3D Motion Estimation of Left Ventricular Dynamics Using MRI and Track-to-Track Fusion.

OBJECTIVE: This study investigates the estimation of three dimensional (3D) left ventricular (LV) motion using the fusion of different two dimensional (2D) cine magnetic resonance (CMR) sequences acquired during routine imaging sessions. Although standard clinical cine CMR data is inherently 2D, the actual underlying LV dynamics lies in 3D space and cannot be captured entirely using single 2D CMR image sequences. By utilizing the image information from various short-axis and long-axis image sequences, the proposed method intends to estimate the dynamic state vectors consisting of the position and velocity information of the myocardial borders in 3D space. METHOD: The proposed method comprises two main components: tracking myocardial points in 2D CMR sequences and fusion of multiple trajectories correspond to the tracked points. The tracking which yields the set of corresponding temporal points representing the myocardial points is performed using a diffeomorphic nonrigid image registration approach. The trajectories obtained from each cine CMR sequence is then fused with the corresponding trajectories from other CMR views using an unscented Kalman smoother (UKS) and a track-to-track fusion algorithm. RESULTS: We evaluated the proposed method by comparing the results against CMR imaging with myocardial tagging. We report a quantitative performance analysis by projecting the state vector estimates we obtained onto 2D tagged CMR images acquired from the same subjects and comparing them against harmonic phase estimates. The proposed algorithm yielded a competitive performance with a mean root mean square error of 1.3±0.5 pixels (1.8±0.6 mm) evaluated over 118 image sequences acquired from 30 subjects. CONCLUSION: This study demonstrates that fusing the information from short and long-axis views of CMR improves the accuracy of cardiac tissue motion estimation. Clinical Impact: The proposed method demonstrates that the fusion of tissue tracking information from long and short-axis views improves the binary classification of the automated regional function assessment.

MRI-based automated detection of implanted low dose rate (LDR) brachytherapy seeds using quantitative susceptibility mapping (QSM) and unsupervised machine learning (ML).

BACKGROUND AND PURPOSE: Permanent seed brachytherapy is an established treatment option for localized prostate cancer. Currently, post-implant dosimetry is performed on CT images despite challenging target delineation due to limited soft tissue contrast. This work aims to develop an MRI-only workflow for post-implant dosimetry of prostate brachytherapy seeds. MATERIAL AND METHODS: A prostate mimicking phantom containing twenty stranded I-125 dummy seeds and calcifications was constructed. A three-dimensional gradient-echo MR sequence was employed on 3T and 1.5T MR scanners. An optimized quantitative susceptibility mapping (QSM) technique was applied to generate positive contrast for the seeds and calcifications. Seed numbers, centroids, and orientations were determined using unsupervised machine learning algorithms (K-means and K-medoids clustering). The geometrical seed positions and the resulting dose distribution were compared to the clinical CT-based approach. RESULTS: The optimized QSM-based method generated high quality positive contrast for the seeds that were significantly different from that for calcifications and could be easily differentiated by thresholding. The estimated seed centroids from both 3T and 1.5T MR data were in perfect agreement with the standard CT-based seed detection algorithm (maximum difference of 0.7 mm). The estimated seed orientations were highly correlated with the actual orientations (R > 0.98). CONCLUSIONS: The proposed MRI-based workflow enabling an accurate and robust means to localize the seeds (position and orientation) upon validation on complex seed configurations, has the potential to replace the current widely practiced CT-based workflow.

Direction modulated brachytherapy (DMBT) tandem applicator for cervical cancer treatment: Choosing the optimal shielding material.

