Potential of Photon-Counting Detector CT for Radiation Dose Reduction for the Assessment of Interstitial Lung Disease in Patients With Systemic Sclerosis.

OBJECTIVE: The aim of this study was to determine the potential of photon-counting detector computed tomography (PCD-CT) for radiation dose reduction compared with conventional energy-integrated detector CT (EID-CT) in the assessment of interstitial lung disease (ILD) in systemic sclerosis (SSc) patients. METHODS: In this retrospective study, SSc patients receiving a follow-up noncontrast chest examination on a PCD-CT were included between May 2021 and December 2021. Baseline scans were generated on a dual-source EID-CT by selecting the tube current-time product for each of the 2 x-ray tubes to obtain a 100% (D 100 ), a 66% (D 66 ), and a 33% dose image (D 33 ) from the same data set. Slice thickness and kernel were adjusted between the 2 scans. Image noise was assessed by placing a fixed region of interest in the subcutaneous fat. Two independent readers rated subjective image quality (5-point Likert scale), presence, extent, diagnostic confidence, and accuracy of SSc-ILD. D 100 interpreted by a radiologist with 22 years of experience served as reference standard. Interobserver agreement was calculated with Cohen κ, and mean variables were compared by a paired t test. RESULTS: Eighty patients (mean 56 ± 14; 64 women) were included. Although CTDI vol of PCD-CT was comparable to D 33 (0.72 vs 0.76 mGy, P = 0.091), mean image noise of PCD-CT was comparable to D 100 (131 ± 15 vs 113 ± 12, P > 0.05). Overall subjective image quality of PCD-CT was comparable to D 100 (4.72 vs 4.71; P = 0.874). Diagnostic accuracy was higher in PCD-CT compared with D 33 /D 66 (97.6% and 92.5%/96.3%, respectively) and comparable to D 100 (98.1%). CONCLUSIONS: With PCD-CT, a radiation dose reduction of 66% compared with EID-CT is feasible, without penalty in image quality and diagnostic performance for the evaluation of ILD.

Computed tomography radiomics for the prediction of thymic epithelial tumor histology, TNM stage and myasthenia gravis.

OBJECTIVES: To evaluate CT-derived radiomics for machine learning-based classification of thymic epithelial tumor (TET) stage (TNM classification), histology (WHO classification) and the presence of myasthenia gravis (MG). METHODS: Patients with histologically confirmed TET in the years 2000-2018 were retrospectively included, excluding patients with incompatible imaging or other tumors. CT scans were reformatted uniformly, gray values were normalized and discretized. Tumors were segmented manually; 15 scans were re-segmented after 2 weeks by two readers. 1316 radiomic features were calculated (pyRadiomics). Features with low intra-/inter-reader agreement (ICC<0.75) were excluded. Repeated nested cross-validation was used for feature selection (Boruta algorithm), model training, and evaluation (out-of-fold predictions). Shapley additive explanation (SHAP) values were calculated to assess feature importance. RESULTS: 105 patients undergoing surgery for TET were identified. After applying exclusion criteria, 62 patients (28 female; mean age, 57±14 years; range, 22-82 years) with 34 low-risk TET (LRT; WHO types A/AB/B1), 28 high-risk TET (HRT; WHO B2/B3/C) in early stage (49, TNM stage I-II) or advanced stage (13, TNM III-IV) were included. 14(23%) of the patients had MG. 334(25%) features were excluded after intra-/inter-reader analysis. Discriminatory performance of the random forest classifiers was good for histology(AUC, 87.6%; 95% confidence interval, 76.3-94.3) and TNM stage(AUC, 83.8%; 95%CI, 66.9-93.4) but poor for the prediction of MG (AUC, 63.9%; 95%CI, 44.8-79.5). CONCLUSIONS: CT-derived radiomic features may be a useful imaging biomarker for TET histology and TNM stage.

Can magnetic resonance imaging radiomics of the pancreas predict postoperative pancreatic fistula?

