Correction of intensity nonuniformity in spin-echo T(1)-weighted images

Jointly estimating bias field and reconstructing uniform MRI image by deep learning

Bias field is one of the main artifacts that degrade the quality of magnetic resonance images. It introduces intensity inhomogeneity and affects image analysis such as segmentation. In this work, we proposed a deep learning approach to jointly estimate bias field and reconstruct uniform image. By modeling the quality degradation process as the product of a spatially varying field and a uniform image, the network was trained on 800 images with true bias fields from 12 healthy subjects. A network structure of bias field estimation and uniform image reconstruction was designed to compensate for the intensity loss. To further evaluate the benefit of bias field correction, a quantitative analysis was made on image segmentation. Experimental results show that the proposed BFCNet improves the image uniformity by 8.3% and 10.1%, the segmentation accuracy by 4.1% and 6.8% on white and grey matter in T2-weighted brain images. Moreover, BFCNet outperforms the state-of-the-art traditional methods and deep learning methods on estimating bias field and preserving image structure, and BFCNet is robust to different levels of bias field and noise.

Empirical transmit field bias correction of T1w/T2w myelin maps

T1-weighted divided by T2-weighted (T1w/T2w) myelin maps were initially developed for neuroanatomical analyses such as identifying cortical areas, but they are increasingly used in statistical comparisons across individuals and groups with other variables of interest. Existing T1w/T2w myelin maps contain radiofrequency transmit field (B1+) biases, which may be correlated with these variables of interest, leading to potentially spurious results. Here we propose two empirical methods for correcting these transmit field biases using either explicit measures of the transmit field or alternatively a 'pseudo-transmit' approach that is highly correlated with the transmit field at 3T. We find that the resulting corrected T1w/T2w myelin maps are both better neuroanatomical measures (e.g., for use in cross-species comparisons), and more appropriate for statistical comparisons of relative T1w/T2w differences across individuals and groups (e.g., sex, age, or body-mass-index) within a consistently acquired study at 3T. We recommend that investigators who use the T1w/T2w approach for mapping cortical myelin use these B1+ transmit field corrected myelin maps going forward.

B1 inhomogeneity correction of RARE MRI at low SNR: Quantitative in vivo 19 F MRI of mouse neuroinflammation with a cryogenically-cooled transceive surface radiofrequency probe

PURPOSE

Low SNR in fluorine-19 (¹⁹ F) MRI benefits from cryogenically-cooled transceive surface RF probes (CRPs), but strong B₁ inhomogeneities hinder quantification. Rapid acquisition with refocused echoes (RARE) is an SNR-efficient method for MRI of neuroinflammation with perfluorinated compounds but lacks an analytical signal intensity equation to retrospectively correct B₁ inhomogeneity. Here, a workflow was proposed and validated to correct and quantify ¹⁹ F-MR signals from the inflamed mouse brain using a ¹⁹ F-CRP.

METHODS

In vivo ¹⁹ F-MR images were acquired in a neuroinflammation mouse model with a quadrature ¹⁹ F-CRP using an imaging setup including 3D-printed components to acquire co-localized anatomical and ¹⁹ F images. Model-based corrections were validated on a uniform ¹⁹ F phantom and in the neuroinflammatory model. Corrected ¹⁹ F-MR images were benchmarked against reference images and overlaid on in vivo ¹ H-MR images. Computed concentration uncertainty maps using Monte Carlo simulations served as a measure of performance of the B₁ corrections.

RESULTS

Our study reports on the first quantitative in vivo ¹⁹ F-MR images of an inflamed mouse brain using a ¹⁹ F-CRP, including in vivo T₁ calculations for ¹⁹ F-nanoparticles during pathology and B₁ corrections for ¹⁹ F-signal quantification. Model-based corrections markedly improved ¹⁹ F-signal quantification from errors > 50% to < 10% in a uniform phantom (p < 0.001). Concentration uncertainty maps ex vivo and in vivo yielded uncertainties that were generally < 25%. Monte Carlo simulations prescribed SNR ≥ 10.1 to reduce uncertainties < 10%, and SNR ≥ 4.25 to achieve uncertainties < 25%.

