A joint convolutional-recurrent neural network with an attention mechanism for detecting intracranial hemorrhage on noncontrast head CT.

To investigate the performance of a joint convolutional neural networks-recurrent neural networks (CNN-RNN) using an attention mechanism in identifying and classifying intracranial hemorrhage (ICH) on a large multi-center dataset; to test its performance in a prospective independent sample consisting of consecutive real-world patients. All consecutive patients who underwent emergency non-contrast-enhanced head CT in five different centers were retrospectively gathered. Five neuroradiologists created the ground-truth labels. The development dataset was divided into the training and validation set. After the development phase, we integrated the deep learning model into an independent center's PACS environment for over six months for assessing the performance in a real clinical setting. Three radiologists created the ground-truth labels of the testing set with a majority voting. A total of 55,179 head CT scans of 48,070 patients, 28,253 men (58.77%), with a mean age of 53.84 ± 17.64 years (range 18-89) were enrolled in the study. The validation sample comprised 5211 head CT scans, with 991 being annotated as ICH-positive. The model's binary accuracy, sensitivity, and specificity on the validation set were 99.41%, 99.70%, and 98.91, respectively. During the prospective implementation, the model yielded an accuracy of 96.02% on 452 head CT scans with an average prediction time of 45 ± 8 s. The joint CNN-RNN model with an attention mechanism yielded excellent diagnostic accuracy in assessing ICH and its subtypes on a large-scale sample. The model was seamlessly integrated into the radiology workflow. Though slightly decreased performance, it provided decisions on the sample of consecutive real-world patients within a minute.

Association between left ventricular strain and cardiac iron load in beta-thalassaemia major: a cardiac magnetic resonance study.

OBJECTIVE: To evaluate the utility of cardiac magnetic resonance feature tracking-derived left ventricular strain in assessing cardiac dysfunction and investigate the correlation between left ventricular strain and myocardial T2* in patients with beta-thalassaemia major. METHODS: Forty-two patients with beta-thalassaemia major, having a mean age of 22.49 ± 8.48 years, and age-matched healthy controls were enrolled in the study. The observer drew regions of interest on the interventricular septum, and T2* decay curves were calculated accordingly. The short-axis cine images were used to derive left ventricular circumferential and radial strains, and the long-axis four-chamber and two-chamber images were used to assess left ventricular longitudinal strain. RESULTS: The mean global left ventricular strains were lower in beta-thalassaemia major patients than the controls (p < 0.05). Left ventricular strains of beta-thalassaemia major patients with cardiac T2* values of > 20 ms were also significantly reduced compared with the controls (p < 0.05); there was no difference between the mean left ventricular ejection fractions of the two groups (p = 0.84). Cardiac T2* showed a weak correlation with left ventricular ejection fraction (r = 0.33, p = 0.03), while the left ventricular circumferential strain showed a good positive correlation with cardiac T2* (r = 0.6, p < 0.0001). CONCLUSION: Compared with healthy controls, patients with beta-thalassaemia major, including those with myocardial T2* values of >20 ms, showed reduced global left ventricular strains. Left ventricular circumferential strain was positively correlated with myocardial T2*. Left ventricular strain analysis using cardiac magnetic resonance feature tracking may have utility in beta-thalassaemia major assessment.Key FindingsPatients with beta-thalassaemia major, including those with myocardial T2* values of >20 ms, had reduced global left ventricular strains.Cardiac T2* showed a weak correlation with left ventricular ejection fraction, while the left ventricular circumferential strain showed a good positive correlation with cardiac T2*.ImportanceLeft ventricular strain using cardiac magnetic resonance feature tracking might be used as an adjunct in assessing cardiac functions in beta-thalassaemia major.

The Association between the Extent of Late Gadolinium Enhancement and Diastolic Dysfunction in Hypertrophic Cardiomyopathy.

