Automated observer-independent acquisition of cardiac short-axis MR images: a pilot study

Present and Future Innovations in AI and Cardiac MRI

Cardiac MRI is used to diagnose and treat patients with a multitude of cardiovascular diseases. Despite the growth of clinical cardiac MRI, complicated image prescriptions and long acquisition protocols limit the specialty and restrain its impact on the practice of medicine. Artificial intelligence (AI)-the ability to mimic human intelligence in learning and performing tasks-will impact nearly all aspects of MRI. Deep learning (DL) primarily uses an artificial neural network to learn a specific task from example data sets. Self-driving scanners are increasingly available, where AI automatically controls cardiac image prescriptions. These scanners offer faster image collection with higher spatial and temporal resolution, eliminating the need for cardiac triggering or breath holding. In the future, fully automated inline image analysis will most likely provide all contour drawings and initial measurements to the reader. Advanced analysis using radiomic or DL features may provide new insights and information not typically extracted in the current analysis workflow. AI may further help integrate these features with clinical, genetic, wearable-device, and "omics" data to improve patient outcomes. This article presents an overview of AI and its application in cardiac MRI, including in image acquisition, reconstruction, and processing, and opportunities for more personalized cardiovascular care through extraction of novel imaging markers.

Implementation and prospective clinical validation of AI-based planning and shimming techniques in cardiac MRI

PURPOSE

Cardiovascular magnetic resonance (CMR) is a vital diagnostic tool in the management of cardiovascular diseases. The advent of advanced CMR technologies combined with artificial intelligence (AI) has the potential to simplify imaging, reduce image acquisition time without compromising image quality (IQ), and improve magnetic field uniformity. Here, we aim to implement two AI-based deep learning techniques for automatic slice alignment and cardiac shimming and evaluate their performance in clinical cardiac magnetic resonance imaging (MRI).

METHODS

Two deep neural networks were developed, trained, and validated on pre-acquired cardiac MRI datasets (>500 subjects) to achieve automatic slice planning and shimming (implemented in the scanner) for CMR. To examine the performance of our automated cardiac planning (EasyScan) and AI-based shim (AI shim), two prospective studies were performed subsequently. For the EasyScan validation, 10 healthy subjects underwent two identical CMR protocols: with manual cardiac planning and with AI-based EasyScan to assess protocol scan time difference and accuracy of cardiac plane prescriptions on a 1.5 T clinical MRI scanner. For the AI shim validation, a total of 20 subjects were recruited: 10 healthy and 10 cardio-oncology patients with referrals for a CMR examination. Cine images were obtained with standard cardiac volume shim and with AI shim to assess signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), overall IQ (sharpness and MR image degradation), ejection fraction (EF), and absolute wall thickening. A hybrid statistical method using of nonparametric (Wilcoxon) and parametric (t-test) assessments was employed for statistical analyses.

RESULTS

CMR protocol with AI-based plane prescriptions, EasyScan, minimized operator dependence and reduced overall scanning time by over 2 min (∼13 % faster, p < 0.001) compared to the protocol with manual cardiac planning. EasyScan plane prescriptions also demonstrated more accurate (less plane angulation errors from planes manually prescribed by a certified cardiac MRI technologist) cardiac planes than previously reported strategies. Additionally, AI shim resulted in improved B0 field homogeneity. Cine images obtained with AI shim revealed a significantly higher SNR (12.49%; p = 0.002) than those obtained with volume shim (volume shim: 32.90 ± 7.42 vs. AI shim: 37.01 ± 8.87) for the left ventricle (LV) myocardium. LV myocardium CNR was 12.48% higher for cine imaging with AI shim (149.02 ± 39.15) than volume shim (132.49 ± 33.94). Images obtained with AI shim resulted in sharper images than those obtained with volume shim (p = 0.012). The LVEF and absolute wall thickening also showed that differences exist between the two shimming methods. The LVEF by AI shim was shown to be slightly larger than LVEF by volume shim in two groups: 2.87% higher with AI shim for the healthy group and 1.70% higher with AI shim for the patient group. The LV absolute wall thickening (in mm) also showed that differences exist between shimming methods for each group with larger changes observed in the patient group (healthy: 3.31%, p = 0.234 and patient group: 7.29%, p = 0.059).