CONCLUSION

Our model-based correction method facilitated ¹⁹ F signal quantification in the inflamed mouse brain when using the SNR-boosting ¹⁹ F-CRP technology, paving the way for future low-SNR ¹⁹ F-MRI applications in vivo.

ABCnet: Adversarial bias correction network for infant brain MR images

Automatic correction of intensity nonuniformity (also termed as the bias correction) is an essential step in brain MR image analysis. Existing methods are typically developed for adult brain MR images based on the assumption that the image intensities within the same brain tissue are relatively uniform. However, this assumption is not valid in infant brain MR images, due to the dynamic and regionally-heterogeneous image contrast and appearance changes, which are caused by the underlying spatiotemporally-nonuniform myelination process. Therefore, it is not appropriate to directly use existing methods to correct the infant brain MR images. In this paper, we propose an end-to-end 3D adversarial bias correction network (ABCnet), tailored for direct prediction of bias fields from the input infant brain MR images for bias correction. The "ground-truth" bias fields for training our network are carefully defined by an improved N4 method, which integrates manually-corrected tissue segmentation maps as anatomical prior knowledge. The whole network is trained alternatively by minimizing generative and adversarial losses. To handle the heterogeneous intensity changes, our generative loss includes a tissue-aware local intensity uniformity term to reduce the local intensity variation in the corrected image. Besides, it also integrates two additional terms to enhance the smoothness of the estimated bias field and to improve the robustness of the proposed method, respectively. Comprehensive experiments with different sizes of training datasets have been carried out on a total of 1492 T1w and T2w MR images from neonates, infants, and adults, respectively. Both qualitative and quantitative evaluations on simulated and real datasets consistently demonstrate the superior performance of our ABCnet in both accuracy and efficiency, compared with popularly available methods.

B1 inhomogeneity correction of RARE MRI with transceive surface radiofrequency probes

PURPOSE

The use of surface radiofrequency (RF) coils is common practice to boost sensitivity in (pre)clinical MRI. The number of transceive surface RF coils is rapidly growing due to the surge in cryogenically cooled RF technology and ultrahigh-field MRI. Consequently, there is an increasing need for effective correction of the excitation field ( B 1 + ) inhomogeneity inherent in these coils. Retrospective B₁ correction permits quantitative MRI, but this usually requires a pulse sequence-specific analytical signal intensity (SI) equation. Such an equation is not available for fast spin-echo (Rapid Acquisition with Relaxation Enhancement, RARE) MRI. Here we present, test, and validate retrospective B₁ correction methods for RARE.

METHODS

We implemented the commonly used sensitivity correction and developed an empirical model-based method and a hybrid combination of both. Tests and validations were performed with a cryogenically cooled RF probe and a single-loop RF coil. Accuracy of SI quantification and T₁ contrast were evaluated after correction.

RESULTS

The three described correction methods achieved dramatic improvements in B₁ homogeneity and significantly improved SI quantification and T₁ contrast, with mean SI errors reduced from >40% to >10% following correction in all cases. Upon correction, images of phantoms and mouse heads demonstrated homogeneity comparable to that of images acquired with a volume resonator. This was quantified by SI profile, SI ratio (error < 10%), and percentage of integral uniformity (PIU > 80% in vivo and ex vivo compared to PIU > 87% with the reference RF coil).

CONCLUSION

This work demonstrates the efficacy of three B₁ correction methods tailored for transceive surface RF probes and RARE MRI. The corrected images are suitable for quantification and show comparable results between the three methods, opening the way for T₁ measurements and X-nuclei quantification using surface transceiver RF coils. This approach is applicable to other MR techniques for which no analytical SI exists.

Magnetic Resonance Imaging Texture Analysis Predicts Recurrence in Patients with Nasopharyngeal Carcinoma

Background

The personalization of oncologic treatment using radiomic signatures is mounting in nasopharyngeal carcinoma (NPC). We ascertain the predictive ability of 3D volumetric magnetic resonance imaging (MRI) texture features on NPC disease recurrence.

Methods

A retrospective study of 58 patients with NPC undergoing primary curative-intent treatment was performed. Forty-two image texture features were extracted from pre-treatment MRI in addition to clinical factors. A multivariate logistic regression was used to model the texture features. A receiver operating characteristic curve on 100 bootstrap samples was used to maximize generalizability to out-of-sample data. A Cox proportional model was used to predict disease recurrence in the final model.

