Contrast-enhanced three-dimensional fast spoiled gradient-echo renal MR imaging: evaluation of vascular and nonvascular disease

Classification of myocardial fibrosis in DE-MRI based on semi-supervised semantic segmentation and dual attention mechanism

OBJECTIVE

It is essential to utilize cardiac delayed-enhanced magnetic resonance imaging (DE-MRI) to diagnose cardiovascular disease. By segmenting myocardium DE-MRI images, it provides critical information for the evaluation and treatment of myocardial infarction. As a consequence, it is vital to investigate the segmentation and classification technique of myocardial DE-MRI.

METHODS

Firstly, an end-to-end minimally supervised and semi-supervised semantic DE-MRI myocardial fibrosis segmentation framework is proposed, which combines image classification and semantic segmentation branches based on the self-attention mechanism. Following that, a residual hole network fused with the dual attention mechanism was built, and a double attention metabolic pathway classification method for cardiac fibrosis in DE-MRI images was developed.

RESULTS

By adding pixel-level labels to an extra 40 training images, the segmentation model may enhance semantic segmentation performance by 2.6 percent (from 61.2 percent to 63.8 percent). When the number of pixel-level labels is increased to 80, semi-supervised feature extraction increases by 4.7 percent when compared to weakly guided semantic segmentation. Adding an attention mechanism to the critical network DRN (Deep Residual Network) can increase the classifier's performance by a small amount. Experiments revealed that the models worked effectively.

CONCLUSION

This paper investigates the segmentation and classification of cardiac fibrosis in DE-MRI data using a semi-supervised semantic segmentation and dual attention mechanism, dealing with the issue that existing segmentation algorithms have difficulty segmenting myocardial fibrosis tissue. In the future, we can consider optimizing the design of the attention module to reduce the module computation.

Machine Learning-Based Segmentation of Left Ventricular Myocardial Fibrosis from Magnetic Resonance Imaging

PURPOSE OF REVIEW

Myocardial fibrosis (MF) arises due to myocardial infarction and numerous cardiac diseases. MF may lead to several heart disorders, such as heart failure, arrhythmias, and ischemia. Cardiac magnetic resonance (CMR) imaging techniques, such as late gadolinium enhancement (LGE) CMR, enable non-invasive assessment of MF in the left ventricle (LV). Manual assessment of MF on CMR is a tedious and time-consuming task that is subject to high observer variability. Automated segmentation and quantification of MF is important for risk stratification and treatment planning in patients with heart disorders. This article aims to review the machine learning (ML)-based methodologies developed for MF quantification in the LV using CMR images.

RECENT FINDINGS

With the availability of relatively large labeled datasets supervised learning methods based on both conventional ML and state-of-the-art deep learning (DL) methods have been successfully applied for automated segmentation of MF. The incorporation of ML algorithms into imaging techniques such as 3D LGE CMR permits fast characterization of MF on CMR imaging and may enhance the diagnosis and prognosis of patients with heart disorders. Concurrently, the studies using cine CMR images have revealed that accurate segmentation of MF on non-contrast CMR imaging might be possible. The application of ML/DL tools in CMR image interpretation is likely to result in accurate and efficient quantification of MF.

Three-dimensional Cardiac MR Imaging: Related Techniques and Clinical Applications

Three-dimensional (3D) cardiac magnetic resonance (MR) imaging has several advantages, including the easy coverage of the entire heart without misregistration, reduction of breath-holding times, and availability for postprocessing reconstruction. These advantages are associated with some techniques such as breath-hold or navigator gating and parallel imaging. However, the image quality of 3D cardiac MR images is compromised by the use of a shorter repetition time and parallel imaging. Thus, a steady-state free precession sequence, contrast agent administration, and presaturation pulses are used to maintain the image quality. In this review, we introduce the MR imaging techniques used in 3D cardiac MR imaging and demonstrate the typical 3D cardiac MR images, followed by discussion about their advantages and disadvantages.