Background Diastolic dysfunction in hypertrophic cardiomyopathy (HCM) patients is a frequent, yet poorly understood phenomenon. Purpose The purpose of this study is to assess the relationship between the myocardial fibrosis and diastolic dysfunction in patients with HCM. Materials and Methods We retrospectively investigated the impact of the myocardial fibrosis, as assessed by the extent of late gadolinium enhancement (LGE-%) on cardiac magnetic resonance imaging (CMRI), on diastolic dysfunction in 110 patients with HCM. The diastolic dysfunction was evaluated by the left atrial (LA) volume index measured on CMRI and lateral septal E/E' ratio calculated on echocardiography. Results : There was a moderate correlation between the LGE-% and LA volume ( r = 0.59, p < 0.0001). The logistic regression model of LGE-%, mitral regurgitation, and total left ventricular mass that investigated the independent predictors of LA volume identified LGE-% as the only independent parameter associated with the LA volume index ( ß = 0.30, p = 0.003). No correlation was observed between the LGE-% and E/E'( r = 0.24, p = 0.009). Conclusions Myocardial fibrosis in HCM patients is associated with a chronic diastolic burden as represented by increased LA volume. However, the fibrosis does not influence the E/E' ratio, which is a well-known parameter of ventricular relaxation, restoring forces, and filling pressure.

Inter-vendor performance of deep learning in segmenting acute ischemic lesions on diffusion-weighted imaging: a multicenter study.

There is little evidence on the applicability of deep learning (DL) in the segmentation of acute ischemic lesions on diffusion-weighted imaging (DWI) between magnetic resonance imaging (MRI) scanners of different manufacturers. We retrospectively included DWI data of patients with acute ischemic lesions from six centers. Dataset A (n = 2986) and B (n = 3951) included data from Siemens and GE MRI scanners, respectively. The datasets were split into the training (80%), validation (10%), and internal test (10%) sets, and six neuroradiologists created ground-truth masks. Models A and B were the proposed neural networks trained on datasets A and B. The models subsequently fine-tuned across the datasets using their validation data. Another radiologist performed the segmentation on the test sets for comparisons. The median Dice scores of models A and B were 0.858 and 0.857 for the internal tests, which were non-inferior to the radiologist's performance, but demonstrated lower performance than the radiologist on the external tests. Fine-tuned models A and B achieved median Dice scores of 0.832 and 0.846, which were non-inferior to the radiologist's performance on the external tests. The present work shows that the inter-vendor operability of deep learning for the segmentation of ischemic lesions on DWI might be enhanced via transfer learning; thereby, their clinical applicability and generalizability could be improved.

Does Employing a Flowchart Improve the Diagnostic Performance of Cardiac Magnetic Resonance Imaging in Left Ventricular Noncompaction?

BACKGROUND: To test the hypothesis that making a diagnosis of left ventricular noncompaction (LVNC) on cardiac magnetic resonance imaging (CMRI) using a noncompacted-to-compacted (NC/C) myocardium ratio > 2.3 would yield significant errors, and also to test a diagnostic flowchart in patients who undergo CMRI and have clinical and echocardiographic findings suggesting LVNC could improve the diagnosis of LVNC. METHODS: A total of 84 patients with LVNC and 162 controls consisting of patients with other diseases and healthy participants who had CMRI and echocardiograms were selected. The diagnostic flowchart of the study involved the use of CMRI with all available sequences for patients with a high pre-test probability of LVNC. Two blinded independent cardiologists evaluated echocardiograms, and patients with suggestive echocardiographic and clinical findings for LVNC were enrolled in the high pre-test probability of LVNC group. Two independent blinded radiologists established the diagnosis of LVNC based on NC/C ratio > 2.3 on CMRI, and they were allowed to re-assess the patients following the diagnostic flowchart. RESULTS: An NC/C ratio > 2.3 identified 83 of 84 LVNC patients, yet incorrectly classified 48 of the 162 controls as having LVNC. Radiologists changed their decision in 23 of 48 patients with incorrect diagnoses, resulted in improved specificity (70.4% to 84.6%). The use of the CMRI diagnostic flowchart in the high pre-test probability group yielded a high specificity (97.2%) and accuracy (95.9%). CONCLUSIONS: LVNC diagnosed by CMRI based on the NC/C criterion can lead to overdiagnosis, whereas only using CMRI in patients with a high pre-test probability of LVNC with all available sequences may improve the diagnostic performance.

Diagnostic value of left atrial strain in pediatric hypertrophic cardiomyopathy with normal maximum left atrial volume index: preliminary cardiac magnetic resonance study.