Results

A total of 58 patients were included in the study. MRI texture features predicted disease recurrence with an area under the curve (AUC), sensitivity, and specificity of 0.79, 0.73, and 0.71, respectively. Loco-regional recurrence was predicted with AUC, sensitivity, and specificity of 0.82, 0.73 and 0.74 respectively while prediction for distant metastasis had an AUC, sensitivity, and specificity of 0.92, 0.79 and 0.84, respectively. Texture features on MRI had a hazard ratio of 4.37 (95% confidence interval 1.72–20.2) for disease recurrence when adjusting for age, sex, smoking, and TNM staging.

Conclusion

Texture features on MRI are independent predictors of NPC recurrence in patients undergoing curative-intent treatment.

Comparison of quality control methods for automated diffusion tensor imaging analysis pipelines

The processing of brain diffusion tensor imaging (DTI) data for large cohort studies requires fully automatic pipelines to perform quality control (QC) and artifact/outlier removal procedures on the raw DTI data prior to calculation of diffusion parameters. In this study, three automatic DTI processing pipelines, each complying with the general ENIGMA framework, were designed by uniquely combining multiple image processing software tools. Different QC procedures based on the RESTORE algorithm, the DTIPrep protocol, and a combination of both methods were compared using simulated ground truth and artifact containing DTI datasets modeling eddy current induced distortions, various levels of motion artifacts, and thermal noise. Variability was also examined in 20 DTI datasets acquired in subjects with vascular cognitive impairment (VCI) from the multi-site Ontario Neurodegenerative Disease Research Initiative (ONDRI). The mean fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) were calculated in global brain grey matter (GM) and white matter (WM) regions. For the simulated DTI datasets, the measure used to evaluate the performance of the pipelines was the normalized difference between the mean DTI metrics measured in GM and WM regions and the corresponding ground truth DTI value. The performance of the proposed pipelines was very similar, particularly in FA measurements. However, the pipeline based on the RESTORE algorithm was the most accurate when analyzing the artifact containing DTI datasets. The pipeline that combined the DTIPrep protocol and the RESTORE algorithm produced the lowest standard deviation in FA measurements in normal appearing WM across subjects. We concluded that this pipeline was the most robust and is preferred for automated analysis of multisite brain DTI data.

Surface Coil Intensity Correction in Magnetic Resonance Imaging in Spinal Metastases

Objective

To evaluate the clinical application of phased-array surface coil intensity correction in magnetic resonance imaging (MRI) in spinal metastases.

Methods

3 phantoms and 50 patients with a corresponding total number of 80 spinal metastases were included in this study. Fast spin echo T1- and T2- weighted MRI with and without surface coil intensity correction was routinely performed for all phantoms and patients. Phantoms were evaluated by means of variance to mean ratio of signal intensity on both T1- and T2- weighted MRI obtained with and without surface coil intensity correction. Spinal metastases were evaluated by image quality scores; reading time per case on both T1- and T2- weighted MRI obtained with and without surface coil intensity correction.

Results

Spinal metastases were diagnosed more successfully on MRI with surface coil intensity correction than on MRI with conventional surface coil technique. The variance to mean ratio of signal intensity was 53.36% for original T1-weighted MRI and 53.58% for original T2-weighted MRI. The variance to mean ratio of signal intensity was reduced to 18.99% for T1-weighted MRI with surface coil intensity correction and 22.77% for T2-weighted MRI with surface coil intensity correction. The overall image quality scores (interface conspicuity of lesion and details of lesion) were significantly higher than those of the original MRI. The reading time per case was shorter for MRI with surface coil intensity correction than for MRI without surface coil intensity correction.

Conclusions

Phased-array surface coil intensity correction in MRIs of spinal metastases provides improvements in image quality that leads to more successfully detection and assessment of spinal metastases than original MRI.