CONCLUSIONS

CMR exams using EasyScan for cardiac planning demonstrated accelerated cardiac exam compared to the CMR protocol with manual cardiac planning. Improved and more uniform B0 magnetic field homogeneity also achieved using AI shim technique compared to volume shimming.

A novel deep-learning-based approach for automatic reorientation of 3D cardiac SPECT images

PURPOSE

Reconstructed transaxial cardiac SPECT images need to be reoriented into standard short-axis slices for subsequent accurate processing and analysis. We proposed a novel deep-learning-based method for fully automatic reorientation of cardiac SPECT images and evaluated its performance on data from two clinical centers.

METHODS

We used a convolutional neural network to predict the 6 rigid-body transformation parameters and a spatial transformation network was then implemented to apply these parameters on the input images for image reorientation. A novel compound loss function which balanced the parametric similarity and penalized discrepancy of the prediction and training dataset was utilized in the training stage. Data from a set of 322 patients underwent data augmentation to 6440 groups of images for the network training, and a dataset of 52 patients from the same center and 23 patients from another center were used for evaluation. Similarity of the 6 parameters was analyzed between the proposed and the manual methods. Polar maps were generated from the output images and the averaged count values of the 17 segments were computed from polar maps to evaluate the quantitative accuracy of the proposed method.

RESULTS

All the testing patients achieved automatic reorientation successfully. Linear regression results showed the 6 predicted rigid parameters and the average count value of the 17 segments having good agreement with the reference manual method. No significant difference by paired t-test was noticed between the rigid parameters of our method and the manual method (p > 0.05). Average count values of the 17 segments show a smaller difference of the proposed and manual methods than those between the existing and manual methods.

CONCLUSION

The results strongly indicate the feasibility of our method in accurate automatic cardiac SPECT reorientation. This deep-learning-based reorientation method has great promise for clinical application and warrants further investigation.

Closed-loop control of k-space sampling via physiologic feedback for cine MRI

BACKGROUND

Segmented cine cardiac MRI combines data from multiple heartbeats to achieve high spatiotemporal resolution cardiac images, yet predefined k-space segmentation trajectories can lead to suboptimal k-space sampling. In this work, we developed and evaluated an autonomous and closed-loop control system for radial k-space sampling (ARKS) to increase sampling uniformity.

METHODS

The closed-loop system autonomously selects radial k-space sampling trajectory during live segmented cine MRI and attempts to optimize angular sampling uniformity by selecting views in regions of k-space that were not previously well-sampled. Sampling uniformity and the ability to detect cardiac phase in vivo was assessed using ECG data acquired from 10 normal subjects in an MRI scanner. The approach was then implemented with a fast gradient echo sequence on a whole-body clinical MRI scanner and imaging was performed in 4 healthy volunteers. The closed-loop k-space trajectory was compared to random, uniformly distributed and golden angle view trajectories via measurement of k-space uniformity and the point spread function. Lastly, an arrhythmic dataset was used to evaluate a potential application of the approach.

RESULTS

The autonomous trajectory increased k-space sampling uniformity by 15±7%, main lobe point spread function (PSF) signal intensity by 6±4%, and reduced ringing relative to golden angle sampling. When implemented, the autonomous pulse sequence prescribed radial view angles faster than the scan TR (0.98 ± 0.01 ms, maximum = 1.38 ms) and increased k-space sampling mean uniformity by 10±11%, decreased uniformity variability by 44±12%, and increased PSF signal ratio by 6±6% relative to golden angle sampling.

CONCLUSION

The closed-loop approach enables near-uniform radial sampling in a segmented acquisition approach which was higher than predetermined golden-angle radial sampling. This can be utilized to increase the sampling or decrease the temporal footprint of an acquisition and the closed-loop framework has the potential to be applied to patients with complex heart rhythms.

Deep Learning-based Prescription of Cardiac MRI Planes

PURPOSE

To develop and evaluate a system to prescribe imaging planes for cardiac MRI based on deep learning (DL)-based localization of key anatomic landmarks.