PURPOSE: To investigate the dose modulation capability of a novel MRI-compatible direction modulated brachytherapy (DMBT) tandem applicator design with various high-density shielding materials for brachytherapy treatment of cervical cancer. The shield materials that have been evaluated are tantalum (Ta), pure tungsten (W), gold (Au), rhenium (Re), osmium (Os), platinum (Pt), iridium (Ir), and W' tungsten alloy (95%W, 3.5%Ni, 1.5%Cu). MATERIALS AND METHODS: The recently proposed six-channel DMBT tandem is composed of nonmagnetic tungsten alloy (W') rod with diameter of 5.4 mm and coated with 0.3-mm thick bio-safe plastic sheath. The tandem shielding material can, however, be individually replaced with various other shields to create directional radiation. Monte Carlo N-Particle (MCNP) code was used to calculate the three-dimensional (3D) dose distributions in a water phantom for an HDR 192 Ir (mHDR-v2) source inside each DMBT tandem with various shields and a plastic conventional tandem (Con.T). Then, the 3D dose distributions were imported into an in-house-coded inverse planning optimization algorithm to obtain optimal plans for 12 clinical cases chosen at random from the international RetroEMBRACE dataset involving conventional tandem and ring (Con.T&R) applicators. All plans generated by the DMBT tandem and ring (DMBT&R) with the tungsten alloy [DMBT(W')&R] were compared with the corresponding Con.T&R plans, to generate benchmark results. These benchmark results were then considered as reference plans for other shields performances. Plans were normalized to receive the same high-risk clinical target volume (CTVHR ) D90 . The D100 , D10 , and V100 for CTVHR , and D2cm3 for organs at risk (OARs) of bladder, sigmoid, and rectum were calculated and compared. RESULTS: Transmission factor (TF), that is, the dose in the backside of the DMBT shield over that in the front opening, at a 5 cm distance, were 36.6%, 34.8%, 31.9%, 28.9%, 27.9%, 26.2%, 26.2%, and 25.5%, for Ta, W', W, Re, Au, Os, Pt, and Ir shields, respectively. On average, the CTVHR values for D100 , V100 , D10 were not significantly different across all DMBT&R shields and the Con.T&R plans (P > 0.219). For the D2cm3 , the benchmark results showed significant reductions (P < 0.03), that is, on average, -8.3% for bladder, -10.7% for rectum, and -10.1% for sigmoid, compared to the Con.T&R plans. However, the various shields showed little improvement from the tungsten alloy (W'), where on average, rectum (bladder) [sigmoid] D2cm3 were reduced by -1.32% (-0.85%) [-1.01%], -1.25% (-0.78%) [-0.91%], -1.22% (-0.75%) [-0.86%], -0.94% (-0.60%) [-0.70%], -0.84% (-0.51%) [-0.59%], and -0.38% (-0.24%) [-0.23%] for Ir, Pt, Os, Au, Re, and W shields, relative to the benchmark W' DMBT plans, respectively. These corresponding values for Ta increased by +0.28% (+0.08%) [+0.25%], respectively. CONCLUSION: The Ir, Pt, Os, Au, Re, and W shielding materials, respectively, in descending order, lead to better OAR sparing than the DMBT(W')&R plans. However, the amount of improvement is limited and clinically insignificant. This finding suggests that the initial W' shield remains a suitable choice given the proven MR compatibility, for use in MR-guided adaptive brachytherapy of cervical cancer.

Quantitative CT assessment of a novel direction-modulated brachytherapy tandem applicator.

PURPOSE: The purpose of this study was to quantitatively assess the CT metal-induced artifacts from a novel direction-modulated brachytherapy (DMBT) tandem applicator prototype, recently designed for cervical cancer treatments. METHODS AND MATERIALS: A water-based pelvic phantom was constructed for CT scanning. The DMBT applicator was imaged using our institutional protocol, one with higher kVp and mAs settings, and repetition of these protocols using 3-mm slices. A conventional stainless steel applicator was also scanned. In addition to the standard reconstructed images, applicator images were reconstructed using a commercial metal artifact-reduction (MAR) algorithm and an in-house-developed research algorithm. Subsequently, image quality and artifact severity were evaluated. RESULTS: Artifact severity, measured in terms of SDs in CT numbers, decreased asymptotically to background water levels with the distance away from the applicator. Artifact-reduction algorithms lead to significant and visible improvements in image quality, with >50% and >20% decrease in artifact severity achieved at a 10-mm distance for the DMBT and stainless steel applicators, respectively. Differences in artifact severity were minimal between the four imaging protocols. DMBT dimensions were the same on images with and without the commercial MAR algorithm, within <1 mm of the theoretical value. Both the commercial and in-house algorithms restored the CT numbers outside the applicator, albeit a better performance was achieved by the in-house algorithm. CONCLUSIONS: The artifacts produced by both applicators were minimized with the use of MAR algorithms. Adoption of the DMBT and stainless steel applicators for CT-guided brachytherapy is anticipated as MAR algorithms are widely available on CT scanners.

Direction modulated brachytherapy (DMBT) for treatment of cervical cancer: A planning study with 192 Ir, 60 Co, and 169 Yb HDR sources.