Noise-Compensated, Bias-Corrected Diffusion Weighted Endorectal Magnetic Resonance Imaging via a Stochastically Fully-Connected Joint Conditional Random Field Model

Diffusion weighted magnetic resonance imaging (DW-MR) is a powerful tool in imaging-based prostate cancer screening and detection. Endorectal coils are commonly used in DW-MR imaging to improve the signal-to-noise ratio (SNR) of the acquisition, at the expense of significant intensity inhomogeneities (bias field) that worsens as we move away from the endorectal coil. The presence of bias field can have a significant negative impact on the accuracy of different image analysis tasks, as well as prostate tumor localization, thus leading to increased inter- and intra-observer variability. Retrospective bias correction approaches are introduced as a more efficient way of bias correction compared to the prospective methods such that they correct for both of the scanner and anatomy-related bias fields in MR imaging. Previously proposed retrospective bias field correction methods suffer from undesired noise amplification that can reduce the quality of bias-corrected DW-MR image. Here, we propose a unified data reconstruction approach that enables joint compensation of bias field as well as data noise in DW-MR imaging. The proposed noise-compensated, bias-corrected (NCBC) data reconstruction method takes advantage of a novel stochastically fully connected joint conditional random field (SFC-JCRF) model to mitigate the effects of data noise and bias field in the reconstructed MR data. The proposed NCBC reconstruction method was tested on synthetic DW-MR data, physical DW-phantom as well as real DW-MR data all acquired using endorectal MR coil. Both qualitative and quantitative analysis illustrated that the proposed NCBC method can achieve improved image quality when compared to other tested bias correction methods. As such, the proposed NCBC method may have potential as a useful retrospective approach for improving the consistency of image interpretations.

Multigrid Nonlocal Gaussian Mixture Model for Segmentation of Brain Tissues in Magnetic Resonance Images

We propose a novel segmentation method based on regional and nonlocal information to overcome the impact of image intensity inhomogeneities and noise in human brain magnetic resonance images. With the consideration of the spatial distribution of different tissues in brain images, our method does not need preestimation or precorrection procedures for intensity inhomogeneities and noise. A nonlocal information based Gaussian mixture model (NGMM) is proposed to reduce the effect of noise. To reduce the effect of intensity inhomogeneity, the multigrid nonlocal Gaussian mixture model (MNGMM) is proposed to segment brain MR images in each nonoverlapping multigrid generated by using a new multigrid generation method. Therefore the proposed model can simultaneously overcome the impact of noise and intensity inhomogeneity and automatically classify 2D and 3D MR data into tissues of white matter, gray matter, and cerebral spinal fluid. To maintain the statistical reliability and spatial continuity of the segmentation, a fusion strategy is adopted to integrate the clustering results from different grid. The experiments on synthetic and clinical brain MR images demonstrate the superior performance of the proposed model comparing with several state-of-the-art algorithms.

Segmentation and quantification of adipose tissue by magnetic resonance imaging

In this brief review, introductory concepts in animal and human adipose tissue segmentation using proton magnetic resonance imaging (MRI) and computed tomography are summarized in the context of obesity research. Adipose tissue segmentation and quantification using spin relaxation-based (e.g., T1-weighted, T2-weighted), relaxometry-based (e.g., T1-, T2-, T2*-mapping), chemical-shift selective, and chemical-shift encoded water-fat MRI pulse sequences are briefly discussed. The continuing interest to classify subcutaneous and visceral adipose tissue depots into smaller sub-depot compartments is mentioned. The use of a single slice, a stack of slices across a limited anatomical region, or a whole body protocol is considered. Common image post-processing steps and emerging atlas-based automated segmentation techniques are noted. Finally, the article identifies some directions of future research, including a discussion on the growing topic of brown adipose tissue and related segmentation considerations.