MATERIALS AND METHODS

Annotated landmarks on 892 long-axis (LAX) and 493 short-axis (SAX) cine steady-state free precession series from cardiac MR images were retrospectively collected between February 2012 and June 2017. U-Net-based heatmap regression was used for localization of cardiac landmarks, which were used to compute cardiac MRI planes. Performance was evaluated by comparing localization distances and plane angle differences between DL predictions and ground truth. The plane angulations from DL were compared with those prescribed by the technologist at the original time of acquisition. Data were split into 80% for training and 20% for testing, and results confirmed with fivefold cross-validation.

RESULTS

On LAX images, DL localized the apex within mean 12.56 mm ± 19.11 (standard deviation) and the mitral valve (MV) within 7.68 mm ± 6.91. On SAX images, DL localized the aortic valve within 5.78 mm ± 5.68, MV within 5.90 mm ± 5.24, pulmonary valve within 6.55 mm ± 6.39, and tricuspid valve within 6.39 mm ± 5.89. On the basis of these localizations, average angle bias and mean error of DL-predicted imaging planes relative to ground truth annotations were as follows: SAX, -1.27° ± 6.81 and 4.93° ± 4.86; four chambers, 0.38° ± 6.45 and 5.16° ± 3.80; three chambers, 0.13° ± 12.70 and 9.02° ± 8.83; and two chamber, 0.25° ± 9.08 and 6.53° ± 6.28, respectively.

CONCLUSION

DL-based anatomic localization is a feasible strategy for planning cardiac MRI planes. This approach can produce imaging planes comparable to those defined by ground truth landmarks.© RSNA, 2019 Supplemental material is available for this article.

Bias associated with left ventricular quantification by multimodality imaging: a systematic review and meta-analysis

Purpose

Cardiac MR (CMR) is the gold standard for left ventricular (LV) quantification. However, two-dimensional echocardiography (2DE) is the most common approach, and both three-dimensional echocardiography (3DE) and multidetector CT (MDCT) are increasingly available. The clinical significance and interchangeability of these modalities remains under-investigated. Therefore, we undertook a systemic review to evaluate the accuracy and absolute bias in LV quantification of all the commonly available non-invasive imaging modalities (2DE, CE-2DE, 3DE, MDCT) compared to cardiac MR (CMR).

Methods

Studies were included that reported LV echocardiographic (2DE, CE-2DE, 3DE) and/or MDCT measurements compared to CMR. Only modern CMR (SSFP sequences) was considered. Studies involving small sample size (<10 patients) and unusual cardiac geometry (ie, congenital heart diseases) were excluded. We evaluated LV end-diastolic volume (LVEDV), end-systolic volume (LVESV) and ejection fraction (LVEF).

Results

1604 articles were initially considered: 65 studies were included (total of 4032 scans (echo, CT, MRI) performed in 2888 patients). Compared to CMR, significant biased underestimation of LV volumes with 2DE was seen (LVEDV—33.30 mL, LVESV −16.20 mL, p<0.0001). This difference was reduced but remained significant with CE-2DE (LVEDV −18.05, p<0.0001) and 3DE (LVEDV −14.41, p<0.001), while MDCT values were similar to CMR (LVEDV −1.20, p=0.43; LVESV −0.13, p=0.91). However, excellent agreement for echocardiographic LVEF evaluation (2DE LVEF 0.78–1.01%, p=0.37) was observed, especially with 3DE (LVEF 0.14%, p=0.88).

Conclusions

Comparing imaging modalities to CMR as reference standard, 3DE had the highest accuracy in LVEF estimation: 2DE and 3DE-derived LV volumes were significantly underestimated. Newer generation CT showed excellent accuracy for LV volumes.

Clinical application of an automatic slice-alignment method for cardiac MR imaging

PURPOSE

We evaluated the usefulness of an automatic slice-alignment method to simplify planning of cardiac magnetic resonance (MR) scans with a 3-tesla scanner.

METHODS

We obtained 2-dimensional (2D) axial multislice images using steady-state free precession (SSFP) sequences covering the whole heart at the end-diastole phase with electrocardiography (ECG) gating in 38 patients. We detected several anatomical feature points of the heart and calculated all planes required for cardiac imaging based on those points. We visually evaluated the acceptability of an acquired imaging plane and measured the angular differences of each view between the results obtained by this method and by a conventional manual pointing approach.