BACKGROUND: The maximum left atrial volume index is the most widely used metric for assessing the left atrium in hypertrophic cardiomyopathy; however, it may be normal in the early phases of the disease. OBJECTIVE: To assess whether pediatric hypertrophic cardiomyopathy patients with normal maximum left atrial volume index have impaired atrial functions on cardiac magnetic resonance imaging (MRI). MATERIALS AND METHOD: A total of 26 pediatric hypertrophic cardiomyopathy patients and 24 age-matched children, as controls, were enrolled in the study. The left atrial reservoir, conduit and booster strain were calculated from two orthogonal planes and the left atrial volumes were calculated using the biplanar method. The extent of left ventricular late gadolinium enhancement (LGE-%) was calculated using the thresholding method. The left ventricular early diastolic longitudinal strain rate was calculated to assess diastolic dysfunction. RESULTS: The maximum left atrial volume index of the children with hypertrophic cardiomyopathy and the controls were not significantly different (P>0.05). Most of the left atrial functional indices were worse in children with hypertrophic cardiomyopathy (P<0.05), yet no difference was observed between the left atrial booster strains of the two groups (P>0.05). The left atrial conduit strain showed moderate to good negative correlations with left ventricular LGE-% and diastolic dysfunction. CONCLUSION: Left atrial conduit and reservoir strains are impaired in pediatric hypertrophic cardiomyopathy patients with normal maximum left atrial volumetric indices. Most left atrial strain parameters are closely linked to left ventricular LGE-% and diastolic dysfunction. Left atrial strain analysis may reveal subtle functional changes in the atrium before the increase in the maximum volume index.

The influence of cardiac motion on radiomics features: radiomics features of non-enhanced CMR cine images greatly vary through the cardiac cycle.

OBJECTIVES: The cardiac cycle might impair the reproducibility of radiomics features of cardiac magnetic resonance (CMR) cine images, yet this issue has not been addressed in the previous research. We aim to evaluate whether radiomics features of CMR cine images vary during the cardiac cycle and investigate the reproducibility of radiomics features of CMR cine images. METHODS: This retrospective study enrolled 59 healthy adults who underwent CMR examination. Two observers segmented the myocardium on a 4D stack of three consecutive mid-ventricular short-axis cine images covering the cardiac cycle. A total of 352 radiomics features were extracted. The coefficient of variation and intraclass correlation coefficient were used to assess the feature variability through the cycle and inter-observer reproducibility, respectively. RESULTS: Approximately 55% of radiomics features showed large variability through the cardiac cycle. The original features showed more variability than the Laplacian of Gaussian-filtered features (73.8% vs. 48%). The features of 4D stack cine images had a higher proportion of reproducible features (92.0%, 87.7%, and 76.1%) compared with the end-diastolic (77.8%, 62.2%, and 41.7%) and the end-systolic images (81.5%, 74.1%, and 58.8%) for intraclass correlation cut-off values of 30.80, > 0.85, and > 0.90, respectively. CONCLUSIONS: Radiomics features of CMR cine images greatly vary during the cardiac cycle. The radiomics features of 4D stack of cine images are more robust compared with end-diastolic and end-systolic cine images in terms of reproducibility. The impact of the cardiac cycle on the reproducibility of the features should be considered when employing CMR cine images radiomics. KEY POINTS: • There is limited evidence on the impact of cardiac motion on radiomics features of CMR cine images and the reproducibility of the radiomics features of CMR cine images. • Radiomics features of non-enhanced CMR cine images greatly vary during the cardiac cycle, and the number of "reproducible" features shows significant variations according to the cardiac phases. • The impact of cardiac cycle on the reproducibility of the radiomics features should be considered when employing CMR cine images radiomics.

Patients with ACL graft deficiency showed a higher frequency of knee osteoarthritis compared with patients with intact ACL graft in the medium term.

AIM: To evaluate whether patients with anterior cruciate ligament (ACL) graft tear who do not undergo revision reconstruction surgery are more likely to exhibit osteoarthritic features than patients with intact ACL graft using MRI. MATERIALS AND METHODS: We retrospectively evaluated our hospital medical archive for identifying all consecutive patients with ACL graft failure who underwent MRI between January 2015 and January 2019. An equal number of patients with intact ACL graft was selected for the control group. An observer with 6 years of musculoskeletal radiology experience evaluated the baseline and last available follow-up MRIs of the patients for bone marrow lesions (BMLs), cartilage defects, osteophytes, anterior tibial translation, and effusion-synovitis. RESULTS: A total of 130 patients, 118 males (90.8%), and 12 females (9.2%), with the mean age of 29.22 ± 9.3 years, were enrolled in the study. Of 130 patients, 65 had a deficient ACL graft, and 65 had an intact ACL graft. The time from the index injury to the last follow-up MRI was 45 months (IQR, 19) for the study cohort. On the follow-up MRIs, patients with ACL graft deficiency showed more prominent degenerative features compared with the patients with intact ACL graft characterized as follows: higher grades of medial meniscal tears (P = 0.06); higher grades of BMLs in the anteromedial (P = 0.014) and posteromedial (P = 0.006) femur, higher grades of cartilage defects in the medial patella (p = 0.0P); higher grades of osteophytes in the anteromedial (P = 0.018) and central medial femur (p = 0.048), central medial tibia (P = 0.048), and medial patella (P = 0.07); and more frequent anterior tibial translation (P = 0.022). CONCLUSION: Patients with ACL graft deficiency present with more prominent degenerative features, particularly on the medial side, suggestive of osteoarthritis compared with patients with intact ACL grafts in the medium-term follow-up.