OBJECTIVES: To evaluate whether a magnetic resonance imaging (MRI) radiomics-based machine learning classifier can predict postoperative pancreatic fistula (POPF) after pancreaticoduodenectomy (PD) and to compare its performance to T1 signal intensity ratio (T1 SIratio). METHODS: Sixty-two patients who underwent 3 T MRI before PD between 2008 and 2018 were retrospectively analyzed. POPF was graded and split into clinically relevant POPF (CR-POPF) vs. biochemical leak or no POPF. On T1- and T2-weighted images, 2 regions of interest were placed in the pancreatic corpus and cauda. 173 radiomics features were extracted using pyRadiomics. Additionally, the pancreas-to-muscle T1 SIratio was measured. The dataset was augmented and split into training (70 %) and test sets (30 %). A Boruta algorithm was used for feature reduction. For prediction of CR-POPF models were built using a gradient-boosted tree (GBT) and logistic regression from the radiomics features, T1 SIratio and a combination of the two. Diagnostic accuracy of the models was compared using areas under the receiver operating characteristics curve (AUCs). RESULTS: Five most important radiomics features were identified for prediction of CR-POPF. A GBT using these features achieved an AUC of 0.82 (95 % Confidence Interval [CI]: 0.74 - 0.89) when applied on the original (non-augmented) dataset. Using T1 SIratio, a GBT model resulted in an AUC of 0.75 (CI: 0.63 - 0.84) and a logistic regression model delivered an AUC of 0.75 (CI: 0.63 - 0.84). A GBT model combining radiomics features and T1 SIratio resulted in an AUC of 0.90 (CI 0.84 - 0.95). CONCLUSION: MRI-radiomics with routine sequences provides promising prediction of CR-POPF.

Prognostic value of texture analysis from cardiac magnetic resonance imaging in patients with Takotsubo syndrome: a machine learning based proof-of-principle approach.

Cardiac magnetic resonance (CMR) imaging has become an important technique for non-invasive diagnosis of takotsubo syndrome (TTS). The long-term prognostic value of CMR imaging in TTS has not been fully elucidated yet. This study sought to evaluate the prognostic value of texture analysis (TA) based on CMR images in patients with TTS using machine learning. In this multicenter study (InterTAK Registry), we investigated CMR imaging data of 58 patients (56 women, mean age 68 ± 12 years) with TTS. CMR imaging was performed in the acute to subacute phase (median time after symptom onset 4 days) of TTS. TA of the left ventricle was performed using free-hand regions-of-interest in short axis late gadolinium-enhanced and on T2-weighted (T2w) images. A total of 608 TA features adding the parameters age, gender, and body mass index were included. Dimension reduction was performed removing TA features with poor intra-class correlation coefficients (ICC ≤ 0.6) and those being redundant (correlation matrix with Pearson correlation coefficient r > 0.8). Five common machine-learning classifiers (artificial neural network Multilayer Perceptron, decision tree J48, NaïveBayes, RandomForest, and Sequential Minimal Optimization) with tenfold cross-validation were applied to assess 5-year outcome including major adverse cardiac and cerebrovascular events (MACCE). Dimension reduction yielded 10 TA features carrying prognostic information, which were all based on T2w images. The NaïveBayes machine learning classifier showed overall best performance with a sensitivity of 82.9% (confidence interval (CI) 80-86.2), specificity of 83.7% (CI 75.7-92), and an area-under-the receiver operating characteristics curve of 0.88 (CI 0.83-0.92). This proof-of-principle study is the first to identify unique T2w-derived TA features that predict long-term outcome in patients with TTS. These features might serve as imaging prognostic biomarkers in TTS patients.

Artificial Intelligence and Texture Analysis in Cardiac Imaging.

PURPOSE OF REVIEW: The aim of this structured review is to summarize the current research applications and opportunities arising from artificial intelligence (AI) and texture analysis with regard to cardiac imaging. RECENT FINDINGS: Current research findings suggest tremendous potential for AI in cardiac imaging, especially with regard to objective image analyses, overcoming the limitations of an observer-dependent subjective image interpretation. Researchers have used this technique across multiple imaging modalities, for instance to detect myocardial scars in cardiac MR imaging, to predict contrast enhancement in non-contrast studies, and to improve image acquisition and reconstruction. AI in medical imaging has the potential to provide novel, much-needed applications for improving patient care pertaining to the cardiovascular system. While several shortcomings are still present in the current methodology, AI may serve as a resourceful assistant to radiologists and clinicians alike.

Machine learning-based CT fractional flow reserve assessment in acute chest pain: first experience.