PURPOSE: To evaluate plan quality of a novel MRI-compatible direction modulated brachytherapy (DMBT) tandem applicator using 192 Ir, 60 Co, and 169 Yb HDR brachytherapy sources, for various cervical cancer high-risk clinical target volumes (CTVHR ). MATERIALS AND METHODS: The novel DMBT tandem applicator has six peripheral grooves of 1.3-mm diameter along a 5.4-mm thick nonmagnetic tungsten alloy rod. Monte Carlo (MC) simulations were used to benchmark the dosimetric parameters of the 192 Ir, 60 Co, and 169 Yb HDR sources in a water phantom against the literature data. 45 clinical cases that were treated using conventional tandem-and-ring applicators with 192 Ir source (192 Ir-T&R) were selected consecutively from intErnational MRI-guided BRAchytherapy in CErvical cancer (EMBRACE) trial. Then, for each clinical case, 3D dose distribution of each source inside the DMBT and conventional applicators were calculated and imported onto an in-house developed inverse planning optimization code to generate optimal plans. All plans generated by the DMBT tandem-and-ring (DMBT T&R) from all three sources were compared to the respective 192 Ir-T&R plans. For consistency, all plans were normalized to the same CTVHR D90 achieved in clinical plans. The D2 cm3 for organs at risk (OAR) such as bladder, rectum, and sigmoid, and D90, D98, D10, V100, and V200 for CTVHR were calculated. RESULTS: In general, plan quality significantly improved when a conventional tandem (Con.T) is replaced with the DMBT tandem. The target coverage metrics were similar across 192 Ir-T&R and DMBT T&R plans with all three sources (P > 0.093). 60 Co-DMBT T&R generated greater hot spots and less dose homogeneity in the target volumes compared with the 192 Ir- and 169 Yb-DMBT T&R plans. Mean OAR doses in the DMBT T&R plans were significantly smaller (P < 0.0084) than the 192 Ir-T&R plans. Mean bladder D2 cm3 was reduced by 4.07%, 4.15%, and 5.13%, for the 192 Ir-, 60 Co-, and 169 Yb-DMBT T&R plans respectively. Mean rectum (sigmoid) D2 cm3 was reduced by 3.17% (3.63%), 2.57% (3.96%), and 4.65% (4.34%) for the 192 Ir-, 60 Co-, and 169 Yb-DMBT T&R plans respectively. The DMBT T&R plans with the 169 Yb source generally resulted in the greatest OAR sparing when the CTVHR were larger and irregular in shape, while for smaller and regularly shaped CTVHR (<30 cm3 ), OAR sparing between the sources were comparable. CONCLUSIONS: The DMBT tandem provides a promising alternative to the Con.T design with significant improvement in the plan quality for various target volumes. The DMBT T&R plans generated with the three sources of varying energies generated superior plans compared to the conventional T&R applicators. Plans generated with the 169 Yb-DMBT T&R produced best results for larger and irregularly shaped CTVHR in terms of OAR sparing. Thus, this study suggests that the combination of the DMBT tandem applicator with varying energy sources can work synergistically to generate improved plans for cervical cancer brachytherapy.

A realistic phantom for validating MRI-based synthetic CT images of the human skull.

PURPOSE: To introduce a new realistic human skull phantom for the validation of synthetic CT images of cortical bone from ultra-short echo-time (UTE) sequences. METHODS: A human skull of an adult female was utilized as a realistic representation of skull cortical bone. The skull was stabilized in a special acrylic container and was filled with contrast agents that have T1 and T2 relaxation times similar to human brain. The phantom was MR scanned at 3T with UTE and T2 -weighted sequences, followed by CT. A clustering approach was developed to extract the cortical bone signal from MR images. T2∗ maps of the skull were calculated. Synthetic CT images of the bone were compared to cortical bone signal extracted from CT images and confounding factors, such as registration errors, were analyzed. RESULTS: Dice similarity coefficient (DSC) of UTE-detected cortical bone was 0.84 and gradually decreased with decreasing number of spokes. DSC did not significantly depend on echo-time. Registration errors were found to be significant confounding factors, with 25% decrease in DSC for consistent 2 mm error at each axis. CONCLUSION: This work introduced a new realistic human skull phantom, specifically for the evaluation and analysis of synthetic CT images of cortical bone.

Sensitivity of clinically relevant dosimetric parameters to contouring uncertainty in postimplant dosimetry of low-dose-rate prostate permanent seed brachytherapy.

