Deep learning-assisted model-based off-resonance correction for non-Cartesian SWI.

PURPOSE: Patient-induced inhomogeneities in the static magnetic field cause distortions and blurring (off-resonance artifacts) during acquisitions with long readouts such as in SWI. Conventional versatile correction methods based on extended Fourier models are too slow for clinical practice in computationally demanding cases such as 3D high-resolution non-Cartesian multi-coil acquisitions. THEORY: Most reconstruction methods can be accelerated when performing off-resonance correction by reducing the number of iterations, compressed coils, and correction components. Recent state-of-the-art unrolled deep learning architectures could help but are generally not adapted to corrupted measurements as they rely on the standard Fourier operator in the data consistency term. The combination of correction models and neural networks is therefore necessary to reduce reconstruction times. METHODS: Hybrid pipelines using UNets were trained stack-by-stack over 99 SWI 3D SPARKLING 20-fold accelerated acquisitions at 0.6 mm isotropic resolution using different off-resonance correction methods. Target images were obtained using slow model-based corrections based on self-estimated Δ B 0 $$ \Delta {B}_0 $$ field maps. The proposed strategies, tested over 11 volumes, are compared to model-only and network-only pipelines. RESULTS: The proposed hybrid pipelines achieved scores competing with two to three times slower baseline methods, and neural networks were observed to contribute both as pre-conditioner and through inter-iteration memory by allowing more degrees of freedom over the model design. CONCLUSION: A combination of model-based and network-based off-resonance correction was proposed to significantly accelerate conventional methods. Different promising synergies were observed between acceleration factors (iterations, coils, correction) and model/network that could be expanded in the future.

Deep Learning Reconstruction for Accelerated Spine MRI: Prospective Analysis of Interchangeability.

Background Deep learning (DL)-based MRI reconstructions can reduce examination times for turbo spin-echo (TSE) acquisitions. Studies that prospectively employ DL-based reconstructions of rapidly acquired, undersampled spine MRI are needed. Purpose To investigate the diagnostic interchangeability of an unrolled DL-reconstructed TSE (hereafter, TSEDL) T1- and T2-weighted acquisition method with standard TSE and to test their impact on acquisition time, image quality, and diagnostic confidence. Materials and Methods This prospective single-center study included participants with various spinal abnormalities who gave written consent from November 2020 to July 2021. Each participant underwent two MRI examinations: standard fully sampled T1- and T2-weighted TSE acquisitions (reference standard) and prospectively undersampled TSEDL acquisitions with threefold and fourfold acceleration. Image evaluation was performed by five readers. Interchangeability analysis and an image quality-based analysis were used to compare the TSE and TSEDL images. Acquisition time and diagnostic confidence were also compared. Interchangeability was tested using the individual equivalence index regarding various degenerative and nondegenerative entities, which were analyzed on each vertebra and defined as discordant clinical judgments of less than 5%. Interreader and intrareader agreement and concordance (κ and Kendall τ and W statistics) were computed and Wilcoxon and McNemar tests were used. Results Overall, 50 participants were evaluated (mean age, 46 years ± 18 [SD]; 26 men). The TSEDL method enabled up to a 70% reduction in total acquisition time (100 seconds for TSEDL vs 328 seconds for TSE, P < .001). All individual equivalence indexes were less than 4%. TSEDL acquisition was rated as having superior image noise by all readers (P < .001). No evidence of a difference was found between standard TSE and TSEDL regarding frequency of major findings, overall image quality, or diagnostic confidence. Conclusion The deep learning (DL)-reconstructed turbo spin-echo (TSE) method was found to be interchangeable with standard TSE for detecting various abnormalities of the spine at MRI. DL-reconstructed TSE acquisition provided excellent image quality, with a 70% reduction in examination time. German Clinical Trials Register no. DRKS00023278 © RSNA, 2022 Online supplemental material is available for this article. See also the editorial by Hallinan in this issue.

Iterative static field map estimation for off-resonance correction in non-Cartesian susceptibility weighted imaging.