BACKGROUND: Computed tomography (CT)-derived fractional flow reserve (FFRCT) enables the non-invasive functional assessment of coronary artery stenosis. We evaluated the feasibility and potential clinical role of FFRCT in patients presenting to the emergency department with acute chest pain who underwent chest-pain CT (CPCT). METHODS: For this retrospective IRB-approved study, we included 56 patients (median age: 62 years, 14 females) with acute chest pain who underwent CPCT and who had at least a mild (≥25% diameter) coronary artery stenosis. CPCT was evaluated for the presence of acute plaque rupture and vulnerable plaque features. FFRCT measurements were performed using a machine learning-based software. We assessed the agreement between the results from FFRCT and patient outcome (including results from invasive catheter angiography and from any non-invasive cardiac imaging test, final clinical diagnosis and revascularization) for a follow-up of 3 months. RESULTS: FFRCT was technically feasible in 38/56 patients (68%). Eleven of the 38 patients (29%) showed acute plaque rupture in CPCT; all of them underwent immediate coronary revascularization. Of the remaining 27 patients (71%), 16 patients showed vulnerable plaque features (59%), of whom 11 (69%) were diagnosed with acute coronary syndrome (ACS) and 10 (63%) underwent coronary revascularization. In patients with vulnerable plaque features in CPCT, FFRCT had an agreement with outcome in 12/16 patients (75%). In patients without vulnerable plaque features (n=11), one patient showed myocardial ischemia (9%). In these patients, FFRCT and patient outcome showed an agreement in 10/11 patients (91%). CONCLUSIONS: Our preliminary data show that FFRCT is feasible in patients with acute chest pain who undergo CPCT provided that image quality is sufficient. FFRCT has the potential to improve patient triage by reducing further downstream testing but appears of limited value in patients with CT signs of acute plaque rupture.

Novel Magnetic Resonance Late Gadolinium Enhancement With Fixed Short Inversion Time in Ischemic Myocardial Scars.

AIMS: Late gadolinium enhancement (LGE) visualizes scar tissue after myocardial infarction. However, in clinically used LGE sequences, subendocardial infarcts can be missed due to low contrast between blood pool and subendocardium. The purpose of his study was to compare scar visibility in a novel 3-dimensional (3D) single breath-hold inversion recovery sequence with fixed, short inversion time (TI = 100 milliseconds) (short LGE) and standard 3D LGE imaging with individually adjusted TI (LGE). METHODS: Short LGE and LGE (both sequences with the same settings: spatial resolution, 1.2 × 1.2 mm; slice thickness, 8 mm; field of view, 350 × 350 mm; single breath-hold) were acquired in 64 patients with previous MI (13 female; mean age, 57 ± 19 years) at 1.5 T. Inversion time was set to 100 milliseconds in short LGE and adjusted individually in LGE according to the Look-Locker sequence. Two independent readers evaluated 1088 segments (17-segment model), identified infarcted segments, and categorized scar visibility (5 = excellent, 1 = poor scar visibility) and scar transmurality (4 = transmural, 0 = no scar) using a 5-point Likert scale. Signal intensity ratios between short LGE and LGE for scar and blood pool, for scar and remote myocardium, and for remote myocardium and blood pool were calculated. RESULTS: Short LGE showed 197 infarcted segments out of 1088 (18.1%); LGE revealed 191 segments (17.6%). Short LGE with dark scar and bright blood pool demonstrated better overall scar visibility, especially in subendocardially infarcted segments compared with LGE (4.2 vs 3.0, 5 = excellent visibility; P = 0.01). Signal intensity ratios for short LGE relative to LGE were 1.42 for scar/blood pool, 0.8 for scar/remote myocardium, and 0.22 for remote myocardium/blood.Overall transmurality was not rated higher in short LGE compared with LGE (P = 0.8). More fibrous tissue and total fibrous percentage (P = 0.04) were measured in short LGE compared with LGE, whereas myocardial mass was not significantly different (P = 0.5). Acquisition time was similar between short LGE and LGE (26 ± 4 seconds vs 25 ± 9 seconds, P = 0.7). CONCLUSIONS: Short LGE is a fast, single breath-hold 3D LGE sequence with no need for myocardial nulling due to fixed inversion time with improved scar visibility, especially in subendocardial infarcts.

Cardiovascular magnetic resonance T2* mapping for the assessment of cardiovascular events in hypertrophic cardiomyopathy.