Late gadolinium enhancement on CMRI in patients with LV noncompaction: An overestimated phenomenon?

BACKGROUND AND PURPOSE: Subendocardial fibrosis is recognized finding in left ventricular noncompaction (LVNC); however, the evidence regarding the patterns and the frequency of late gadolinium enhancement (LGE) on cardiac magnetic resonance imaging (CMRI) is controversial. The present study sought to assess the frequency and patterns of LGE in LVNC. MATERIALS AND METHODS: Patients with a diagnosis of LVNC based on the echocardiographic CMRI criteria were enrolled in this retrospective study. The myocardial noncompacted-to-compacted ratio (NC/C) was perpendicularly measured on short-axis cine images. Two observers jointly assessed the presence of LGE on short-axis LGE images. The long-axis four-chamber and long-axis two-chamber images were used to confirm the presence of LGE if needed. RESULTS: A total of 42 patients, 20 females (47.7%) and 22 were males (52.3%), were included in the study. The median age of the patients was 32.4 years (range 18-63). LGE was identified in 2 out of 42 patients (4.7%) with LVNC. LGE was identified in the interventricular septum involving the subendocardial layer and noncompacted lateral myocardial wall involving the trabeculae at mid-ventricular and basal levels. CONCLUSION: LGE is uncommon in patients with LVNC. We highlight that the diagnosis of LVNC in patients with atypical LGE patterns, such as epicardial or transmural enhancement, should be reappraised and the other cardiac diseases should be discarded before establishing the final diagnosis.

Diagnostic values of edema-sensitive T2-weighted imaging, TSE T1-weighted early contrast-enhanced imaging, late gadolinium enhancement, and the Lake Louise criteria in assessing acute myocarditis: A single-center cardiac magnetic resonance study.

OBJECTIVE: The aim of this study was to evaluate the diagnostic accuracy of the Lake Louise consensus criteria using cardiac magnetic resonance (CMR) imaging assessment of edema, hyperemia, and late gadolinium enhancement (LGE) in the diagnostic determination of acute myocarditis. METHODS: A total of 44 patients with acute myocarditis and 24 healthy controls were included in this retrospective study. The presence of edema was defined as a myocardial mean signal intensity >1.9 times that of the skeletal muscle in the same slice on T2-weighted short tau inversion-recovery sequences. Hyperemia was defined as an early gadolinium enhancement ratio (EGEr) ≥4 calculated using the contrast-enhancement of the myocardium and skeletal muscle on TSE T1-weighted sequences, and LGE was assessed by visual examination. The reference methods used to determine the presence of myocarditis were based on the European Society of Cardiology Working Group on Myocardial and Pericardial Diseases guidelines for clinical and biochemical findings. RESULTS: The diagnostic accuracy of edema, hyperemia, LGE, and the Lake Louise criteria (at least 2 of 3 components) was 75.7%, 64.2%, 88.5%, and 84.2%, respectively. Among the 3 components of the Lake Louise criteria, edema had the highest specificity (100%), and LGE had the highest sensitivity (86%). The use of LGE and/or edema as a criterion for acute myocarditis revealed the highest diagnostic accuracy (92.8%) among the CMR sequences and combinations of components examined. CONCLUSION: LGE and/or edema as a sole criterion for the diagnosis of acute myocarditis demonstrated better diagnostic accuracy than the Lake Louise criteria. The use of EGEr did not improve the performance of CMR-based diagnosis. Alternatives to the use of EGEr are recommended for the diagnosis of acute myocarditis.