PURPOSE: Patient-induced inhomogeneities in the magnetic field cause distortions and blurring during acquisitions with long readouts such as in susceptibility-weighted imaging (SWI). Most correction methods require collecting an additional ΔB0$$ \Delta {\mathrm{B}}_0 $$ field map to remove these artifacts. THEORY: The static ΔB0$$ \Delta {\mathrm{B}}_0 $$ field map can be approximated with an acceptable error directly from a single echo acquisition in SWI. The main component of the observed phase is linearly related to ΔB0$$ \Delta {\mathrm{B}}_0 $$ and the echo time (TE), and the relative impact of non- ΔB0$$ \Delta {\mathrm{B}}_0 $$ terms becomes insignificant with TE$$ \mathrm{TE} $$ >20 ms at 3 T for a well-tuned system. METHODS: The main step is to combine and unfold the multi-channel phase maps wrapped many times, and several competing algorithms are compared for this purpose. Four in vivo brain data sets collected using the recently proposed 3D spreading projection algorithm for rapid k-space sampling (SPARKLING) readouts are used to assess the proposed method. RESULTS: The estimated 3D field maps generated with a 0.6 mm isotropic spatial resolution provide overall similar off-resonance corrections compared to reference corrections based on an external ΔB0$$ \Delta {\mathrm{B}}_0 $$ acquisitions, and even improved for 2 of 4 individuals. Although a small estimation error is expected, no aftermath was observed in the proposed corrections, whereas degradations were observed in the references. CONCLUSION: A static ΔB0$$ \Delta {\mathrm{B}}_0 $$ field map estimation method was proposed to take advantage of acquisitions with long echo times, and outperformed the reference technique based on an external field map. The difference can be attributed to an inherent robustness to mismatches between volumes and external ΔB0$$ \Delta {\mathrm{B}}_0 $$ maps, and diverse other sources investigated.

Feasibility and Implementation of a Deep Learning MR Reconstruction for TSE Sequences in Musculoskeletal Imaging.

Magnetic Resonance Imaging (MRI) of the musculoskeletal system is one of the most common examinations in clinical routine. The application of Deep Learning (DL) reconstruction for MRI is increasingly gaining attention due to its potential to improve the image quality and reduce the acquisition time simultaneously. However, the technology has not yet been implemented in clinical routine for turbo spin echo (TSE) sequences in musculoskeletal imaging. The aim of this study was therefore to assess the technical feasibility and evaluate the image quality. Sixty examinations of knee, hip, ankle, shoulder, hand, and lumbar spine in healthy volunteers at 3 T were included in this prospective, internal-review-board-approved study. Conventional (TSES) and DL-based TSE sequences (TSEDL) were compared regarding image quality, anatomical structures, and diagnostic confidence. Overall image quality was rated to be excellent, with a significant improvement in edge sharpness and reduced noise compared to TSES (p < 0.001). No difference was found concerning the extent of artifacts, the delineation of anatomical structures, and the diagnostic confidence comparing TSES and TSEDL (p > 0.05). Therefore, DL image reconstruction for TSE sequences in MSK imaging is feasible, enabling a remarkable time saving (up to 75%), whilst maintaining excellent image quality and diagnostic confidence.

A parallel spatial and Bloch manifold regularized iterative reconstruction method for MR Fingerprinting.

Magnetic Resonance Fingerprinting (MRF) reconstructs tissue maps based on a sequence of very highly undersampled images. In order to be able to perform MRF reconstruction, state-of-the-art MRF methods rely on priors such as the MR physics (Bloch equations) and might also use some additional low-rank or spatial regularization. However to our knowledge these three regularizations are not applied together in a joint reconstruction. The reason is that it is indeed challenging to incorporate effectively multiple regularizations in a single MRF optimization algorithm. As a result most of these methods are not robust to noise especially when the sequence length is short. In this paper, we propose a family of new methods where spatial and low-rank regularizations, in addition to the Bloch manifold regularization, are applied on the images. We show on digital phantom and NIST phantom scans, as well as volunteer scans that the proposed methods bring significant improvement in the quality of the estimated tissue maps.

Automated detection of critical findings in multi-parametric brain MRI using a system of 3D neural networks.