Three-dimensional mapping of the creatine kinase enzyme reaction rate in muscles of the lower leg

Phosphorus ((31) P) magnetization transfer (MT) techniques enable the non-invasive measurement of metabolic turnover rates of important enzyme-catalyzed reactions, such as the creatine kinase reaction (CK), a major transducing reaction involving adenosine triphosphate and phosphocreatine. Alteration in the kinetics of the CK reaction rate appears to play a central role in many disease states. In this study, we developed and implemented at ultra-high field (7T) a novel three-dimensional (31) P-MT imaging sequence that maps the kinetics of CK in the entire volume of the lower leg at relatively high resolution (0.52 mL voxel size), and within acquisition times that can be tolerated by patients (below 60 min). We tested the sequence on five healthy and two clinically diagnosed type 2 diabetic subjects. Overall, we obtained measurements that are in close agreement with measurements reported previously using spectroscopic methods. Importantly, our spatially resolved method allowed us to measure local CK reaction rate constants and metabolic fluxes in individual muscles in a non-invasive manner. Furthermore, it allowed us to detect variations of the CK rates of different muscles, which would not have been possible using unlocalized MRS methods. The results of this work suggest that 3D mapping of the CK reaction rates and metabolic fluxes can be achieved in the skeletal muscle in vivo at relatively high spatial resolution and with acquisition times well tolerated by patients. The ability to measure bioenergetics simultaneously in large areas of muscles will bring new insights into possible heterogeneous patterns of muscle metabolism associated with several diseases and serve as a valuable tool for monitoring the efficacy of interventions.

3D-mapping of phosphocreatine concentration in the human calf muscle at 7 T: comparison to 3 T

PURPOSE

The development and implementation of a spectrally selective 3D-Turbo Spin Echo sequence for quantitative mapping of phosphocreatine (PCr) concentration in different muscles of the lower leg of healthy volunteers both at 3 T and 7 T.

METHODS

Nine healthy volunteers were recruited, all of whom where scanned at 3 T and 7 T. Three dimensional PCr concentration maps were obtained after images were corrected for B1 inhomogeneities, T1 relaxation weighting, and partial volume of fatty tissue in the muscles. Two volunteers performed plantar flexions inside the magnet, and the oxidative capacity of their muscles was estimated.

RESULTS

Three dimensional PCr concentration maps were obtained, with full muscle coverage and nominal voxel size of 0.52 mL at both fields. At 7 T a 2.7-fold increase of signal-to-noise ratio was achieved compared to 3 T.

CONCLUSION

Imaging (31) P metabolites at 7 T allowed for significant increase in signal to noise ratio compared to imaging at 3 T, while quantification of the PCr concentration remained unaffected. The importance of such an increase in signal-to-noise ratio is 2-fold, first higher resolution images with reduced partial volume effects can be acquired, and second multiple measurements such as dynamic imaging of PCr post-exercise, (31) P magnetization transfer, or other (1) H measurements, can be acquired in a single imaging session.

Rapid quantification of muscle fat content and subcutaneous adipose tissue in fish using MRI

The potentiality of MRI to quantify fat content in flesh and subcutaneous fat in fish cutlets was investigated. Low measurement time was aimed at in a view to handling large number of samples needed in selective breeding programs for example. Results on fresh and frozen-thawed cutlets were compared to assess this way of conservation. As MRI generates unwanted spatial variations of the signal, a correction method was developed enabling the measurement on several cutlets simultaneously in less than 3 min per sample. For subcutaneous fat, the results were compared with vision measurements. High correlations between both techniques were found (R(2)=0.77 and 0.87 for the ventral and dorsal part). Fat in flesh was validated vs NMR measurements. No statistical difference was found between fresh and frozen-thawed cutlets. RMSE was respectively 0.8% and 0.89%. These results confirmed the potentiality of MRI for fat measurement in fish particularly for a large number of samples.

Sequential MR imaging of denervated and reinnervated skeletal muscle as correlated to functional outcome

PURPOSE

To prospectively assess the short inversion time inversion-recovery (STIR) magnetic resonance (MR) signal intensity changes of denervated and reinnervated skeletal muscle over time in clinical patients.

MATERIALS AND METHODS

This study was approved by the institutional review board, and informed consent was obtained from all patients. Twenty-three patients with complete traumatic transection of the median or ulnar nerve in the forearm were prospectively followed for 12 months after surgical nerve repair. STIR MR images of selected intrinsic hand muscles were obtained 1, 3, 6, 9, and 12 months after nerve repair, and signal intensities of denervated and reinnervated muscles were measured semiquantitatively. After 12 months, hand function was assessed. Signal intensity ratios were correlated to functional outcome with analysis of variance.

RESULTS

Of the 23 patients, 10 had good function recovery, while 13 had poor recovery. For the group with good function recovery, mean signal intensity ratios of 1.267 ± 0.060 (standard deviation), 1.357 ± 0.116, 1.297 ± 0.111, 1.205 ± 0.096, and 1.086 ± 0.104 were found at 1-, 3-, 6-, 9-, and 12-month follow-up, respectively. In the group with poor recovery, mean signal intensity ratios of 1.299 ± 0.056, 1.377 ± 0.094, 1.419 ± 0.117, 1.398 ± 0.111, and 1.342 ± 0.095 were found at 1-, 3-, 6-, 9-, and 12-month follow-up, respectively. Comparison of the group with poor function recovery and the group with good function recovery showed significant differences at 6-, 9-, and 12-month follow-up (P = .035, P = .001, and P < .001, respectively), with normalizing signal intensities in the group with good function recovery and sustained high signal intensity in the group with poor function recovery.

CONCLUSION

STIR MR imaging can be used to differentiate between denervated and reinnervated muscles for at least 12 months after nerve transection.

An improved level set method for brain MR images segmentation and bias correction

Intensity inhomogeneities cause considerable difficulty in the quantitative analysis of magnetic resonance (MR) images. Thus, bias field estimation is a necessary step before quantitative analysis of MR data can be undertaken. This paper presents a variational level set approach to bias correction and segmentation for images with intensity inhomogeneities. Our method is based on an observation that intensities in a relatively small local region are separable, despite of the inseparability of the intensities in the whole image caused by the overall intensity inhomogeneity. We first define a localized K-means-type clustering objective function for image intensities in a neighborhood around each point. The cluster centers in this objective function have a multiplicative factor that estimates the bias within the neighborhood. The objective function is then integrated over the entire domain to define the data term into the level set framework. Our method is able to capture bias of quite general profiles. Moreover, it is robust to initialization, and thereby allows fully automated applications. The proposed method has been used for images of various modalities with promising results.

A review of methods for correction of intensity inhomogeneity in MRI

Medical image acquisition devices provide a vast amount of anatomical and functional information, which facilitate and improve diagnosis and patient treatment, especially when supported by modern quantitative image analysis methods. However, modality specific image artifacts, such as the phenomena of intensity inhomogeneity in magnetic resonance images (MRI), are still prominent and can adversely affect quantitative image analysis. In this paper, numerous methods that have been developed to reduce or eliminate intensity inhomogeneities in MRI are reviewed. First, the methods are classified according to the inhomogeneity correction strategy. Next, different qualitative and quantitative evaluation approaches are reviewed. Third, 60 relevant publications are categorized according to several features and analyzed so as to reveal major trends, popularity, evaluation strategies and applications. Finally, key evaluation issues and future development of the inhomogeneity correction field, supported by the results of the analysis, are discussed.

MRI intensity nonuniformity correction using simultaneously spatial and gray-level histogram information

A novel approach for correcting intensity nonuniformity in magnetic resonance imaging (MRI) is presented. This approach is based on the simultaneous use of spatial and gray-level histogram information. Spatial information about intensity nonuniformity is obtained using cubic B-spline smoothing. Gray-level histogram information of the image corrupted by intensity nonuniformity is exploited from a frequential point of view. The proposed correction method is illustrated using both physical phantom and human brain images. The results are consistent with theoretical prediction, and demonstrate a new way of dealing with intensity nonuniformity problems. They are all the more significant as the ground truth on intensity nonuniformity is unknown in clinical images.

Noninvasive characterization of gonad maturation and determination of the sex of Pacific oysters by MRI

The aim of this study was to test the ability of magnetic resonance imaging (MRI) technique to characterize gonad development and to determine the sex of live Pacific oysters through their shells. A preliminary nuclear magnetic resonance (NMR) relaxometry study was conducted to characterize T1 and T2 NMR relaxation parameters for the main oyster organs. This showed that T1-weighted MRI sequences were most appropriate to optimize contrasts between tissues in images. The results showed that gray levels of gonads in images acquired with gradient-echo sequence were variably affected by T2* weighting effect. However, the ovaries systematically gave a hypersignal in spin-echo T1-weighted images, and stack histograms of female oysters showed a peak well separated from that of male oysters. An automated method is proposed to quantify the development of oysters and their gonad maturation and to identify their sex.