PURPOSE: There is strong evidence relating postimplant dosimetry for low-dose-rate prostate seed brachytherapy to local control rates. The delineation of the prostate on CT images, however, represents a challenge due to the lack of soft-tissue contrast to identify the prostate borders. This study aims at quantifying the sensitivity of prostate V100 and D90 to contouring uncertainty as clinically relevant parameters for evaluation of target coverage in postimplant dosimetry. METHODS AND MATERIALS: CT images, postoperative plans, and contours of a cohort of patients (n = 43; low risk = 55.8%, intermediate risk = 39.5%, high risk = 4.7%), who had received prostate seed brachytherapy, were imported into MIM Symphony treatment planning system. The prostate contours in postimplant CT images were expanded or contracted uniformly in extents of ±1.00 mm, ±2.00 mm, ±3.00 mm, ±4.00 mm, and ±5.00 mm. The values for V100 and D90 were extracted from dose-volume histograms for each contour and compared. RESULTS: Significant changes were observed in the values of D90 and V100 as well as the number of inacceptable plans for expansion or contraction of only few millimeters. Evaluation of a plan coverage based on D90 was found to be less sensitive to systematic contouring errors compared with V100. Number of plans incorrectly identified for lack or adequacy of coverage is lower using D90 compared with V100 for the same margin of error. CONCLUSIONS: Evaluation of a plan coverage based on V100 is too sensitive to systematic contouring errors of prostate. D90 increases the accuracy of CT-based postimplant quality assurance in identifying plans with insufficient coverage compared with V100.

Direction Modulated Brachytherapy for Treatment of Cervical Cancer. II: Comparative Planning Study With Intracavitary and Intracavitary-Interstitial Techniques.

PURPOSE: To perform a comprehensive comparative planning study evaluating the utility of the proposed direction modulated brachytherapy (DMBT) tandem applicator against standard applicators, in the setting of image guided adaptive brachytherapy of cervical cancer. METHODS AND MATERIALS: A detailed conceptual article was published in 2014. The proposed DMBT tandem applicator has 6 peripheral grooves of 1.3-mm width, along a 5.4-mm-thick nonmagnetic tungsten alloy rod of density 18.0 g/cm(3), capable of generating directional dose profiles. We performed a comparative planning study with 45 cervical cancer patients enrolled consecutively in the prospective observational EMBRACE study. In all patients, MRI-based planning was performed while utilizing various tandem-ring (27 patients) and tandem-ring-needles (18 patients) applicators, in accordance with the Groupe Européen de Curiethérapie-European Society for Radiotherapy and Oncology recommendations. For unbiased comparisons, all cases were replanned with an in-house-developed inverse optimization code while enforcing a uniform set of constraints that are reflective of the clinical practice. All plans were normalized to the same high-risk clinical target volume D90 values achieved in the original clinical plans. RESULTS: In general, if the standard tandem was replaced with the DMBT tandem while maintaining all other planning conditions the same, there was consistent improvement in the plan quality. For example, among the 18 tandem-ring-needles cases, the average D2cm(3) reductions achieved were -2.48% ± 11.03%, -4.45% ± 5.24%, and -5.66% ± 6.43% for the bladder, rectum, and sigmoid, respectively. An opportunity may also exist in avoiding use of needles altogether for when the total number of needles required is small (approximately 2 to 3 needles or less), if DMBT tandem is used. CONCLUSIONS: Integrating the novel DMBT tandem onto both intracavitary and intracavitary-interstitial applicator assembly enabled consistent improvement in the sparing of the OARs, over a standard "single-channel" tandem, though individual variations in benefit were considerable. Although at an early stage of development, the DMBT concept design is demonstrated to be useful and pragmatic for potential clinical translation.

Metal artefacts in MRI-guided brachytherapy of cervical cancer.

The importance of assessing the metal-induced artefacts in magnetic resonance imaging (MRI)-guided brachytherapy is growing along with the increasing interest of integrating MRI into the treatment procedure of cervical cancer. Examples of metal objects in use include intracavitary cervical applicators and interstitial needles. The induced artefacts increase the uncertainties in the clinical workflow and can be a potential obstacle for the accurate delivery of the treatment. Overcoming this problem necessitates a good understanding of its originating sources. Several efforts are recorded in the literature to quantify the extent of such artefacts, in phantoms and in clinical practice. Here, we elaborate on the origin of metal-induced artefacts in the light of brachytherapy applications, while summarizing recent efforts that have been made to assess and overcome the induced distortions.