RESULTS

The average visual scores were 3.4 ± 1.0 for short-axis images, 3.2 ± 0.9 for 4-chamber images, 3.2 ± 0.8 for 2-chamber images, and 3.3 ± 0.8 for 3-chamber images; average angular differences were 5.8 ± 5.1 (short axis), 7.7 ± 5.7 (4-chamber), 11.5 ± 6.7 (2-chamber), and 9.1 ± 4.6 degrees (3-chamber). Processing time was within 1.8 s in all subjects.

CONCLUSION

The proposed method can provide planes within the clinically acceptable range and within a short time in cardiac imaging of patients with various cardiac shapes and diseases without the need for high level operator proficiency in performing the examination and interpreting results.

Automatic slice alignment method for cardiac magnetic resonance imaging

OBJECTIVES

Automatic slice alignment is important for easier operation and shorter examination times in cardiac magnetic resonance imaging (MRI) examinations. We propose a new automatic slice alignment method for six cardiac planes (short-axis, vertical long-axis, horizontal long-axis, 4-chamber, 2-chamber, and 3-chamber views).

MATERIALS AND METHODS

ECG-gated 2D steady-state free precession axial multislice images were acquired using a 1.5-T MRI scanner during a single breath-hold. The scanning time was set to <20 s in 23 volumes from 23 healthy volunteers. In this method, the positions of the mitral valve, cardiac apex, left ventricular outflow tract, tricuspid valve, anterior wall of the heart, and right ventricular corner are detected to determine the positions of six reference planes by combining knowledge-based recognition and image processing techniques. In order to evaluate the results of automatic slice alignment for the short-axis, 4-chamber, 2-chamber, and 3-chamber views, the angular and positional errors between the results obtained by our proposed method and by manual annotation were measured.

RESULTS

The average angular errors for the short-axis, 4-chamber, 2-chamber, and 3-chamber views were 3.05°, 4.52°, 7.28°, and 5.79°, respectively. The average positional errors for the short-axis (base), short-axis (apex), 4-chamber, 2-chamber, and 3-chamber views were 6.61°, 3.80°, 1.55°, 1.52°, and 1.48°, respectively.

CONCLUSION

The experimental results showed that our proposed method can detect the cardiac planes quickly and accurately. Our method is therefore beneficial to both patients and operators.

Fully automatic geometry planning for cardiac MR imaging and reproducibility of functional cardiac parameters

PURPOSE

To establish operator-independent, fully automated planning of standard cardiac geometries and to determine the impact on interstudy reproducibility of cardiac functional parameters.

MATERIALS AND METHODS

Cardiac MR imaging was done in 50 patients referred for left-ventricular function assessment. In all patients, first standard manual planning was performed followed by automatic planning (AUTO1) and repeat automatic planning (AUTO2) after repositioning the patient to investigate interstudy reproducibility. Cardiac functional parameters were assessed and cine scans were visually graded on a 4-point scale from nondiagnostic to excellent.

RESULTS

Overall success rate of AUTO was 94% with good to excellent geometry planning in >94% of cine standard views. Comparing manual versus fully automated planning, a high agreement of cardiac functional parameters (Lin's concordance correlation coefficient, 0.91 to 0.99) with minimal percent bias (0.24 to 3.84%) was found. In addition, a high interstudy reproducibility of automatic planning was demonstrated (Lin's concordance correlation coefficient, 0.89 to 0.99; percent bias, 0.38 to 5.04%; precision, 3.46 to 9.09%).

CONCLUSION

Fully automated planning of cardiac geometries could reliably be performed in patients showing a variety of cardiovascular pathologies. Standard cardiac geometries were precisely replicated and functional parameters were highly accurate.