With the rapid growth and increasing use of brain MRI, there is an interest in automated image classification to aid human interpretation and improve workflow. We aimed to train a deep convolutional neural network and assess its performance in identifying abnormal brain MRIs and critical intracranial findings including acute infarction, acute hemorrhage and mass effect. A total of 13,215 clinical brain MRI studies were categorized to training (74%), validation (9%), internal testing (8%) and external testing (8%) datasets. Up to eight contrasts were included from each brain MRI and each image volume was reformatted to common resolution to accommodate for differences between scanners. Following reviewing the radiology reports, three neuroradiologists assigned each study to abnormal vs normal, and identified three critical findings including acute infarction, acute hemorrhage, and mass effect. A deep convolutional neural network was constructed by a combination of localization feature extraction (LFE) modules and global classifiers to identify the presence of 4 variables in brain MRIs including abnormal, acute infarction, acute hemorrhage and mass effect. Training, validation and testing sets were randomly defined on a patient basis. Training was performed on 9845 studies using balanced sampling to address class imbalance. Receiver operating characteristic (ROC) analysis was performed. The ROC analysis of our models for 1050 studies within our internal test data showed AUC/sensitivity/specificity of 0.91/83%/86% for normal versus abnormal brain MRI, 0.95/92%/88% for acute infarction, 0.90/89%/81% for acute hemorrhage, and 0.93/93%/85% for mass effect. For 1072 studies within our external test data, it showed AUC/sensitivity/specificity of 0.88/80%/80% for normal versus abnormal brain MRI, 0.97/90%/97% for acute infarction, 0.83/72%/88% for acute hemorrhage, and 0.87/79%/81% for mass effect. Our proposed deep convolutional network can accurately identify abnormal and critical intracranial findings on individual brain MRIs, while addressing the fact that some MR contrasts might not be available in individual studies.

Accelerated real-time cardiac MRI using iterative sparse SENSE reconstruction: comparing performance in patients with sinus rhythm and atrial fibrillation.

OBJECTIVES: To compare accelerated real-time cardiac MRI (CMR) using sparse spatial and temporal undersampling and non-linear iterative SENSE reconstruction (RT IS SENSE) with real-time CMR (RT) and segmented CMR (SEG) in a cohort that included atrial fibrillation (AF) patients. METHODS: We evaluated 27 subjects, including 11 AF patients, by acquiring steady-state free precession cine images covering the left ventricle (LV) at 1.5 T with SEG (acceleration factor 2, TR 42 ms, 1.8 × 1.8 × 6 mm3), RT (acceleration factor 3, TR 62 ms, 3.0 × 3.0 × 7 mm3), and RT IS SENSE (acceleration factor 9.9-12, TR 42 ms, 2.0 × 2.0 × 7 mm3). We performed quantitative LV functional analysis in sinus rhythm (SR) patients and qualitatively scored image quality, noise and artefact using a 5-point Likert scale in the complete cohort and AF and SR subgroups. RESULTS: There was no difference between LV functional parameters between acquisitions in SR patients. RT IS SENSE short-axis image quality was superior to SEG (4.5 ± 0.6 vs. 3.9 ± 1.1, p = 0.007) and RT (3.8 ± 0.4, p = 0.003). There was reduced artefact in RT IS SENSE compared to SEG (4.4 ± 0.6 vs. 3.8 ± 1.2, p = 0.04), driven by arrhythmia performance. RT IS SENSE short-axis image quality was superior to SEG (4.6 ± 0.5 vs. 3.1 ± 1.0, p < 0.001) in the AF subgroup. CONCLUSION: Accelerated real-time CMR with iterative sparse SENSE provides excellent clinical performance, especially in patients with AF. KEY POINTS: • Iterative sparse SENSE significantly accelerates real-time cardiovascular MRI acquisitions. • It provides excellent qualitative and quantitative performance in sinus rhythm patients. • It outperforms standard segmented acquisitions in patients with atrial fibrillation. • It improves the trade-off between temporal and spatial resolution in real-time imaging.

AIR-MRF: Accelerated iterative reconstruction for magnetic resonance fingerprinting.

Existing approaches for reconstruction of multiparametric maps with magnetic resonance fingerprinting (MRF) are currently limited by their estimation accuracy and reconstruction time. We aimed to address these issues with a novel combination of iterative reconstruction, fingerprint compression, additional regularization, and accelerated dictionary search methods. The pipeline described here, accelerated iterative reconstruction for magnetic resonance fingerprinting (AIR-MRF), was evaluated with simulations as well as phantom and in vivo scans. We found that the AIR-MRF pipeline provided reduced parameter estimation errors compared to non-iterative and other iterative methods, particularly at shorter sequence lengths. Accelerated dictionary search methods incorporated into the iterative pipeline reduced the reconstruction time at little cost of quality.

