, areas under the receiver operating characteristic curve for the experts (AUC, 0.77 95% CI, 0.69-0.89; AUC, 0.8 95% CI, 0.72-0.88) and the SVM (0.79 0.71-0.87) were the same (
= NS) and were higher than those for the unadjusted VP (0.65 0.56-0.75) (
< 0.01).
Machine learning with SVM allows automatic and accurate VP localization, decreasing user dependence in SPECT MPI quantification.
Objectives
To evaluate the diagnostic performance of a deep learning algorithm for automated detection of small
18
F-FDG-avid pulmonary nodules in PET scans, and to assess whether novel block ...sequential regularized expectation maximization (BSREM) reconstruction affects detection accuracy as compared to ordered subset expectation maximization (OSEM) reconstruction.
Methods
Fifty-seven patients with 92
18
F-FDG-avid pulmonary nodules (all ≤ 2 cm) undergoing PET/CT for oncological (re-)staging were retrospectively included and a total of 8824 PET images of the lungs were extracted using OSEM and BSREM reconstruction. Per-slice and per-nodule sensitivity of a deep learning algorithm was assessed, with an expert readout by a radiologist/nuclear medicine physician serving as standard of reference. Receiver-operator characteristic (ROC) curve of OSEM and BSREM were assessed and the areas under the ROC curve (AUC) were compared. A maximum standardized uptake value (SUV
max
)–based sensitivity analysis and a size-based sensitivity analysis with subgroups defined by nodule size was performed.
Results
The AUC of the deep learning algorithm for nodule detection using OSEM reconstruction was 0.796 (CI 95%; 0.772–0.869), and 0.848 (CI 95%; 0.828–0.869) using BSREM reconstruction. The AUC was significantly higher for BSREM compared to OSEM (
p
= 0.001). On a per-slice analysis, sensitivity and specificity were 66.7% and 79.0% for OSEM, and 69.2% and 84.5% for BSREM. On a per-nodule analysis, the overall sensitivity of OSEM was 81.5% compared to 87.0% for BSREM.
Conclusions
Our results suggest that machine learning algorithms may aid detection of small
18
F-FDG-avid pulmonary nodules in clinical PET/CT. AI performed significantly better on images with BSREM than OSEM.
Key Points
• The diagnostic value of deep learning for detecting small lung nodules (≤ 2 cm) in PET images using BSREM and OSEM reconstruction was assessed.
• BSREM yields higher SUV
max
of small pulmonary nodules as compared to OSEM reconstruction.
• The use of BSREM translates into a higher detectability of small pulmonary nodules in PET images as assessed with artificial intelligence
.
Purpose Left ventricular non-compaction (LVNC) is characterized by a 2-layered myocardium composed of a noncompacted (NC) and a compacted (C) layer. The echocardiographic NC:C ratio is difficult to ...assess in many patients. The aim of the study was to assess the value of cardiac computed tomography (CCT) for the diagnosis of LVNC. Methods In this prospective controlled study, segmental analysis of transthoracic echocardiography (TTE) and prospective ECG-triggered CCT was performed in 17 patients with LVNC and 19 healthy controls. In TTE maximal NC and C thickness was measured at enddiastole and endsystole in the segment with most prominent trabeculation in short axis views. In CCT, maximal segmental NC and C thickness was measured during diastole, and NC:C ratio was determined. Spearman’s correlation coefficient and receiver operating characteristic curves were calculated. Results The median IQR radiation dose was 1.31.2–1.5mSv. The CCT thickness of the C layer was significantly lower in patients with LVNC as compared to controls in the inferolateral, midventricular, lateral-, inferior-, and septal-apical segments. The CCT NC:C ratio differed significantly between LVNC and controls in the inferior-midventricular and all the apical segments. NC:C ratio correlated significantly between TTE and CCT at enddiastole (σ = 0.8) and endsystole (σ = 0.9). Using a CCT NC:C ratio ≥1.8, all LVNC patients could be identified. Conclusion LVNC can be diagnosed with ECG-triggered low-dose CCT and discriminated from normal individuals using a NC:C ratio of ≥1.8 in diastole. There is a very good correlation of NC:C ratio in TTE and CCT.
Full text
Available for:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
Purpose
The DMI PET/CT is a modular silicon photomultiplier–based scanner with an axial field-of-view (FOV) between 15 and 25 cm depending on ring configuration (3, 4, or 5 rings). A new generation ...of the system includes a reengineered detector module, featuring improved electronics and an additional 6th ring, extending the axial FOV to 30 cm. We report on the performance evaluation of the 6-ring upgraded Generation 2 (Gen2) system while values are also reported for the 5-ring configuration of the very same system prior to the upgrade.
Methods
PET performance was evaluated using the NEMA NU 2–2018 standard for spatial resolution, sensitivity, image quality, count rate performance, timing resolution, and image co-registration accuracy. Patient images were used to assess image quality.