Assessment of aortoiliac and renal arteries: MR angiography with parallel acquisition versus conventional MR angiography and digital subtraction angiography

PURPOSE

To prospectively compare the image quality, sensitivity, and specificity of three-dimensional gadolinium-enhanced magnetic resonance (MR) angiography accelerated by parallel acquisition (ie, fast MR angiography) with MR angiography not accelerated by parallel acquisition (ie, conventional MR angiography) for assessment of aortoiliac and renal arteries, with digital subtraction angiography (DSA) as the reference standard.

MATERIALS AND METHODS

The study was approved by the institutional review board; informed consent was obtained from all patients. Forty consecutive patients (33 men, seven women; mean age, 63 years) suspected of having aortoiliac and renal arterial stenoses and thus examined with DSA underwent both fast (mean imaging time, 17 seconds) and conventional (mean imaging time, 29 seconds) MR angiography. The arterial tree was divided into segments for image analysis. Two readers independently evaluated all MR angiograms for image quality, presence of arterial stenosis, and renal arterial variants. Image quality, sensitivity, and specificity were analyzed on per-patient and per-segment bases for multiple comparisons (with Bonferroni correction) and for dependencies between segments (with patient as the primary sample unit). Interobserver agreement was evaluated by using kappa statistics.

RESULTS

Overall, the image quality with fast MR angiography was significantly better (P=.001) than that with conventional MR angiography. At per-segment analysis, the image quality of fast MR angiograms of the distal renal artery tended to be better than that of conventional MR angiograms of these vessels. Differences in sensitivity for the detection of arterial stenosis between the two MR angiography techniques were not significant for either reader. Interobserver agreement in the detection of variant renal artery anatomy was excellent with both conventional and fast MR angiography (kappa=1.00).

CONCLUSION

Fast MR angiography and conventional MR angiography do not differ significantly in terms of arterial stenosis grading or renal arterial variant detection.

Imaging of kidney cancer

Advances in molecular genetics have expanded the understanding of renal cell tumors. Now it is understood that renal cortical tumors are a family of neoplasms with distinct cytogenetics and molecular defects, unique histopathologic features, and different malignant potentials. Imaging contributes to clinical management of patients with renal tumors in providing diagnostic information for tumor detection, characterization, staging, treatment planning, and follow-up.

Utilizing generalized autocalibrating partial parallel acquisition (GRAPPA) to achieve high-resolution contrast-enhanced MR angiography of hepatic artery: initial experience in orthotopic liver transplantation candidates

OBJECTIVE

To evaluate feasibility of using GRAPPA to acquire high-resolution 3D contrast-enhanced MR angiography (CE-MRA) of hepatic artery and value of GRAPPA for displaying vessels anatomy.

MATERIALS AND METHODS

High-resolution CE-MRA using GRAPPA was performed in 67 orthotopic liver transplantation recipient candidates. Signal intensity (SI) and relative SI, i.e., Cv-ro (vessel-to-liver contrast) of the aorta and the hepatic common artery (HCA), were measured. The SI and the relative SI were compared and analyzed using T-test. For purpose of qualitative evaluation, the vessel visualization quality and the order of depicted hepatic artery branches were evaluated by two radiologists independently and assessed by weighted kappa analysis. The depiction of hepatic arterial anatomy and variations was evaluated, and results were correlated with the findings in surgery.

RESULTS

The mean SI values were 283.29+/-65.07 (mean+/-S.D.) for aorta and 283.16+/-64.07 for HCA, respectively. The mean relative SI values were 0.698+/-0.09 for aorta and 0.696+/-0.09 for HCA, respectively. Homogeneous enhancement between aorta and HCA was confirmed by statistically insignificant differences (p-values were 0.89 for mean SI values and 0.12 for mean relative SI values, respectively). The average score for vessel visualization ranged from good to excellent for different artery segments. Overall interobserver agreement in the visualization of different artery segments was excellent (kappa value>0.80). The distal intrahepatic segmental arteries were well delineated for majority of patients with excellent interobserver agreement. Normal hepatic arterial anatomy was correctly demonstrated in 53 patients, and arterial anomalies were accurately detected on high-resolution MRA image of all 14 patients.

CONCLUSION

High-resolution hepatic artery MRA acquired using GRAPPA in a reproducible manner excellently depicts and delineates small vessels and can be routinely used for evaluating OLT candidates.