Quantitative MRI assessment of a novel direction modulated brachytherapy tandem applicator for cervical cancer at 1.5T.

BACKGROUND AND PURPOSE: The purpose of this work is to quantitatively investigate the artifacts and image distortions induced in the MR images by a recently proposed direction modulated brachytherapy (DMBT) tandem applicator prototype. This new MRI-compatible applicator allows better sparing of organs-at-risk (OAR) for cervical cancer patients, while providing conformal dose distributions to target volumes. MATERIALS AND METHODS: Specific phantom and tools were designed and manufactured for this study. The phantom was filled with a tissue-like solution and MR images were acquired with clinical protocols as per GEC-ESTRO recommendations. Images were obtained at 6 different orientations that mimic possible clinical settings and full-width-at-half-maximum (FWHM) was recorded at multiple locations/angles. The accuracy of detecting the centerline of the tandem was assessed using a novel radial-fiducials mount. RESULTS: FWHM from all line profiles at all angles and all orientations was 6.14±0.7mm (compared to 6mm of the actual DMBT tandem diameter). The in-plane spatial-shift observed at para-axial and para-sagittal views was less than 0.5mm. CONCLUSIONS: This work demonstrated that the novel DMBT tandem applicator prototype has minimal artifact in T2-weighted images employed in clinical practice, suggesting the applicator might be a good candidate for MRI-guided adaptive brachytherapy.

Fat quantification using an interleaved bipolar acquisition.

PURPOSE: To demonstrate a new multigradient echo bipolar acquisition sequence for fat quantification. THEORY AND METHODS: A multiecho bipolar acquisition is used such that the even echoes have opposite polarity to the odd echoes. In addition, the readout gradients alternate their polarities every other phase-encode line. Each echo, therefore, consists of phase-encode lines with both positive and negative polarities. Phase-encodes acquired with the same polarity are grouped together, and parallel imaging reconstruction is used to obtain two full k-space maps with opposite readout polarities at all the echoes. By complex averaging, the inconsistent phase errors between odd and even echoes are removed and water/fat separation techniques used with conventional unipolar sequences can be performed. RESULTS: Phantoms and in vivo experiments demonstrated accurate fat fraction and increased signal to noise ratio efficiency compared with the established unipolar acquisition. Artefacts due to phase and magnitude errors of bipolar acquisition were eliminated in all experiments. CONCLUSION: The interleaved bipolar sequence is an efficient technique for fat quantification. It demonstrated accurate fat measurements in a shorter scan time compared with the standard unipolar sequence.

Magnetic resonance imaging-guided brachytherapy for cervical cancer: initiating a program.

Over the past decade, the application of magnetic resonance imaging (MRI) has increased, and there is growing evidence to suggest that improvements in accuracy of target delineation in MRI-guided brachytherapy may improve clinical outcomes in cervical cancer. To implement a high quality image guided brachytherapy program, a multidisciplinary team is required with appropriate expertise as well as an adequate patient load to ensure a sustainable program. It is imperative to know that the most important source of uncertainty in the treatment process is related to target delineation and therefore, the necessity of training and expertise as well as quality assurance should be emphasized. A short review of concepts and techniques that have been developed for implementation and/or improvement of workflow of a MRI-guided brachytherapy program are provided in this document, so that institutions can use and optimize some of them based on their resources to minimize their procedure times.

Validation of volumetric and single-slice MRI adipose analysis using a novel fully automated segmentation method.

PURPOSE: To validate a fully automated adipose segmentation method with magnetic resonance imaging (MRI) fat fraction abdominal imaging. We hypothesized that this method is suitable for segmentation of subcutaneous adipose tissue (SAT) and intra-abdominal adipose tissue (IAAT) in a wide population range, easy to use, works with a variety of hardware setups, and is highly repeatable. MATERIALS AND METHODS: Analysis was performed comparing precision and analysis time of manual and automated segmentation of single-slice imaging, and volumetric imaging (78-88 slices). Volumetric and single-slice data were acquired in a variety of cohorts (body mass index [BMI] 15.6-41.76) including healthy adult volunteers, adolescent volunteers, and subjects with nonalcoholic fatty liver disease and lipodystrophies. A subset of healthy volunteers was analyzed for repeatability in the measurements. RESULTS: The fully automated segmentation was found to have excellent agreement with manual segmentation with no substantial bias across all study cohorts. Repeatability tests showed a mean coefficient of variation of 1.2 ± 0.6% for SAT, and 2.7 ± 2.2% for IAAT. Analysis with automated segmentation was rapid, requiring 2 seconds per slice compared with 8 minutes per slice with manual segmentation. CONCLUSION: We demonstrate the ability to accurately and rapidly segment regional adipose tissue using fat fraction maps across a wide population range, with varying hardware setups and acquisition methods.