Results
The average system sensitivity was measured at 32.76 cps/kBq (~ 47% increase to 5 rings at 22.29 cps/kBq) while noise equivalent count rate peaked at 434.3 kcps corresponding to 23.6 kBq/mL (~ 60% increase to Generation 1 (Gen1) and 39% to Gen2 5 rings). Contrast recovery ranged between 54.5 and 85.8% similar to 5 rings, while the 6 rings provided lower background variability (2.3–8.5% for 5 rings vs 1.9–6.8% for 6 rings) and lower lung error (4.0% for the 5 rings and 3.16% for the 6 rings). Transverse/axial full width at half-maximum (FWHM) at 1 cm (3.79/4.26 mm) and 10 cm (4.29/4.55 mm), scatter fraction (40.2%), energy resolution (9.63%), and time-of-flight (TOF) resolution (389.6 ps at 0 kBq/mL) were in line to previously reported values measured across different system configurations. Improved patient image quality is obtained with the 6 rings compared to the 5 rings, while image quality is retained even at reduced scan times, enabling WB dynamic acquisitions.
Conclusions
The higher sensitivity of the 6-ring DMI compared to the 5-ring configuration may lead to improved image quality of clinical images at reduced scan time. Additionally, it could equally be used to allow improved temporal sampling and/or reduced overall scan time in dynamic acquisitions. Conversely, temporal sampling and scan time could be traded per application to further drive injected dose at lower levels.
Full text
Available for:
DOBA, EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, IZUM, KILJ, KISLJ, MFDPS, NLZOH, NUK, OILJ, PILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, SIK, UILJ, UKNU, UL, UM, UPUK, VKSCE, VSZLJ, ZAGLJ
Abstract
Aims
HIV-positive persons have increased cardiovascular event rates but data on the prevalence of subclinical atherosclerosis compared with HIV-negative persons are not uniform. We assessed ...subclinical atherosclerosis utilizing coronary artery calcium (CAC) scoring and coronary computed tomography angiography (CCTA) in 428 HIV-positive participants of the Swiss HIV Cohort Study and 276 HIV-negative controls concurrently referred for clinically indicated CCTA.
Methods and results
We assessed the association of HIV infection, cardiovascular risk profile, and HIV-related factors with subclinical atherosclerosis in univariable and multivariable analyses. HIV-positive participants (median duration of HIV infection, 15 years) were younger than HIV-negative participants (median age 52 vs. 56 years; P < 0.01) but had similar median 10-year Framingham risk scores (9.0% vs. 9.7%; P = 0.40). The prevalence of CAC score >0 (53% vs. 56.2%; P = 0.42) and median CAC scores (47 vs. 47; P = 0.80) were similar, as was the prevalence of any, non-calcified/mixed, and high-risk plaque. In multivariable adjusted analysis, HIV-positive participants had a lower prevalence of calcified plaque than HIV-negative participants 36.9% vs. 48.6%, P < 0.01; adjusted odds ratio (aOR) 0.57; 95% confidence interval (CI) 0.40–0.82; P < 0.01, lower coronary segment severity score (aOR 0.72; 95% CI 0.53–0.99; P = 0.04), and lower segment involvement score (aOR 0.71, 95% CI 0.52–0.97; P = 0.03). Advanced immunosuppression was associated with non-calcified/mixed plaque (aOR 1.97; 95% CI 1.09–3.56; P = 0.02).
Conclusion
HIV-positive persons in Switzerland had a similar degree of non-calcified/mixed plaque and high-risk plaque, and may have less calcified coronary plaque, and lower coronary atherosclerosis involvement and severity scores than HIV-negative persons with similar Framingham risk scores.
As a professional group, physicians are at increased risk of burnout and job stress, both of which are associated with an increased risk of coronary heart disease that is at least as high as that of ...other professionals. This study aimed to examine the association of burnout and job stress with coronary microvascular function, a predictor of major adverse cardiovascular events.
Thirty male physicians with clinical burnout and 30 controls without burnout were included. Burnout was assessed with the Maslach Burnout Inventory and job stress with the effort-reward imbalance and overcommitment questionnaire. All participants underwent myocardial perfusion positron emission tomography to quantify endothelium-dependent (cold pressor test) and endothelium-independent (adenosine challenge) coronary microvascular function. Burnout and job stress were regressed on coronary flow reserve (primary outcome) and two additional measures of coronary microvascular function in the same model while adjusting for age and body mass index.
Burnout and job stress were significantly and independently associated with endothelium-dependent microvascular function. Burnout was positively associated with coronary flow reserve, myocardial blood flow response, and hyperemic myocardial blood flow (r partial = 0.28 to 0.35; p-value = 0.008 to 0.035). Effort-reward ratio (r partial = - 0.32 to - 0.38; p-value = 0.004 to 0.015) and overcommitment (r partial = - 0.30 to - 0.37; p-value = 0.005 to 0.022) showed inverse associations with these measures.
In male physicians, burnout and high job stress showed opposite associations with coronary microvascular endothelial function. Longitudinal studies are needed to show potential clinical implications and temporal relationships between work-related variables and coronary microvascular function. Future studies should include burnout and job stress for a more nuanced understanding of their potential role in cardiovascular health.
Full text
Available for:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
Angiographic severity of coronary artery stenosis has historically been the primary guide to revascularization or medical management of coronary artery disease. However, physiologic severity defined ...by coronary pressure and/or flow has resurged into clinical prominence as a potential, fundamental change from anatomically to physiologically guided management. This review addresses clinical coronary physiology-pressure and flow-as clinical tools for treating patients. We clarify the basic concepts that hold true for whatever technology measures coronary physiology directly and reliably, here focusing on positron emission tomography and its interplay with intracoronary measurements.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP