Inversion Recovery-prepared Gradient Echo Research Articles

Collaborative Learning for Annotation-Efficient Volumetric MR Image Segmentation.

Deep learning has presented great potential in accurate MR image segmentation when enough labeled data are provided for network optimization. However, manually annotating three-dimensional (3D) MR images is tedious and time-consuming, requiring experts with rich domain knowledge and experience. To build a deep learning method exploring sparse annotations, namely only a single two-dimensional slice label for each 3D training MR image. Retrospective. Three-dimensional MR images of 150 subjects from two publicly available datasets were included. Among them, 50 (1377 image slices) are for prostate segmentation. The other 100 (8800 image slices) are for left atrium segmentation. Five-fold cross-validation experiments were carried out utilizing the first dataset. For the second dataset, 80 subjects were used for training and 20 were used for testing. 1.5 T and 3.0 T; axial T2-weighted and late gadolinium-enhanced, 3D respiratory navigated, inversion recovery prepared gradient echo pulse sequence. A collaborative learning method by integrating the strengths of semi-supervised and self-supervised learning schemes was developed. The method was trained using labeled central slices and unlabeled noncentral slices. Segmentation performance on testing set was reported quantitatively and qualitatively. Quantitative evaluation metrics including boundary intersection-over-union (B-IoU), Dice similarity coefficient, average symmetric surface distance, and relative absolute volume difference were calculated. Paired t test was performed, and P < 0.05 was considered statistically significant. Compared to fully supervised training with only the labeled central slice, mean teacher, uncertainty-aware mean teacher, deep co-training, interpolation consistency training (ICT), and ambiguity-consensus mean teacher, the proposed method achieved a substantial improvement in segmentation accuracy, increasing the mean B-IoU significantly by more than 10.0% for prostate segmentation (proposed method B-IoU: 70.3% ± 7.6% vs. ICT B-IoU: 60.3% ± 11.2%) and by more than 6.0% for left atrium segmentation (proposed method B-IoU: 66.1% ± 6.8% vs. ICT B-IoU: 60.1% ± 7.1%). A collaborative learning method trained using sparse annotations can segment prostate and left atrium with high accuracy. 0 TECHNICAL EFFICACY: Stage 1.

Improved high-resolution pediatric vascular cardiovascular magnetic resonance with gadofosveset-enhanced 3D respiratory navigated, inversion recovery prepared gradient echo readout imaging compared to 3D balanced steady-state free precession readout imaging

BackgroundImproved delineation of vascular structures is a common indication for cardiovascular magnetic resonance (CMR) in children and requires high spatial resolution. Currently, pre-contrast 3D, respiratory navigated, T2-prepared, fat saturated imaging with a bSSFP readout (3D bSSFP) is commonly used; however, these images can be limited by blood pool inhomogeneity and exaggeration of metal artifact. We compared image quality of pediatric vasculature obtained using standard 3D bSSFP to 3D, respiratory navigated, inversion recovery prepared imaging with a gradient echo readout (3D IR GRE) performed after administration of gadofosveset trisodium (GT), a blood pool contrast agent.MethodsFor both sequences, VCG triggering was used with acquisition during a quiescent period of the cardiac cycle. 3D bSSFP imaging was performed pre-contrast, and 3D IR GRE imaging was performed 5 min after GT administration. We devised a vascular imaging quality score (VIQS) with subscores for coronary arteries, pulmonary arteries and veins, blood pool homogeneity, and metal artifact. Scoring was performed on axial reconstructions of isotropic datasets by two independent readers and differences were adjudicated. Signal- and contrast-to-noise (SNR and CNR) calculations were performed on each dataset.ResultsThirty-five patients had both 3D bSSFP and 3D IR GRE imaging performed. 3D IR GRE imaging showed improved overall vascular imaging compared to 3D bSSFP when comparing all-patient VIQS scores (n = 35, median 14 (IQR 11–15), vs 6 (4–10), p < 0.0001), and when analyzing the subset of patients with intrathoracic metal (n = 17, 16 (14–17) vs. 5 (2–9), p < 0.0001). 3D IR GRE showed significantly improved VIQS subscores for imaging the RCA, pulmonary arteries, pulmonary veins, and blood pool homogeneity. In addition, 3D IR GRE imaging showed reduced variability in both all-patient and metal VIQS scores compared to 3D bSSFP (p < 0.05). SNR and CNR were higher with 3D IR GRE in the left ventricle and left atrium, but not the pulmonary arteries.ConclusionsRespiratory navigated 3D IR GRE imaging after GT administration provides improved vascular CMR in pediatric patients compared to pre-contrast 3D bSSFP imaging, as well as improved imaging in patients with intrathoracic metal. It is an excellent alternative in this challenging patient population when high spatial resolution vascular imaging is needed.

Role of inflammation in the pathogenesis of hypertrophic cardiomyopathy: a T2-mapping CMR study

Background Hypertrophic cardiomyopathy (HCM) is associated with increased myocardial fibrosis and collagen deposition. Collagen reduces cardiovascular magnetic resonance (CMR) derived T2 times, however, the inflammatory milieu required for its deposition is often associated with localised oedema. Focal areas of high T2 signal intensity have been reported in association with areas of late gadolinium enhancement (LGE) in patients with HCM. However, this previous work has relied upon the subjective visual assessment of local changes in signal intensity. We hypothesised that quantitative CMR T2mapping would reveal marked differences in myocardial T2 characteristics in patients with HCM relative to controls. Methods Twenty-two consecutive patients with HCM (15 male, age 55.8±8.7 years) but no co-morbidity and 22 healthy controls (9 male, age 49.3±13.1 years) were studied. For each subject, a short axis mid-ventricular slice was acquired using a 1.5T Avanto (Siemens Healthcare, Erlangen, Germany). A breath-hold T2-prepared singleshot steady-state free precession (SSFP) sequence was used to produce three T2-weighted images with different T2 preparation times (0 ms, 24 ms, 55 ms). Parallel imaging was used to reduce the imaging time and motioncorrection was applied to improve the accuracy of T2 measurement. T2 maps were produced by fitting the signal intensity-time exponential decay curve using a linear 2-parameter model after logarithmic transformation (CMRtools, Cardiovascular Imaging Solutions, London, UK). An inversion recovery prepared gradient echo sequence with the inversion time optimised to null normal myocardium was used 10 minutes after intravenous gadolinium contrast (Gadovist, Bayer-Schering, Berlin, Germany, 0.1 mmol/kg) to detect late enhancement. Imaging was repeated in an orthogonal phase encoding direction to exclude artifact. Results There were no significant differences in baseline demographic characteristics between HCM and control patients. LV-wall thickness, indexed LV mass and ejection fraction were significantly higher in HCM patients (Table 1). In keeping with LV hypertrophy, the LV volumes were significantly smaller. None of the healthy controls exhibited any late enhancement whereas 18 (82%) HCM patients had areas of mid-wall enhancement, predominantly in the areas of hypertrophy. T2 times in patients with HCM were similar to controls with no evidence of any significant differences (HCM 54.1±2.5ms versus Controls 54.1±1.6ms, P=0.97). Conclusions

Assessment of Myocardial Viability Using Delayed Enhancement Magnetic Resonance Imaging at 3.0 Tesla

Cardiac magnetic resonance imaging (MRI) at 3.0 T has recently become available and potentially provides a significant improvement of tissue contrast in T1-weighted imaging techniques relying on Gd-based contrast enhancement. Imaging at high-field strength may be especially advantageous for methods relying on strong T1-weighting and imaging after contrast material administration. The aim of this study was to compare cardiac delayed enhancement (DE) MRI at 3.0 T and 1.5 T with respect to image quality, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) between infarcted and normal myocardium. Forty consecutive patients with history of myocardial infarction were examined at 3.0 T (n = 20) or at 1.5 T (n = 20). Myocardial function was assessed using cine steady-state-free-precession (SSFP) sequences (TR 3.1 milliseconds, TE 1.6 milliseconds, flip angle 70 degrees , and a matrix of 168 x 256 at 1.5 T and TR 3.4 milliseconds, TE 1.7 milliseconds, flip angle 50 degrees and a matrix of 168 x 256 at 3.0 T), acquired in long- and short-axes views. DE images were obtained 15 minutes after the administration of 0.15 mmol of Gd-DTPA/kg body weight using a segmented inversion recovery prepared gradient echo sequence at 1.5 T (TR 9.6 milliseconds, TE 4.4 milliseconds, flip angle 25 degrees , matrix 160 x 256, bandwidth 140 Hertz/pixel) and at 3.0 T (TR 9.8 milliseconds, TE 4.3 milliseconds, flip angle 30 degrees , matrix 150 x 256, bandwidth 140 Hertz/pixel). For image analysis, standardized SNR and CNR measurements were performed in infarcted and remote myocardial regions. Two independent observers rated image quality on a 4-point scale (0 = poor image quality, 1 = sufficient image quality, 2 = good image quality, 3 = excellent image quality). High diagnostic image quality was obtained in all patients. Rating of mean image quality was 2.2 +/- 0.8 at 1.5 T and 2.5 +/- 0.6 at 3.0 T (P = 0.012) for observer 1 and 2.2 +/- 0.7 at 1.5 T and 2.6 +/- 0.6 at 3.0 T (P = 0.003) for observer 2, respectively. Interobserver agreement was good (kappa = 0.68 at 1.5 T and 0.78 at 3.0 T). SNR measurements yielded a mean SNR of 37.8 +/- 13.9/22.9 +/- 6.0 in infarcted myocardium (P < 0.001) and 5.6 +/- 2.2/5.9 +/- 2.4 in normal myocardium (P = 0.45) at 3.0 T/1.5 T, respectively. CNR measurements revealed mean values of 32.4 +/- 13.0/16.7 +/- 5.4 (P< 0.001) at 3.0 T/1.5 T, respectively. Delayed enhancement MRI at 3.0 T is feasible and provides superior image quality compared with 1.5 T. Furthermore, using identical contrast doses, increased SNR and CNR values were recorded at 3.0 T.