A Review on MR Image Intensity Inhomogeneity Correction

Intensity inhomogeneity (IIH) is often encountered in MR imaging, and a number of techniques have been devised to correct this artifact. This paper attempts to review some of the recent developments in the mathematical modeling of IIH field. Low-frequency models are widely used, but they tend to corrupt the low-frequency components of the tissue. Hypersurface models and statistical models can be adaptive to the image and generally more stable, but they are also generally more complex and consume more computer memory and CPU time. They are often formulated together with image segmentation within one framework and the overall performance is highly dependent on the segmentation process. Beside these three popular models, this paper also summarizes other techniques based on different principles. In addition, the issue of quantitative evaluation and comparative study are discussed.

Method to correct intensity inhomogeneity in MR images for atherosclerosis characterization

We are developing methods to characterize atherosclerotic disease in human carotid arteries using multiple MR images having different contrast mechanisms (T1W, T2W, PDW). To enable the use of voxel gray values for interpretation of disease, we created a new method, local entropy minimization with a bicubic spline model (LEMS), to correct the severe (approximately 80%) intensity inhomogeneity that arises from the surface coil array. This entropy-based method does not require classification and robustly addresses some problems that are more severe than those found in brain imaging, including noise, steep bias field, sensitivity of artery wall voxels to edge artifacts, and signal voids near the artery wall. Validation studies were performed on a synthetic digital phantom with realistic intensity inhomogeneity, a physical phantom roughly mimicking the neck, and patient carotid artery images. We compared LEMS to a modified fuzzy c-means segmentation based method (mAFCM), and a linear filtering method (LINF). Following LEMS correction, skeletal muscles in patient images were relatively isointense across the field of view. In the physical phantom, LEMS reduced the variation in the image to 1.9% and across the vessel wall region to 2.5%, a value which should be sufficient to distinguish plaque tissue types, based on literature measurements. In conclusion, we believe that the correction method shows promise for aiding human and computerized tissue classification from MR signal intensities.

Intensity non-uniformity correction in MRI: existing methods and their validation

Magnetic resonance imaging is a popular and powerful non-invasive imaging technique. Automated analysis has become mandatory to efficiently cope with the large amount of data generated using this modality. However, several artifacts, such as intensity non-uniformity, can degrade the quality of acquired data. Intensity non-uniformity consists in anatomically irrelevant intensity variation throughout data. It can be induced by the choice of the radio-frequency coil, the acquisition pulse sequence and by the nature and geometry of the sample itself. Numerous methods have been proposed to correct this artifact. In this paper, we propose an overview of existing methods. We first sort them according to their location in the acquisition/processing pipeline. Sorting is then refined based on the assumptions those methods rely on. Next, we present the validation protocols used to evaluate these different correction schemes both from a qualitative and a quantitative point of view. Finally, availability and usability of the presented methods is discussed.

Determination of the lean meat percentage of pig carcasses using magnetic resonance imaging

The purpose of Workpackage 3 of the European Eupigclass project was to test indirect methods of measuring the lean meat percentage of a carcass that would be less costly, at least as accurate and more consistent than dissection. Magnetic resonance imaging was one of the three indirect methods tested to measure the lean meat weight and the lean meat percentage of pig carcasses, the other methods being X-ray CT and vision techniques. One hundred and twenty carcasses from three different genotypes and from both sexes were slaughtered. The left parts of the carcasses were fully dissected and the right parts were investigated with an indirect method using a 1.5T MRI system. The acquisition protocol was chosen to give an optimized contrast between fat and muscle tissues. Two different approaches, image segmentation and PLS regression, were used to extract information from the images. Automatic image segmentation was performed to quantify the volume of muscle in the images and gave a standard error of prediction using a linear regression with the dissection of the left half carcasses of 586g and 1.10% for lean meat weight and lean meat percentage, respectively. PLS regression using the signal intensities histograms gave an estimation error of 465g for lean meat weight. These results showed that MRI could be used in place of full dissection for authorizing and monitoring classification equipment of pig carcasses.