Magnetic resonance imaging of abdominal pain during pregnancy

The objective of this review is to demonstrate magnetic resonance imaging as an important adjunct to ultrasound and computed tomography in the evaluation of the pregnant patient with abdominal pain. With the advent of ultrafast T2-weighted pulse sequences, fetal and bowel motion cause minimal artifact on the images. An accurate diagnosis can often be made in a few minutes based on these high-contrast images performed in 2 or 3 planes. T1-weighted gradient echo images with and without fat saturation are useful for identifying blood and fat, especially in the case of adnexal masses. Gadolinium-diethylenetriamine pentaacetic acid is rarely used to diagnose inflammatory or obstructive disease and is reserved for those patients with suspected malignancies.

Sensitivity encoding (SENSE) for contrast-enhanced 3D MR angiography of the abdominal arteries

PURPOSE

To assess sensitivity encoding (SENSE) for contrast-enhanced MR angiography (CE-MRA) of the abdominal arteries in comparison with standard MRA protocols.

MATERIALS AND METHODS

In 22 patients MRA of the abdominal arteries was performed twice (once using a standard protocol, and once with the additional use of SENSE). In 10 patients all examination parameters were kept constant (TR/TE/FA = 3.8 msec/1.3 msec/30 degrees ), and a reduction in scan time from 22 to 11 seconds was realized with the use of SENSE. In 12 patients, using SENSE the acquisition matrix was increased from 208 to 416, keeping the scan time constant. Image quality was scored on a five-point scale by three radiologists. Additionally, ROI-based measurements of CNR were performed.

RESULTS

For both protocols, image quality was significantly improved using SENSE. The time-reducing SENSE protocol yielded an average score of 4.2 points vs. 3.1 for the standard protocol. Using SENSE to increase the acquisition matrix, an average score of 4.3 was reached vs. 3.2 for the standard protocol (P < 0.05). The number of depictable small vessels and their bifurcations was significantly increased by either of the two SENSE protocols as compared to the standard imaging procedure.

CONCLUSION

SENSE for MRA of the abdominal arteries significantly increases image quality and permits a substantial reduction in breath-hold time or a significantly improved spatial resolution.

Is preoperative high-resolution magnetic resonance imaging accurate in predicting neurovascular compression in patients with trigeminal neuralgia?

High-resolution magnetic resonance imaging (HR-MRI) using three-dimensional fast imaging employing steady-state acquisition (3D-FIESTA) and double-dose contrast-enhanced three-dimensional fast spoiled gradient echo (3D-FSPGR) sequences is considered to be a useful tool in detecting neurovascular compression in patients with trigeminal neuralgia. The purpose of this study was to analyze the accuracy and preoperative diagnostic value of these high-resolution imaging techniques in patients with trigeminal neuralgia, in a single-blind study. The preoperative MRI images of 21 consecutive patients were matched to one neuroradiologist, who was blind as to which side exhibited the symptoms. The images and post-processing multiplanar reconstructions were compared with the video-documented operative observations. HR-MRI using only 3D-FSPGR sequences demonstrated neurovascular compression in accordance with the intraoperative finding in 11 patients (52.4%). In the subgroup where, additionally, 3D-FIESTA sequences were available, neurovascular compression was in accordance with the intraoperative finding in 71.4% (n = 7). High-resolution magnetic resonance imaging using double-dose contrast-enhanced 3D-FSPGR and 3D-FIESTA sequences is currently not sufficient enough to make an accurate prediction of neurovascular compression in a single-blind setting. These 3D imaging techniques currently provide only limited information, and one should consider their use carefully when identifying patients with trigeminal neuralgia from operation until image quality is improved by superior image resolution that can accurately discriminate vessels surrounding the trigeminal root entry zone.

Centrally fat-saturated three-dimensional magnetic resonance angiography of the abdomen using selective central fat-saturation of k-space

PURPOSE

To assess the usefulness of centrally fat-saturated three-dimensional magnetic resonance (MR) angiography of the abdomen using an elliptical centric view order and selective placement of fat-saturation pulses in the central 30% portion of the k-space in terms of fat signal reduction, image contrast of post-contrast images, and breath-holding time.