Background: Hypertrophic cardiomyopathy (HCM) is associated with an increased risk of adverse cardiac events. Beyond classic risk factors, relative myocardial ischaemia and succeeding myocardial alterations, which can be detected using either contrast agents or parametric mapping in cardiovascular magnetic resonance (CMR) imaging, have shown an impact on outcome in HCM. CMR may help to risk stratify using parametric T2* mapping. Therefore, the aim of the present study was to evaluate the association of T2* values or fibrosis with cardiovascular events in HCM. Methods: The relationship between T2* with supraventricular, ventricular arrhythmia or heart failure was retrospectively assessed in 91 patients with HCM referred for CMR on a 1.5T MR imaging system. Fibrosis as a reference was added to the model. Patients were subdivided into groups according to T2* value quartiles. Results: 47 patients experienced an event of ventricular arrhythmia, 25 of atrial fibrillation/flutter and 17 of heart failure. T2*≤28.7 ms yielded no association with ventricular events in the whole HCM cohort. T2* of non-obstructive HCM showed a significant association with ventricular events in univariate analysis, but not in multivariate analysis. For the combined endpoint of arrhythmic events, there was already an association for the whole HCM cohort, but again only in univariate analyses. Fibrosis stayed the strongest predictor in all analyses. There was no association for T2* and fibrosis with heart failure. Conclusions: Decreased T2* values by CMR only provide a small association with arrhythmic events in HCM, especially in non-obstructive HCM. No information is added for heart failure.

Multimodal Multiparametric Three-dimensional Image Fusion in Coronary Artery Disease: Combining the Best of Two Worlds.

PURPOSE: To allow for comprehensive noninvasive diagnostics of coronary artery disease (CAD) by using three-dimensional (3D) image fusion of CT coronary angiography, CT-derived fractional flow reserve (CT FFR), whole-heart dynamic 3D cardiac MRI perfusion, and 3D cardiac MRI late gadolinium enhancement (LGE). MATERIALS AND METHODS: Seventeen patients (54 years ± 10 [standard deviation], one female) who underwent cardiac CT and cardiac MRI were included (combined subcohort of three prospective trials). Software facilitating multimodal 3D image fusion was developed. Postprocessing of CT data included segmentation of the coronary tree and heart contours, calculation of CT FFR values, and color coding of the coronary tree according to CT FFR. Postprocessing of cardiac MRI data included segmentation of the left ventricle (LV) in cardiac MRI perfusion and cardiac MRI LGE, co-registration of cardiac MRI to CT data, and projection of cardiac MRI perfusion and LGE values onto the high spatial resolution LV from CT. RESULTS: Image quality was rated as good to excellent (scores: 2.5-2.6; 3 = excellent). CT coronary angiography revealed significant stenoses in seven of 17 cases (41%). CT FFR was possible in 16 of 17 cases (94%) and showed pathologic flow in seven of 17 cases (41%), six of which coincided with cases revealing significant stenoses at CT coronary angiography. Cardiac MRI perfusion identified eight of 17 patients (47%) with hypoperfusion (ischemic burden of 17% ± 5). Cardiac MRI LGE showed myocardial scar in three of 17 cases (18%, scar burden of 7% ± 4). Conventional two-dimensional readout of CT coronary angiography and cardiac MRI resulted in eight of 17 cases (47%) with uncertain findings. Most of these divergent findings could be solved when adding information from CT FFR and 3D image fusion (six of eight, 75%). CONCLUSION: Multimodal 3D cardiac image fusion is feasible and may help with comprehensive noninvasive CAD diagnostics.Supplemental material is available for this article.© RSNA, 2020.

The potential of machine learning to predict postoperative pancreatic fistula based on preoperative, non-contrast-enhanced CT: A proof-of-principle study.

BACKGROUND: Postoperative pancreatic fistula remains an unsolved challenge after pancreatoduodenectomy. Important in this regard is the presence of a soft pancreatic texture which is a major risk factor. Advances in machine learning and texture analysis of medical images allow identification of features of parenchyma that are invisible to the human eye. The aim of this study was to investigate the potential of machine learning to predict postoperative pancreatic fistula based on preoperative, non-contrast-enhanced computed tomography. METHODS: We screened a prospectively assessed database including all patients undergoing pancreatoduodenectomy at a tertiary center from 2008 until 2018 for patients based on the occurrence of postoperative pancreatic fistula. In total, 110 patients were included, consisting of 55 patients who developed a postoperative pancreatic fistula and 55 without postoperative pancreatic fistula. For machine learning-based texture analysis preoperative, non-contrast-enhanced computed tomography axial images were used. Machine learning results were tested using 10-fold cross validation. Previously validated clinical fistula risk scores (original and alternative fistula risk scores) served as reference tests. RESULTS: Both the original and the alternative fistula risk scores showed good discrimination between patients without and with postoperative pancreatic fistula (area under the curve 0.76 and 0.72, respectively). Machine learning-based texture analysis showed potential to detect histologic fibrosis (area under the curve 0.84, sensitivity 75%; specificity 92%), histologic lipomatosis (area under the curve 0.82, sensitivity 78%; specificity 89%), and intraoperative pancreatic hardness (area under the curve 0.70, sensitivity 78%; specificity 74%). The features of the machine learning-based texture analysis were most accurate in predicting the occurrence of postoperative pancreatic fistula (area under the curve 0.95, sensitivity of 96%; specificity 98%) after pancreatoduodenectomy. CONCLUSION: This proof-of-principle study suggests the ability of machine learning in recognizing important features of pancreatic texture associated with an increased risk of postoperative pancreatic fistula based on preoperative computed tomography.

Medication-Related Osteonecrosis of the Jaw-Comparison of Bone Imaging Using Ultrashort Echo-Time Magnetic Resonance Imaging and Cone-Beam Computed Tomography.

OBJECTIVE: The aim of this study was to compare bone imaging between ultrashort echo-time (UTE) magnetic resonance (MR) imaging and cone-beam computed tomography (CBCT) as the reference standard in patients with medication-related osteonecrosis of the jaw (MRONJ). MATERIALS AND METHODS: A 1-year retrospective, blinded, and randomized qualitative analysis of UTE MR images and CBCT from 19 patients with clinically diagnosed MRONJ was performed by 2 independent radiologists. Medication-related osteonecrosis of the jaw imaging hallmarks such as osteolysis, periosteal thickening, and medullary osteosclerosis were rated visually (0 and 1 to 3 for normal and mild to severe changes) for defined anatomic regions of the jaw. In addition, segmentation of these regions was performed on coregistered MR/CBCT images for the following quantitative comparison of signal intensity (SI) on MR and gray values (GVs) on CBCT images. Interreader/modality agreement (Cohen kappa), standard testing for significant differences of (non)parametric values, and Pearson correlation of signal intensity/GV were used for statistical analysis. RESULTS: The anterior corpus of the mandible was most often affected by MRONJ (P < 0.001). Overall, interreader agreement of qualitative MRONJ hallmark scores was almost perfect (κ = 0.81) and without significant differences between modalities (κ = 0.81 vs 0.82, CBCT vs MR, respectively). Intermodality agreement for qualitative gradings was substantial for both readers (κ = 0.77 and 0.70). Signal intensity/GV in MRONJ-affected areas differed significantly from healthy bone (P < 0.001) as well as correlation significantly between modalities (r = -0.77; P < 0.001). CONCLUSIONS: Qualitative assessment of MRONJ with radiation-free UTE MR imaging is comparable to reference standard CBCT. Quantitative measurements of both modalities significantly distinguish diseased from normal bone with strong correlations among the quantitative values in both modalities.

Cardiac magnetic resonance imaging to detect ischemia in chronic coronary syndromes: state of the art.

The new 2019 European Society of Cardiology guidelines for the diagnosis and management of chronic coronary syndromes emphasize the role of noninvasive functional imaging of myocardial ischemia in diagnosing coronary artery disease to guide decision making regarding revascularization. Cardiac magnetic resonance imaging (CMR) stands out relative to other imaging modalities given its high safety profile, absence of ionizing radiation, and its versatility in encoding various image contrasts. It also allows an assessment of myocardial function, ischemia, and viability as well as permits tissue characterization including detection of edema in a single examination. In recent years, a number of meta­analyses and studies considering the role of CMR for detecting ischemia have been published. The recent multicenter randomized MR­INFORM trial has demonstrated the clinical utility of CMR in patients with stable angina and cardiovascular risk factors. This landmark study has proved that a perfusion CMR­based strategy leads to a lower number of revascularizations while being noninferior to an invasive coronary angiography with fractional flow reserve-guided therapy in terms of major adverse cardiac events at 1 year. In light of recent and future technical improvements, CMR will become increasingly important in the assessment of myocardial ischemia in patients with chronic coronary syndromes.

Right coronary artery motion analysis: a novel method to measure right ventricular systolic function by selective coronary angiography.

Right ventricular systolic dysfunction is prognostic in various cardiovascular diseases. Right ventricular systolic function is not commonly assessed in the catheterization laboratory. Therefore, we developed a novel, reproducible method to measure right ventricular systolic function during selective coronary angiography. We analyzed the angiographic systolic translational motion and maximum speed of the right coronary artery (RCA) in 97 consecutive patients and compared it to the tricuspid annular plane systolic excursion (TAPSE) as measured by echocardiography. All measurements were performed by two independent operators on two occasions. Inter-observer variability and intra-observer variability were excellent for RCA motion distance and for RCA maximum speed. There was a significant correlation of the RCA motion distance and RCA maximum speed with the TAPSE measured by echocardiography (Pearson's correlation for RCA distance: r = 0.59, p < 0.001, r2 = 0.35; for RCA speed: r = 0.40, p < 0.001, r2 = 0.16). The area under the receiver operating curve for the RCA motion distance was 0.88 (95% CI 0.80-0.96) for discrimination of normal and abnormal right ventricular systolic function. A cut-off value less than 22.3 mm systolic RCA motion had a specificity of 93.3% and a sensitivity of 75.6% for identifying an abnormal right ventricular systolic function. Analysis of the RCA motion is a reproducible and reliable method to measure right ventricular systolic function during selective coronary angiography. It is a simple and useful tool to assess right ventricular function in the catheterization laboratory and may serve for risk assessment for right ventricular failure. CLINICAL TRIAL REGISTRATION: Data for this study was collected retrospectively from Swiss Transcatheter Aortic Valve Implantation Registry (NCT01368250). https://clinicaltrials.gov/show/NCT01368250 .

Texture analysis of myocardial infarction in CT: Comparison with visual analysis and impact of iterative reconstruction.

OBJECTIVES: To compare texture analysis (TA) with subjective visual diagnosis of myocardial infarction (MI) in cardiac computed tomography (CT) and to evaluate the impact of iterative reconstruction (IR). METHODS: Ten patients (4 women, mean age 68 ± 11 years) with confirmed chronic MI and 20 controls (8 women, mean age 52 ± 11 years) with no cardiac abnormality underwent contrast-enhanced cardiac CT with the same protocol. Images were reconstructed with filtered back projection (FBP) and with advanced modeled IR at strength levels 3-5. Subjective diagnosis of MI was made by three independent, blinded readers with different experience levels. Classification of MI was performed using machine learning-based decision tree models for the entire data set and after splitting into training and test data to avoid overfitting. RESULTS: Subjective visual analysis for diagnosis of MI showed excellent intrareader (kappa: 0.93) but poor interreader agreement (kappa: 0.3), with variable performance at different image reconstructions. TA showed high performance for all image reconstructions (correct classifications: 94%-97%, areas under the curve: 0.94-0.99). After splitting into training and test data, overall lower performances were observed, with best results for IR at level 5 (correct classifications: 73%, area under the curve: 0.65). CONCLUSIONS: As compared with subjective, nonreliable visual analysis of inexperienced readers, TA enables objective and reproducible diagnosis of chronic MI in cardiac CT with higher accuracy. IR has a considerable impact on both subjective and objective image analysis.

Cardiovascular magnetic resonance T2* mapping for structural alterations in hypertrophic cardiomyopathy.

PURPOSE: Hypertrophic cardiomyopathy (HCM) is characterized by a heterogeneous morphology and variable prognosis. A mismatch between left ventricular mass (LVM) and microvascular circulation with corresponding relative ischemia has been implicated to cause myocardial replacement fibrosis that deteriorates prognosis. Besides parametric T1 mapping, Cardiovascular Magnetic Resonance (CMR) T2* mapping is able to identify ischemia as well as fibrosis in cardiac and extracardiac diseases. Therefore, we aimed to investigate the value of T2* mapping to characterize structural alterations in patients with HCM. METHODS: CMR was performed on a 1.5 T MR imaging system (Achieva, Philips, Best, Netherlands) using a 5-channel coil in patients with HCM (n = 103, 50.6 ± 16.4 years) and in age- and gender-matched controls (n = 20, 44.8 ± 16.9 years). T2* mapping (1 midventricular short axis slice) was acquired in addition to late gadolinium enhancement (LGE). T2* values were compared between patients with HCM and controls as well as between HCM patients with- and without fibrosis. RESULTS: HCM patients showed significantly decreased T2* values compared to controls (26.2 ± 4.6 vs. 31.3 ± 4.3 ms, p < 0.001). Especially patients with myocardial fibrosis presented with decreased T2* values in comparison to those without fibrosis (25.2 ± 4.0 vs. 28.7 ± 5.3 ms, p = 0.003). A regression model including maximum wall thickness, LVM and T2* values provided good overall diagnostic accuracy of 80% to diagnose HCM with and without fibrosis. CONCLUSION: In this study, parametric mapping identified lower T2* values in HCM patients compared to controls, especially in a sub-group of patients with myocardial fibrosis. As myocardial fibrosis has been suggested to influence prognosis of patients with HCM, T2* mapping may add information for identifying a higher risk sub-group of HCM patients.

Radiomics for Distinguishing Myocardial Infarction from Myocarditis at Late Gadolinium Enhancement at MRI: Comparison with Subjective Visual Analysis.

PURPOSE: To evaluate whether radiomics features of late gadolinium enhancement (LGE) regions at cardiac MRI enable distinction between myocardial infarction (MI) and myocarditis and to compare radiomics with subjective visual analyses by readers with different experience levels. MATERIALS AND METHODS: In this retrospective, institutional review board-approved study, consecutive MRI examinations of 111 patients with MI and 62 patients with myocarditis showing LGE were included. By using open-source software, classification performances attained from two-dimensional (2D) and three-dimensional (3D) texture analysis, shape, and first-order descriptors were compared, applying five different machine learning algorithms. A nested, stratified 10-fold cross-validation was performed. Classification performances were compared through Wilcoxon signed-rank tests. Supervised and unsupervised feature selection techniques were tested; the effect of resampling MR images was analyzed. Subjective image analysis was performed on 2D and 3D image sets by two independent, blinded readers with different experience levels. RESULTS: When trained with recursive feature elimination (RFE), a support vector machine achieved the best results (accuracy: 88%) for 2D features, whereas linear discriminant analysis (LDA) showed the highest accuracy (85%) for 3D features (P <.05). When trained with principal component analysis (PCA), LDA attained the highest accuracy with both 2D (86%) and 3D (89%; P =.4) features. Results found for classifiers trained with spline resampling were less accurate than those achieved with one-dimensional (1D) nearest-neighbor interpolation (P <.05), whereas results for classifiers trained with 1D nearest-neighbor interpolation and without resampling were similar (P =.1). As compared with the radiomics approach, subjective visual analysis performance was lower for the less experienced and higher for the experienced reader for both 2D and 3D data. CONCLUSION: Radiomics features of LGE permit the distinction between MI and myocarditis with high accuracy by using either 2D features and RFE or 3D features and PCA.© RSNA, 2019Supplemental material is available for this article.

Photon Counting Computed Tomography With Dedicated Sharp Convolution Kernels: Tapping the Potential of a New Technology for Stent Imaging.