Free-breathing Sparse Sampling Cine MR Imaging with Iterative Reconstruction for the Assessment of Left Ventricular Function and Mass at 3.0 T.

Purpose To prospectively evaluate the accuracy of left ventricle (LV) analysis with a two-dimensional real-time cine true fast imaging with steady-state precession (trueFISP) magnetic resonance (MR) imaging sequence featuring sparse data sampling with iterative reconstruction (SSIR) performed with and without breath-hold (BH) commands at 3.0 T. Materials and Methods Ten control subjects (mean age, 35 years; range, 25-56 years) and 60 patients scheduled to undergo a routine cardiac examination that included LV analysis (mean age, 58 years; range, 20-86 years) underwent a fully sampled segmented multiple BH cine sequence (standard of reference) and a prototype undersampled SSIR sequence performed during a single BH and during free breathing (non-BH imaging). Quantitative analysis of LV function and mass was performed. Linear regression, Bland-Altman analysis, and paired t testing were performed. Results Similar to the results in control subjects, analysis of the 60 patients showed excellent correlation with the standard of reference for single-BH SSIR (r = 0.93-0.99) and non-BH SSIR (r = 0.92-0.98) for LV ejection fraction (EF), volume, and mass (P < .0001 for all). Irrespective of breath holding, LV end-diastolic mass was overestimated with SSIR (standard of reference: 163.9 g ± 58.9, single-BH SSIR: 178.5 g ± 62.0 [P < .0001], non-BH SSIR: 175.3 g ± 63.7 [P < .0001]); the other parameters were not significantly different (EF: 49.3% ± 11.9 with standard of reference, 48.8% ± 11.8 with single-BH SSIR, 48.8% ± 11 with non-BH SSIR; P = .03 and P = .12, respectively). Bland-Altman analysis showed similar measurement errors for single-BH SSIR and non-BH SSIR when compared with standard of reference measurements for EF, volume, and mass. Conclusion Assessment of LV function with SSIR at 3.0 T is noninferior to the standard of reference irrespective of BH commands. LV mass, however, is overestimated with SSIR. © RSNA, 2016 Online supplemental material is available for this article.

Right ventricular assessment at cardiac MRI: initial clinical experience utilizing an IS-SENSE reconstruction.

Cardiac MR is considered the gold standard in assessing RV function. The purpose of this study is to evaluate the clinical utility of an investigational iterative reconstruction algorithm in the quantitative assessment of RV function. This technique has the potential to improve the clinical utility of CMR in the evaluation of RV pathologies, particularly in patients with dyspnea, by shortening acquisition times without adversely influencing imaging performance. Segmented cine images were acquired on 9 healthy volunteers and 29 patients without documented RV pathologies using conventional GRAPPA acquisition with factor 2 acceleration (GRAPPA 2), a spatio-temporal TSENSE acquisition with factor 4 acceleration (TSENSE 4), and iteratively reconstructed Sparse SENSE acquisition with factor 4 acceleration (IS-SENSE 4). 14 subjects were re-analyzed and intraclass correlation coefficients (ICC) were calculated and Bland-Altman plots generated to assess agreement. Two independent reviewers qualitatively scored images. Comparison of acquisition techniques was performed using univariate analysis of variance (ANOVA). Differences in RV EF, BSA-indexed ESV (ESVi), BSA-indexed EDV (EDVi), and BSA-indexed SV (SVi) were shown to be statistically insignificant via ANOVA testing. R(2) values for linear regression of TSENSE 4 and IS-SENSE 4 versus GRAPPA 2 were 0.34 and 0.72 for RV-EF, and 0.61 and 0.76 for RV-EDVi. ICC values for intraobserver and interobserver quantification yielded excellent agreement, and Bland-Altman plots assessing agreement were generated as well. Qualitative review yielded small, but statistically significant differences in image quality and noise between TSENSE 4 and IS-SENSE 4. All three techniques were rated nearly artifact free. Segmented imaging acquisitions with IS-SENSE reconstruction and an acceleration factor of 4 accurately and reliably quantitates RV systolic function parameters, while maintaining image quality. TSENSE-4 accelerated acquisitions showed poorer correlation to standard imaging, and inferior interobserver and intraobserver agreement. IS-SENSE has the potential to shorten cine acquisition times by 50 %, improving imaging options in patients with intermittent arrhythmias or difficulties with breath holding.