Automatic view planning for cardiac MRI acquisition

Conventional cardiac MRI acquisition involves a multi-step approach, requiring a few double-oblique localizers in order to locate the heart and prescribe long- and short-axis views of the heart. This approach is operator-dependent and time-consuming. We propose a new approach to automating and accelerating the acquisition process to improve the clinical workflow. We capture a highly accelerated static 3D full-chest volume through parallel imaging within one breath-hold. The left ventricle is localized and segmented, including left ventricle outflow tract. A number of cardiac landmarks are then detected to anchor the cardiac chambers and calculate standard 2-, 3-, and 4-chamber long-axis views along with a short-axis stack. Learning-based algorithms are applied to anatomy segmentation and anchor detection. The proposed algorithm is evaluated on 173 localizer acquisitions. The entire view planning is fully automatic and takes less than 10 seconds in our experiments.

Clinical evaluation of automated scan prescription of knee MR images

PURPOSE

To compare an automated scan planning method to manual scan positioning in routine knee magnetic resonance imaging (MRI) studies.

MATERIALS AND METHODS

The automated scan planning method uses anatomical landmarks in a 3D survey of the knee. The method is trained by example plannings, consisting of manual slice positioning by an experienced technologist in 15 MRI studies. Automated knee MR examinations obtained in three geometries in 50 consecutive patients were compared to those obtained in 50 consecutive control patients, where imaging planes were planned manually. Anatomical coverage and slice angulation were scored for each geometry on a 4-grade scale by an experienced radiologist blinded to the way of planning; groups were compared using a Mann-Whitney U-test.

RESULTS

In 150 automated sequences the technologist adapted slice positioning in four cases (addition of slices to adapt to the size of the knee), representing the only automated sequences that received a poor rating. Thirteen sequences with manual planning received a poor rating. No difference in quality was found (P > 0.05) between automated and manual plannings for coronal coverage, sagittal coverage and angulation, and transverse angulation. Rating of automated planning was higher for transverse coverage, but lower than manual planning for coronal angulation.

CONCLUSION

Automated sequence prescription for knee MRI is feasible in clinical practice, with similar quality as manual positioning.

MR image analysis: Longitudinal cardiac motion influences left ventricular measurements

BACKGROUND

Software for the analysis of left ventricular (LV) volumes and mass using border detection in short-axis images only, is hampered by through-plane cardiac motion. Therefore we aimed to evaluate software that involves longitudinal cardiac motion.

METHODS

Twenty-three consecutive patients underwent 1.5-Tesla cine magnetic resonance (MR) imaging of the entire heart in the long-axis and short-axis orientation with breath-hold steady-state free precession imaging. Offline analysis was performed using software that uses short-axis images (Medis MASS) and software that includes two-chamber and four-chamber images to involve longitudinal LV expansion and shortening (CAAS-MRV). Intraobserver and interobserver reproducibility was assessed by using Bland-Altman analysis.

RESULTS

Compared with MASS software, CAAS-MRV resulted in significantly smaller end-diastolic (156+/-48ml versus 167+/-52ml, p=0.001) and end-systolic LV volumes (79+/-48ml versus 94+/-52ml, p<0.001). In addition, CAAS-MRV resulted in higher LV ejection fraction (52+/-14% versus 46+/-13%, p<0.001) and calculated LV mass (154+/-52g versus 142+/-52g, p=0.004). Intraobserver and interobserver limits of agreement were similar for both methods.

CONCLUSION

MR analysis of LV volumes and mass involving long-axis LV motion is a highly reproducible method, resulting in smaller LV volumes, higher ejection fraction and calculated LV mass.

Sparse registration for three-dimensional stress echocardiography

Three-dimensional (3-D) stress echocardiography is a novel technique for diagnosing cardiac dysfunction. It involves evaluating wall motion of the left ventricle, by visually analyzing ultrasound images obtained in rest and in different stages of stress. Since the acquisitions are performed minutes apart, variabilities may exist in the visualized cross-sections. To improve anatomical correspondence between rest and stress, aligning the images is essential. We developed a new intensity-based, sparse registration method to retrieve standard anatomical views from 3-D stress images that were equivalent to the manually selected views in the rest images. Using sparse image planes, the influence of common image artifacts could be reduced. We investigated different similarity measures and different levels of sparsity. The registration was tested using data of 20 patients and quantitatively evaluated based on manually defined anatomical landmarks. Alignment was best using sparse registration with two long-axis and two short-axis views; registration errors were reduced significantly, to the range of interobserver variabilities. In 91% of the cases, the registration result was qualitatively assessed as better than or equal to the manual alignment. In conclusion, sparse registration improves the alignment of rest and stress images, with a performance similar to manual alignment. This is an important step towards objective quantification in 3-D stress echocardiography.