METHODS

Fat signal in abdomen and breath-holding time were compared between centrally fat-saturated three-dimensional sequence and partially fat-saturated three-dimensional sequence or conventional fat-saturated three-dimensional sequence. Abdominal contrast-enhanced centrally fat-saturated three-dimensional MR angiography was obtained at arterial and equilibrium phases, and image quality was quantitatively and visually evaluated.

RESULTS

Centrally fat-saturated three-dimensional sequence suppressed fat signal, as did the conventional fat-saturated three-dimensional sequence, and the breath-hold was prolonged only by 1.5 seconds compared to the partially fat-saturated three-dimensional sequence. Contrast-enhanced centrally fat-saturated three-dimensional MR angiography provided abdominal MR arteriography with large signal difference between vessels and fat, while venous signal was insufficient at equilibrium phase.

CONCLUSION

Abdominal contrast-enhanced centrally fat-saturated three-dimensional MR angiography using an elliptical centric view order and selective central fat-saturation of k-space reduced fat signal comparable to conventional fat-saturated three-dimensional sequence, and provided contrast-enhanced MR arteriography with high vascular contrast and minimum prolongation of breath-hold.

MR techniques for renal imaging

This article describes the principles, attributes, and pitfalls of the many MR imaging approaches available for assessment of renal-related disorders. Tables 1 and 2 summarize the specific approach and rationale.

Fat-suppressed three-dimensional MR angiography technique with elliptical centric view order and no prolonged breath-holding time

PURPOSE

To determine the appropriate rate of fat-suppression pulses (using spec IR-spectral selective inversion recovery) for fat-suppressed 3D magnetic resonance angiography (MRA) with an elliptical centric view order.

MATERIALS AND METHODS

In abdominal 3D fast spoiled gradient echo (fast SPGR) with an elliptical centric view order, the spec IR pulse rate was changed from zero to one every 15 repetitions (in nine steps) in eight volunteers. In the equilibrium phase, abdominal contrast-enhanced 3D MRA was obtained by 3D fast SPGR using an elliptical centric view order without fat-suppression and with two spec IR, and by fat-suppressed 3D fast SPGR with a sequential-centric view order (efgre3D) in 18 cases. Fat and vascular signals were estimated.

RESULTS

Although 3D fast SPGR using an elliptical centric view order with spec IR placed every 15 TR and efgre3D effectively decreased fat signals, these sequences lengthened the breath-hold by 4-6 seconds compared with non-fat-suppressed sequence. 3D fast SPGR using an elliptical centric view order and two spec IR reduced the fat signal by 30% and provided good 3D MR angiography without substantial prolongation of breath-hold.

CONCLUSION

Two spec IR can be used for generation of partially fat-suppressed abdominal 3D MRA without prolongation of the breath-hold when performing 3D fast SPGR using an elliptical centric view order.

Body MR venography

MRV offers unique diagnostic possibilities for detection and characterization of venous disease. It allows evaluation of perivascular and vascular anatomy, evolution of thromboembolic events, and assessment of vascular flow. MRI is a diagnostic tool that can be tailored for a variety of clinical dilemmas, not only DVTs. Continued improvements in hardware and software will expand the role of MRV.

Measurement of renal volumes with contrast-enhanced MRI

PURPOSE

To determine the accuracy of in vivo magnetic resonance imaging (MRI) measurement of total renal parenchymal volume and medullary fraction.

MATERIALS AND METHODS

Sixteen kidneys in eight pigs were imaged with a multiphasic contrast-enhanced fast three-dimensional sequence on a 1.5-T imager. Kidney segmentation, followed by a process of signal intensity thresholding for cortical and nephrographic phase datasets, allowed for MRI measurements of parenchymal volume and medullary fraction. Autopsy provided reference standards of renal volume, weight, and medullary fraction.

RESULTS

An excellent correlation was found between MRI measurement of total renal parenchymal volume and autopsy volume (R2 = 0.86) and weight (R2 = 0.90). Medullary fraction (mean +/- SD) measured with MRI was 0.120 +/- 0.067, and with autopsy was 0.116 +/- 0.025 (t-test P = 0.84, F-test P = 0.001).

CONCLUSION

MRI measurements of total renal volume are accurate. MRI measurements of medullary fraction show promise, but precision is limited when using a simple signal intensity thresholding algorithm.