Max-IDEAL: a max-flow based approach for IDEAL water/fat separation.

PURPOSE: To propose a novel approach to water/fat separation using a unique smoothness constraint. THEORY AND METHODS: Chemical-shift based water/fat separation is an established noninvasive imaging tool for the visualization of body fat in various anatomies. Nevertheless, B0 magnetic field inhomogeneities can hamper the water/fat separation process. In this work, B0 variations are mapped using a convex-relaxed labeling model which produces a coarse estimate of the field map, while considering T2* decay during the labeling process. Fat and water components are subsequently resolved using T2*-IDEAL. An adaptive spatial filtering (ASF) was introduced to improve the robustness of the estimate. The method was tested on cardiac and abdominal datasets from healthy volunteers and nonalcoholic fatty liver disease (NAFLD) patients. RESULTS: Out of 168 cardiac and abdominal images, only 1 case has shown water/fat swaps that can hinder the clinical interpretation of the underlying anatomy. CONCLUSION: This work demonstrates a new water/fat separation approach that prevents the occurrence of water/fat swaps, by means of a unique smoothness constraint. Incorporating T2* effect in the labeling procedure and including the ASF processing enhance the robustness of the proposed approach and permit the procedure to handle abrupt B0 variations within the field of view.

Intra-thoracic fat volume is associated with myocardial infarction in patients with metabolic syndrome.

BACKGROUND: Visceral adiposity is increased in those with Metabolic Syndrome (MetS) and atherosclerotic disease burden. In this study we evaluate for associations between intra-thoracic fat volume (ITFV) and myocardial infarction (MI) in patients with MetS. METHODS: Ninety-four patients with MetS, MI or both were identified from a cardiovascular CMR clinical registry. MetS was defined in accordance to published guidelines; where-as MI was defined as the presence of subendocardial-based injury on late gadolinium enhancement imaging in a coronary vascular distribution. A healthy control group was also obtained from the same registry. Patients were selected into the following groups: MetS+/MI- (N = 32), MetS-/MI + (N = 30), MetS+/MI + (N = 32), MetS-/MI- (N = 16). ITFV quantification was performed using signal threshold analysis of sequential sagittal CMR datasets (HASTE) and indexed to body mass index. RESULTS: The mean age of the population was 59.8 ± 12.5 years. MetS+ patients (N=64) demonstrated a significantly higher indexed ITFV compared to MetS- patients (p = 0.05). Patients in respective MetS-/MI-, MetS+/MI-, MetS-/MI+, and MetS+/MI + study groups demonstrated a progressive elevation in the indexed ITFV (22.3 ± 10.6, 28.6 ± 12.6, 30.6 ± 12.3, and 35.2 ± 1.4 ml/kg/m(2), (p = 0.002)). Among MetS+ patients those with MI showed a significantly higher indexed ITFV compared to those without MI (p = 0.02). CONCLUSIONS: ITFV is elevated in patients with MetS and incrementally elevated among those with evidence of prior ischemic myocardial injury. Accordingly, the quantification of ITFV may be a valuable marker of myocardial infarction risk among patients with MetS and warrants further investigation.

A convex relaxation approach to fat/water separation with minimum label description.

While Magnetic Resonance Imaging is capable of separating water and fat components in the body, mapping of magnetic field inhomogeneities is essential for the successful application of this process. In this study, we address the problem of field map estimation using a convex-relaxed max-flow method. We propose a novel two-stage approach that leads to the global optimum of the proposed problem. The first stage minimizes the signal residuals via a convex-relaxed minimum description length (MDL)-based approach. The MDL-based labeling model penalizes the total number of appearing labels, which helps to avoid field map errors when abrupt changes in field homogeneity exist. By exploringthe whole range of possible frequency offsets, this stage ensures limiting the estimated field offset within certain boundaries where the global minimum resides. The second stage employs the output of the labeling model in a commonly used gradient-descent based method (known as IDEAL) to converge to the exact global minimum, i.e. the required value of the field offset. Experimental results for cardiac imaging, where challenging field inhomogeneities exist, showed that our method significantly outperforms over a widely-used technique for fat/water separation in terms of robustness and efficiency.