Avaliação da função cardíaca por ressonância magnética com seqüências em equilíbrio estável: segmentadas × tempo real

OBJETIVO: Comparar os índices de função sistólica ventricular obtidos entre as seqüências de cine-ressonância magnética em equilíbrio estável, em tempo real e acoplada ao eletrocardiograma, em pacientes com ritmo regular ou não. MATERIAIS E MÉTODOS: Foram comparados a fração de ejeção e os volumes diastólico e sistólico finais, em 31 pacientes, 11 com ritmo cardíaco irregular e 20 com ritmo cardíaco sinusal regular, utilizando-se seqüências segmentadas acopladas ao eletrocardiograma e em tempo real. O tratamento estatístico foi feito através da correlação de Pearson e a concordância de Bland-Altman, com p < 0,01. RESULTADOS: As aquisições em tempo real demonstraram borramento dos contornos endocárdicos, mas ambas as seqüências tiveram forte correlação positiva entre os valores obtidos: fração de ejeção, r = 0,94; volume diastólico final, r = 0,93; volume sistólico final, r = 0,98. A análise dos 11 pacientes com ritmo irregular não demonstrou diferença estatisticamente significativa, apesar da menor relação de contraste sangue-miocárdio. CONCLUSÃO: Seqüências em tempo real podem ser utilizadas para a análise da função cardíaca, independente do ritmo cardíaco dos pacientes.

Left-ventricle myocardium segmentation using a coupled level-set with a priori knowledge

This paper presents a coupled level-set segmentation of the myocardium of the left ventricle of the heart using a priori information. From a fast marching initialisation, two fronts representing the endocardium and epicardium boundaries of the left ventricle are evolved as the zero level-set of a higher dimension function. We introduce a novel and robust stopping term using both gradient and region-based information. The segmentation is supervised both with a coupling function and using a probabilistic model built from training instances. The robustness of the segmentation scheme is evaluated by performing a segmentation on four unseen data-sets containing high variation and the performance of the segmentation is quantitatively assessed.

SPASM: A 3D-ASM for segmentation of sparse and arbitrarily oriented cardiac MRI data

A new technique (SPASM) based on a 3D-ASM is presented for automatic segmentation of cardiac MRI image data sets consisting of multiple planes with arbitrary orientations, and with large undersampled regions. Model landmark positions are updated in a two-stage iterative process. First, landmark positions close to intersections with images are updated. Second, the update information is propagated to the regions without image information, such that new locations for the whole set of the model landmarks are obtained. Feature point detection is performed by a fuzzy inference system, based on fuzzy C-means clustering. Model parameters were optimized on a computer cluster and the computational load distributed by grid computing. SPASM was applied to image data sets with an increasing sparsity (from 2 to 11 slices) comprising images with different orientations and stemming from different MRI acquisition protocols. Segmentation outcomes and calculated volumes were compared to manual segmentation on a dense short-axis data configuration in a 3D manner. For all data configurations, (sub-)pixel accuracy was achieved. Performance differences between data configurations were significantly different (p<0.05) for SA data sets with less than 6 slices, but not clinically relevant (volume differences<4 ml). Comparison to results from other 3D model-based methods showed that SPASM performs comparable to or better than these other methods, but SPASM uses considerably less image data. Sensitivity to initial model placement proved to be limited within a range of position perturbations of approximately 20 mm in all directions.

Influence of positional and angular variation of automatically planned short-axis stacks on quantification of left ventricular dimensions and function with cardiovascular magnetic resonance

PURPOSE

To theoretically and experimentally investigate the influence of the automated cardiovascular magnetic resonance (CMR) scan planning pitfalls, namely inaccurate positioning and tilting of short-axis (SA) imaging planes, on quantification of the left ventricular (LV) dimensions and function.