Right Ventricular Imaging in 25 Seconds: Evaluating the Use of Sparse Sampling CINE With Iterative Reconstruction for Volumetric Analysis of the Right Ventricle.

OBJECTIVE: The aim of this study was to prospectively evaluate a 2-dimensional real-time CINE TrueFISP magnetic resonance sequence using sparse data sampling with iterative reconstruction (SSIR) for right ventricular (RV) volumetry in comparison to the criterion standard (CS) acquired at 3 T. MATERIALS AND METHODS: Ten healthy controls and 20 consecutive patients scheduled for cardiac magnetic resonance imaging on a 3-T system (Magnetom Skyra; Siemens Healthcare Sector, Germany) underwent undersampled SSIR sequences with a single breath-hold (BH) as well as with shallow free breathing (NBH) and a fully sampled multi-BH sequence as CS. Right ventricular volumetry was performed with dedicated cardiac magnetic resonance software (cvi42; Circle Cardiovascular Imaging Inc, Calgary, Alberta, Canada). Agreement of SSIR with and without BH and CS for RV functional parameters (end-systolic volume [RVESV], end-diastolic volume [RVEDV], stroke volume [RVSV], and ejection fraction [RVEF]) were assessed with Bland-Altman analysis and paired t test. RESULTS: Analysis of the 30 individuals (19 male; 48 ± 14 years) revealed no significant differences when comparing CS and BH measurements for RVEDV (153.7 vs 153.6 mL, P = 0.96), RVESV (71.6 vs 72.1 mL, P = 0.78), RVSV (82.0 vs 81.6 mL, P = 0.65), and RVEF (54.9% vs 54.2%, P = 0.19). Similar results were shown when comparing CS and NBH measurements for RVEDV (153.7 vs 152.2 mL, P = 0.34), RVESV (71.6 vs 72.8 mL, P = 0.30), RVSV (82.0 vs 81.0 mL, P = 0.46), and RVEF (54.9 vs 54.4, P = 0.48). Time taken for acquisition was 350 seconds for the CS, 34 seconds for BH, and 25 seconds for NBH measurements. Additional time required for iterative reconstruction was 2 minutes and 30 seconds for the sparse sampled data sets. CONCLUSIONS: Our results demonstrate that accurate RV volumetry with SSIR data at 3 T is feasible in clinical routine within 25 seconds even without BH, which is of particular importance in patients with dyspnea.

Highly accelerated cardiac MRI using iterative SENSE reconstruction: initial clinical experience.

To evaluate the qualitative and quantitative performance of an accelerated cardiovascular MRI (CMR) protocol that features iterative SENSE reconstruction and spatio-temporal L1-regularization (IS SENSE). Twenty consecutively recruited patients and 9 healthy volunteers were included. 2D steady state free precession cine images including 3-chamber, 4-chamber, and short axis slices were acquired using standard parallel imaging (GRAPPA, acceleration factor = 2), spatio-temporal undersampled TSENSE (acceleration factor = 4), and IS SENSE techniques (acceleration factor = 4). Acquisition times, quantitative cardiac functional parameters, wall motion abnormalities (WMA), and qualitative performance (scale: 1-poor to 5-excellent for overall image quality, noise, and artifact) were compared. Breath-hold times for IS SENSE (3.0 ± 0.6 s) and TSENSE (3.3 ± 0.6) were both reduced relative to GRAPPA (8.4 ± 1.7 s, p < 0.001). No difference in quantitative cardiac function was present between the three techniques (p = 0.89 for ejection fraction). GRAPPA and IS SENSE had similar image quality (4.7 ± 0.4 vs. 4.5 ± 0.6, p = 0.09) while, both techniques were superior to TSENSE (quality: 4.1 ± 0.7, p < 0.001). GRAPPA WMA agreement with IS SENSE was good (κ > 0.60, p < 0.001), while agreement with TSENSE was poor (κ < 0.40, p < 0.001). IS SENSE is a viable clinical CMR acceleration approach to reduce acquisition times while maintaining satisfactory qualitative and quantitative performance.