OBJECTIVES: The aims of this study were to assess the value of a dedicated sharp convolution kernel for photon counting detector (PCD) computed tomography (CT) for coronary stent imaging and to evaluate to which extent iterative reconstructions can compensate for potential increases in image noise. MATERIALS AND METHODS: For this in vitro study, a phantom simulating coronary artery stenting was prepared. Eighteen different coronary stents were expanded in plastic tubes of 3 mm diameter. Tubes were filled with diluted contrast agent, sealed, and immersed in oil calibrated to an attenuation of -100 HU simulating epicardial fat. The phantom was scanned in a modified second generation 128-slice dual-source CT scanner (SOMATOM Definition Flash, Siemens Healthcare, Erlangen, Germany) equipped with both a conventional energy integrating detector and PCD. Image data were acquired using the PCD part of the scanner with 48 × 0.25 mm slices, a tube voltage of 100 kVp, and tube current-time product of 100 mAs. Images were reconstructed using a conventional convolution kernel for stent imaging with filtered back-projection (B46) and with sinogram-affirmed iterative reconstruction (SAFIRE) at level 3 (I463). For comparison, a dedicated sharp convolution kernel with filtered back-projection (D70) and SAFIRE level 3 (Q703) and level 5 (Q705) was used. The D70 and Q70 kernels were specifically designed for coronary stent imaging with PCD CT by optimizing the image modulation transfer function and the separation of contrast edges. Two independent, blinded readers evaluated subjective image quality (Likert scale 0-3, where 3 = excellent), in-stent diameter difference, in-stent attenuation difference, mathematically defined image sharpness, and noise of each reconstruction. Interreader reliability was calculated using Goodman and Kruskal's γ and intraclass correlation coefficients (ICCs). Differences in image quality were evaluated using a Wilcoxon signed-rank test. Differences in in-stent diameter difference, in-stent attenuation difference, image sharpness, and image noise were tested using a paired-sample t test corrected for multiple comparisons. RESULTS: Interreader and intrareader reliability were excellent (γ = 0.953, ICCs = 0.891-0.999, and γ = 0.996, ICCs = 0.918-0.999, respectively). Reconstructions using the dedicated sharp convolution kernel yielded significantly better results regarding image quality (B46: 0.4 ± 0.5 vs D70: 2.9 ± 0.3; P < 0.001), in-stent diameter difference (1.5 ± 0.3 vs 1.0 ± 0.3 mm; P < 0.001), and image sharpness (728 ± 246 vs 2069 ± 411 CT numbers/voxel; P < 0.001). Regarding in-stent attenuation difference, no significant difference was observed between the 2 kernels (151 ± 76 vs 158 ± 92 CT numbers; P = 0.627). Noise was significantly higher in all sharp convolution kernel images but was reduced by 41% and 59% by applying SAFIRE levels 3 and 5, respectively (B46: 16 ± 1, D70: 111 ± 3, Q703: 65 ± 2, Q705: 46 ± 2 CT numbers; P < 0.001 for all comparisons). CONCLUSIONS: A dedicated sharp convolution kernel for PCD CT imaging of coronary stents yields superior qualitative and quantitative image characteristics compared with conventional reconstruction kernels. Resulting higher noise levels in sharp kernel PCD imaging can be partially compensated with iterative image reconstruction techniques.

3D image fusion of whole-heart dynamic cardiac MR perfusion and late gadolinium enhancement: Intuitive delineation of myocardial hypoperfusion and scar.

BACKGROUND: Since patients with myocardial hypoperfusion due to coronary artery disease (CAD) with preserved viability are known to benefit from revascularization, accurate differentiation of hypoperfusion from scar is desirable. PURPOSE: To develop a framework for 3D fusion of whole-heart dynamic cardiac MR perfusion and late gadolinium enhancement (LGE) to delineate stress-induced myocardial hypoperfusion and scar. STUDY TYPE: Prospective feasibility study. SUBJECTS: Sixteen patients (61 ± 14 years, two females) with known/suspected CAD. FIELD STRENGTH/SEQUENCE: 1.5T (nine patients); 3.0T (seven patients); whole-heart dynamic 3D cardiac MR perfusion (3D-PERF, under adenosine stress); 3D LGE inversion recovery sequences (3D-SCAR). ASSESSMENT: A software framework was developed for 3D fusion of 3D-PERF and 3D-SCAR. Computation steps included: 1) segmentation of the left ventricle in 3D-PERF and 3D-SCAR; 2) semiautomatic thresholding of perfusion/scar data; 3) automatic calculation of ischemic/scar burden (ie, pathologic relative to total myocardium); 4) projection of perfusion/scar values onto artificial template of the left ventricle; 5) semiautomatic coregistration to an exemplary heart contour easing 3D orientation; and 6) 3D rendering of the combined datasets using automatically defined color tables. All tasks were performed by two independent, blinded readers (J.S. and R.M.). STATISTICAL TESTS: Intraclass correlation coefficients (ICC) for determining interreader agreement. RESULTS: Image acquisition, postprocessing, and 3D fusion were feasible in all cases. In all, 10/16 patients showed stress-induced hypoperfusion in 3D-PERF; 8/16 patients showed LGE in 3D-SCAR. For 3D-PERF, semiautomatic thresholding was possible in all patients. For 3D-SCAR, automatic thresholding was feasible where applicable. Average ischemic burden was 11 ± 7% (J.S.) and 12 ± 7% (R.M.). Average scar burden was 8 ± 5% (J.S.) and 7 ± 4% (R.M.). Interreader agreement was excellent (ICC for 3D-PERF = 0.993, for 3D-SCAR = 0.99). DATA CONCLUSION: 3D fusion of 3D-PERF and 3D-SCAR facilitates intuitive delineation of stress-induced myocardial hypoperfusion and scar. LEVEL OF EVIDENCE: 2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2018;48:1129-1138.