MATERIALS AND METHODS

Eleven healthy subjects and eight patients underwent CMR. Manually and automatically planned SA sets were acquired. To obtain the quantitative measurements of LV function, one observer performed image analysis twice. The agreement between planning methods, as well as the decomposition of the total variation into interstudy and intraobserver components was measured.

RESULTS

The decomposition of the total variation showed that the interstudy factor accounts for 70-85% of the total variation, while the rest is due to the intraobserver factor. Moreover, the relative contribution of the interstudy factor remains independent from errors in slice positioning and small angular deviation of SA stacks from the optimal orientation. Good agreement between the theoretical and measured variability factors was observed.

CONCLUSION

Global LV function derived from the automatically planned CMR acquisitions yield accurate quantification of the human cardiovascular system. Inaccurate positioning and tilting of SA images does not affect the quantitative measurements of LV function. The computer-aided system for automated CMR has proven clinical applicability.

Determination of left ventricular long-axis orientation using MRI: changes during the respiratory and cardiac cycles in normal and diseased subjects

BACKGROUND

It has previously been shown that magnetic resonance imaging (MRI) can be used to accurately determine left ventricular (LV) long-axis orientation in healthy individuals. However, the inter- and intra-observer variability in patients with acute coronary syndrome (ACS) and chronic heart failure (CHF) has not been explored. Furthermore, the changes in LV long-axis orientation because of respiration and during the cardiac cycle remain to be determined.

METHODS

LV long-axis orientation was determined by MRI in the frontal and transverse planes in 44 subjects with no cardiac disease, 20 ACS patients and 13 CHF patients. Changes in LV long-axis orientation because of respiration were assessed in a subset of 25 subjects. Changes during the cardiac cycle were assessed in six subjects from each subject group. Reproducibility was assessed by a re-examination of 17 subjects after 28 days.

RESULTS

The inter- and intra-observer variability for LV long-axis orientation was low for all subject groups. The difference between the baseline and the 28 days examinations was -1.4+/-5.9 degrees and -0.8+/-4.4 degrees in the frontal and transverse planes, respectively. No significant change in LV long-axis orientation was found between end-expiration and end-inspiration (frontal plane, P=0.63 and transverse plane, P=0.42; n=25). No significant difference in change of the LV long-axis orientation during the cardiac cycle was found between the subject groups (frontal plane, chi-square 1.8, P=0.40 and transverse plane, chi-square 5.7, P=0.06).

CONCLUSIONS

There is a low inter-and intra-observer variability and a high reproducibility for determining LV long-axis orientation in patients with no cardiac disease as well as in patients with ACS or CHF. There is no significant change in LV long-axis orientation due to respiration, and only small changes during the cardiac cycle in these groups.

Accuracy of short-axis cardiac MRI automatically derived from scout acquisitions in free-breathing and breath-holding modes

To qualitatively assess the accuracy of automated cardiovascular magnetic resonance planning procedures devised from scout acquisitions in free-breathing and breath-holding modes, to quantitatively evaluate the accuracy of the derived left ventricular volumes, mass and function and compare these parameters with the ones obtained from the manually planned acquisitions. Ten healthy volunteers underwent cardiovascular MR (CMR) acquisitions for ventricular function assessment. Short-axis data sets of the left ventricle (LV) were manually planned and generated twice in an automatic fashion. Automated planning parameters were derived from gated scout acquisitions in free-breathing and breath-holding modes. End-diastolic volume (EDV), end-systolic volume (ESV), ejection fraction (EF), and left ventricular mass (LVM) were measured. The agreement between the manual and automatic planning methods, as well as the variability of the aforementioned measurements were assessed. The differences between two automated planning methods were also compared. The mean differences between the manual and automated CMR planning derived from gated scouts in free-breathing mode were 8.05 ml (EDV), 1.84 ml (ESV), 0.69% (EF), and 4.72 g (LVM). The comparison between manual and automated CMR planning derived from gated scouts in breath-holding mode yielded the following differences: 4.22 ml (EDV), 0.34 ml (ESV), 0.3% (EF), and -0.72 mg (LVM). The variability coefficients were 3.72 and 3.66 (EDV), 5.6 and 8.19 (ESV), 3.46 and 4.31 (EF), 6.49 and 5.20 (LVM) for the automated CMR planning methods derived from scouts in free-breathing and breath-holding modes, respectively. Automated CMR planning methods can provide accurate measurements of LV dimensions in normal subjects, and therefore may be utilized in the clinical environment to provide a cost-effective solution for functional assessment of the human cardiovascular system.