Highly undersampled contrast-enhanced MRA with iterative reconstruction: Integration in a clinical setting.

PURPOSE: To integrate, optimize, and evaluate a three-dimensional (3D) contrast-enhanced sparse MRA technique with iterative reconstruction on a standard clinical MR system. METHODS: Data were acquired using a highly undersampled Cartesian spiral phyllotaxis sampling pattern and reconstructed directly on the MR system with an iterative SENSE technique. Undersampling, regularization, and number of iterations of the reconstruction were optimized and validated based on phantom experiments and patient data. Sparse MRA of the whole head (field of view: 265 × 232 × 179 mm(3) ) was investigated in 10 patient examinations. RESULTS: High-quality images with 30-fold undersampling, resulting in 0.7 mm isotropic resolution within 10 s acquisition, were obtained. After optimization of the regularization factor and of the number of iterations of the reconstruction, it was possible to reconstruct images with excellent quality within six minutes per 3D volume. Initial results of sparse contrast-enhanced MRA (CEMRA) in 10 patients demonstrated high-quality whole-head first-pass MRA for both the arterial and venous contrast phases. CONCLUSION: While sparse MRI techniques have not yet reached clinical routine, this study demonstrates the technical feasibility of high-quality sparse CEMRA of the whole head in a clinical setting. Sparse CEMRA has the potential to become a viable alternative where conventional CEMRA is too slow or does not provide sufficient spatial resolution.

Compressed sensing single-breath-hold CMR for fast quantification of LV function, volumes, and mass.

OBJECTIVES: The purpose of this study was to compare a novel compressed sensing (CS)-based single-breath-hold multislice magnetic resonance cine technique with the standard multi-breath-hold technique for the assessment of left ventricular (LV) volumes and function. BACKGROUND: Cardiac magnetic resonance is generally accepted as the gold standard for LV volume and function assessment. LV function is 1 of the most important cardiac parameters for diagnosis and the monitoring of treatment effects. Recently, CS techniques have emerged as a means to accelerate data acquisition. METHODS: The prototype CS cine sequence acquires 3 long-axis and 4 short-axis cine loops in 1 single breath-hold (temporal/spatial resolution: 30 ms/1.5 × 1.5 mm(2); acceleration factor 11.0) to measure left ventricular ejection fraction (LVEF(CS)) as well as LV volumes and LV mass using LV model-based 4D software. For comparison, a conventional stack of multi-breath-hold cine images was acquired (temporal/spatial resolution 40 ms/1.2 × 1.6 mm(2)). As a reference for the left ventricular stroke volume (LVSV), aortic flow was measured by phase-contrast acquisition. RESULTS: In 94% of the 33 participants (12 volunteers: mean age 33 ± 7 years; 21 patients: mean age 63 ± 13 years with different LV pathologies), the image quality of the CS acquisitions was excellent. LVEF(CS) and LVEF(standard) were similar (48.5 ± 15.9% vs. 49.8 ± 15.8%; p = 0.11; r = 0.96; slope 0.97; p < 0.00001). Agreement of LVSV(CS) with aortic flow was superior to that of LVSV(standard) (overestimation vs. aortic flow: 5.6 ± 6.5 ml vs. 16.2 ± 11.7 ml, respectively; p = 0.012) with less variability (r = 0.91; p < 0.00001 for the CS technique vs. r = 0.71; p < 0.01 for the standard technique). The intraobserver and interobserver agreement for all CS parameters was good (slopes 0.93 to 1.06; r = 0.90 to 0.99). CONCLUSIONS: The results demonstrated the feasibility of applying the CS strategy to evaluate LV function and volumes with high accuracy in patients. The single-breath-hold CS strategy has the potential to replace the multi-breath-hold standard cardiac magnetic resonance technique.

Wavelet-based compressed sensing using a Gaussian scale mixture model.

While initial compressed sensing (CS) recovery techniques operated under the implicit assumption that the sparse domain coefficients are independently distributed, recent results have indicated that integrating a statistical or structural dependence model of sparse domain coefficients into CS enhances recovery. In this paper, we present a method for exploiting empirical dependences among wavelet coefficients during CS recovery using a Bayes least-square Gaussian-scale-mixture model. The proposed model is successfully incorporated into several recent CS algorithms, including reweighted l(1) minimization (RL1), iteratively reweighted least squares, and iterative hard thresholding. Extensive experiments including comparisons with a state-of-the-art model-based CS method demonstrate that the proposed algorithms are highly effective at reducing reconstruction error and/or the number of measurements required for a desired reconstruction quality.

Simultaneous ODF estimation and tractography in HARDI.

We consider the problem of tracking white matter fibers in high angular resolution diffusion imaging (HARDI) data while simultaneously estimating the local fiber orientation profile. Prior work showed that an unscented Kalman filter (UKF) can be used for this problem, yet existing algorithms employ parametric mixture models to represent water diffusion and to define the state space. To address this restrictive model dependency, we propose to extend the UKF to HARDI data modeled by orientation distribution functions (ODFs), a more generic diffusion model. We consider the spherical harmonic representation of the HARDI signal as the state, enforce nonnegativity of the ODFs, and perform tractography using the directions at which the ODFs attain their peaks. In simulations, our method outperforms filtered two-tensor tractography at different levels of noise by achieving a reduction in mean Chamfer error of 0.05 to 0.27 voxels; it also produced in vivo fiber tracking that is consistent with the neuroanatomy.

View compensated compression of volume rendered images for remote visualization.

Remote visualization of volumetric images has gained importance over the past few years in medical and industrial applications. Volume visualization is a computationally intensive process, often requiring hardware acceleration to achieve a real time viewing experience. One remote visualization model that can accomplish this would transmit rendered images from a server, based on viewpoint requests from a client. For constrained server-client bandwidth, an efficient compression scheme is vital for transmitting high quality rendered images. In this paper, we present a new view compensation scheme that utilizes the geometric relationship between viewpoints to exploit the correlation between successive rendered images. The proposed method obviates motion estimation between rendered images, enabling significant reduction to the complexity of a compressor. Additionally, the view compensation scheme, in conjunction with JPEG2000 performs better than AVC, the state of the art video compression standard.

Stochastic DT-MRI connectivity mapping on the GPU.

We present a method for stochastic fiber tract mapping from diffusion tensor MRI (DT-MRI) implemented on graphics hardware. From the simulated fibers we compute a connectivity map that gives an indication of the probability that two points in the dataset are connected by a neuronal fiber path. A Bayesian formulation of the fiber model is given and it is shown that the inversion method can be used to construct plausible connectivity. An implementation of this fiber model on the graphics processing unit (GPU) is presented. Since the fiber paths can be stochastically generated independently of one another, the algorithm is highly parallelizable. This allows us to exploit the data-parallel nature of the GPU fragment processors. We also present a framework for the connectivity computation on the GPU. Our implementation allows the user to interactively select regions of interest and observe the evolving connectivity results during computation. Results are presented from the stochastic generation of over 250,000 fiber steps per iteration at interactive frame rates on consumer-grade graphics hardware.

Efficient transmission of compressed data for remote volume visualization.

One of the goals of telemedicine is to enable remote visualization and browsing of medical volumes. There is a need to employ scalable compression schemes and efficient client-server models to obtain interactivity and an enhanced viewing experience. First, we present a scheme that uses JPEG2000 and JPIP (JPEG2000 Interactive Protocol) to transmit data in a multi-resolution and progressive fashion. The server exploits the spatial locality offered by the wavelet transform and packet indexing information to transmit, in so far as possible, compressed volume data relevant to the clients query. Once the client identifies its volume of interest (VOI), the volume is refined progressively within the VOI from an initial lossy to a final lossless representation. Contextual background information can also be made available having quality fading away from the VOI. Second, we present a prioritization that enables the client to progressively visualize scene content from a compressed file. In our specific example, the client is able to make requests to progressively receive data corresponding to any tissue type. The server is now capable of reordering the same compressed data file on the fly to serve data packets prioritized as per the client's request. Lastly, we describe the effect of compression parameters on compression ratio, decoding times and interactivity. We also present suggestions for optimizing JPEG2000 for remote volume visualization and volume browsing applications. The resulting system is ideally suited for client-server applications with the server maintaining the compressed volume data, to be browsed by a client with a low bandwidth constraint.