Automated Short-Axis Cardiac Magnetic Resonance Image Acquisitions

RATIONALE AND OBJECTIVE

This study investigates the use of an automated observer-independent planning system for short-axis cardiovascular magnetic resonance (MR) acquisitions in the clinical environment. The capacity of the automated method to produce accurate measurements of left ventricular dimensions and function was quantitatively assessed in normal subjects and patients.

METHODS

Fourteen healthy volunteers and 8 patients underwent cardiovascular MR (CMR) acquisitions for ventricular function assessment. Short-axis datasets of the left ventricle (LV) were acquired in 2 ways: manually planned and generated in an automatic fashion. End-diastolic volume (EDV), end-systolic volume (ESV), ejection fraction (EF), and left ventricular mass (LVM) were derived from the 2 datasets. The agreement between the manual and automatic planning methods was assessed.

RESULTS

The mean differences between the manual and automated CMR planning methods for the normal subjects and patients were 5.89 mL and 1.93 mL (EDV), 1.14 mL and -0.41 mL (ESV), 0.81% and 0.89% (EF), and 4.35 g and 3.88 g (LVM), respectively. There was no significant difference in ESV and EF. LVM significantly differed in both groups, whereas EDV was significantly different in the normal subjects and insignificantly different in the patients. The variability coefficients were 2.8 and 3.59 (EDV), 3.3 and 5.03 (ESV), 1.79 and 2.65 (EF), and 4.36 and 2.27 (LVM) for the normal subjects and patients, respectively. The mean angular deviation of the LV axes turned out to be 8.58 +/- 5.76 degrees for the normal subjects and 8.35 +/- 5.15 degrees for the patients.

CONCLUSIONS

Automated CMR planning method can provide accurate measurements of LV dimensions in normal subjects and patients, and therefore, can be used in the clinical environment for functional assessment of the human cardiovascular system.

Computerised planning of the acquisition of cardiac MR images

A method to automatically plan acquisition of magnetic resonance images aligned with the cardiac axes is presented. Localiser images are acquired with a mean short axis orientation calculated from a group of (n=50) adult patients. These images are segmented using the expectation maximization algorithm. The borders of the ventricular blood pools are found and used to provide an estimate of the orientation of the cardiac axes. These estimated orientations are compared with corresponding manually aligned orientations. The method has been tested on n=12 volunteers showing an error of within 12 degrees which is sufficiently accurate for clinical use.

Determination of the left ventricular long-axis orientation from a single short-axis MR image: relation to BMI and age

Accurate determination of imaging planes in relation to the left ventricular (LV) long-axis orientation is important for anatomical and functional evaluation as well as for serial comparisons with cardiac magnetic resonance (CMR) imaging. Therefore, a fast and reliable method to test the accuracy of CMR imaging for measuring the orientation of the LV long-axis was developed and validated. In addition, the relationship between LV long-axis orientation and body mass index (BMI), gender and age was assessed. Two approaches were used, a long-axis approach (based on a manually defined vector) and a short-axis approach (based on a calculated vector). The concordance between the two approaches was assessed in 72 healthy volunteers. The accuracy and precision of MR imaging for measuring three-dimensional orientations were tested using a LV phantom. The mean difference between the long- and short-axis approaches for measuring the LV long-axis orientation in the study population was 0 +/- 3 degrees, 0 +/- 2 degrees, and -1 +/- 3 degrees in the frontal, transverse and sagittal plane, respectively. BMI and age were shown to influence LV long-axis orientation, especially in the frontal and sagittal planes. A significant difference in LV long-axis orientation in the frontal and sagittal planes was found between genders. The correlation coefficient between MR-measured phantom orientation and true phantom orientation was >0.98 in all three orthogonal planes. These observations suggest that a single LV short-axis MR image can be used for measuring LV long-axis orientation in patients with no cardiac disease.