Joint relaxation-diffusion measurements can provide new insight about the tissue microstructural properties. Most recent methods have focused on inverting the Laplace transform to recover the joint distribution of relaxation-diffusion. However, as is well-known, this problem is notoriously ill-posed and numerically unstable. In this work, we address this issue by directly computing the joint moments of transverse relaxation rate and diffusivity, which can be robustly estimated. To zoom into different parts of the joint distribution, we further enhance our method by applying multiplicative filters to the joint probability density function of relaxation and diffusion and compute the corresponding moments. We propose an approach to use these moments to compute several novel scalar indices to characterize specific properties of the underlying tissue microstructure. Furthermore, for the first time, we propose an algorithm to estimate diffusion signals that are independent of echo time based on the moments of the marginal probability density function of diffusion. We demonstrate its utility in extracting tissue information not contaminated with multiple intra-voxel relaxation rates. We compare the performance of four types of filters that zoom into tissue components with different relaxation and diffusion properties and demonstrate it on an in-vivo human dataset. Experimental results show that these filters are able to characterize heterogeneous tissue microstructure. Moreover, the filtered diffusion signals are also able to distinguish fiber bundles with similar orientations but different relaxation rates. The proposed method thus allows to characterize the neural microstructure information in a robust and unique manner not possible using existing techniques.
PURPOSE: We present SlicerDMRI, an open-source software suite that enables research using diffusion magnetic resonance imaging (dMRI), the only modality that can map the white matter connections of the living human brain. SlicerDMRI enables analysis and visualization of dMRI data and is aimed at the needs of clinical research users. SlicerDMRI is built upon and deeply integrated with 3D Slicer, a National Institutes of Health-supported open-source platform for medical image informatics, image processing, and three-dimensional visualization. Integration with 3D Slicer provides many features of interest to cancer researchers, such as real-time integration with neuronavigation equipment, intraoperative imaging modalities, and multimodal data fusion. One key application of SlicerDMRI is in neurosurgery research, where brain mapping using dMRI can provide patient-specific maps of critical brain connections as well as insight into the tissue microstructure that surrounds brain tumors. PATIENTS AND METHODS: In this article, we focus on a demonstration of SlicerDMRI as an informatics tool to enable end-to-end dMRI analyses in two retrospective imaging data sets from patients with high-grade glioma. Analyses demonstrated here include conventional diffusion tensor analysis, advanced multifiber tractography, automated identification of critical fiber tracts, and integration of multimodal imagery with dMRI. RESULTS: We illustrate the ability of SlicerDMRI to perform both conventional and advanced dMRI analyses as well as to enable multimodal image analysis and visualization. We provide an overview of the clinical rationale for each analysis along with pointers to the SlicerDMRI tools used in each. CONCLUSION: SlicerDMRI provides open-source and clinician-accessible research software tools for dMRI analysis. SlicerDMRI is available for easy automated installation through the 3D Slicer Extension Manager.
OBJECTIVE: Accurate biopsy sampling of the suspected lesions is critical for the diagnosis and clinical management of prostate cancer. Transperineal in-bore MRI-guided prostate biopsy (tpMRgBx) is a targeted biopsy technique that was shown to be safe, efficient, and accurate. Our goal was to develop an open source software platform to support evaluation, refinement, and translation of this biopsy approach. METHODS: We developed SliceTracker, a 3D Slicer extension to support tpMRgBx. We followed modular design of the implementation to enable customization of the interface and interchange of image segmentation and registration components to assess their effect on the processing time, precision, and accuracy of the biopsy needle placement. The platform and supporting documentation were developed to enable the use of software by an operator with minimal technical training to facilitate translation. Retrospective evaluation studied registration accuracy, effect of the prostate segmentation approach, and re-identification time of biopsy targets. Prospective evaluation focused on the total procedure time and biopsy targeting error (BTE). RESULTS: Evaluation utilized data from 73 retrospective and ten prospective tpMRgBx cases. Mean landmark registration error for retrospective evaluation was 1.88 ± 2.63 mm, and was not sensitive to the approach used for prostate gland segmentation. Prospectively, we observed target re-identification time of 4.60 ± 2.40 min and BTE of 2.40 ± 0.98 mm. CONCLUSION: SliceTracker is modular and extensible open source platform for supporting image processing aspects of the tpMRgBx procedure. It has been successfully utilized to support clinical research procedures at our site.
OBJECTIVE: The cochlear nucleus (CN) is the target of the auditory brainstem implant (ABI). Most ABI candidates have Neurofibromatosis Type 2 (NF2) and distorted brainstem anatomy from bilateral vestibular schwannomas. The CN is difficult to characterize as routine structural MRI does not resolve detailed anatomy. We hypothesize that diffusion tensor imaging (DTI) enables both in vivo localization and quantitative measurements of CN morphology. STUDY DESIGN: We analyzed 7 Tesla (T) DTI images of 100 subjects (200 CN) and relevant anatomic structures using an MRI brainstem atlas with submillimetric (50 μm) resolution. SETTING: Tertiary referral center. PATIENTS: Young healthy normal hearing adults. INTERVENTION: Diagnostic. MAIN OUTCOME MEASURES: Diffusion scalar measures such as fractional anisotropy (FA), mean diffusivity (MD), mode of anisotropy (Mode), principal eigenvectors of the CN, and the adjacent inferior cerebellar peduncle (ICP). RESULTS: The CN had a lamellar structure and ventral-dorsal fiber orientation and could be localized lateral to the inferior cerebellar peduncle (ICP). This fiber orientation was orthogonal to tracts of the adjacent ICP where the fibers run mainly caudal-rostrally. The CN had lower FA compared to the medial aspect of the ICP (0.44 ± 0.09 vs. 0.64 ± 0.08, p < 0.001). CONCLUSIONS: 7T DTI enables characterization of human CN morphology and neuronal substructure. An ABI array insertion vector directed more caudally would better correspond to the main fiber axis of CN. State-of-the-art DTI has implications for ABI preoperative planning and future image guidance-assisted placement of the electrode array.
BACKGROUND: The trigeminal nerve (TGN) is the largest cranial nerve and can be involved in multiple inflammatory, compressive, ischemic or other pathologies. Currently, imaging-based approaches to identify the TGN mostly rely on T2-weighted magnetic resonance imaging (MRI), which provides localization of the cisternal portion of the TGN where the contrast between nerve and cerebrospinal fluid (CSF) is high enough to allow differentiation. The course of the TGN within the brainstem as well as anterior to the cisternal portion, however, is more difficult to display on traditional imaging sequences. An advanced imaging technique, diffusion MRI (dMRI), enables tracking of the trajectory of TGN fibers and has the potential to visualize anatomical regions of the TGN not seen on T2-weighted imaging. This may allow a more comprehensive assessment of the nerve in the context of pathology. To date, most work in TGN tracking has used clinical dMRI acquisitions with a b-value of 1000 s/mm2 and conventional diffusion tensor MRI (DTI) tractography methods. Though higher b-value acquisitions and multi-tensor tractography methods are known to be beneficial for tracking brain white matter fiber tracts, there have been no studies conducted to evaluate the performance of these advanced approaches on nerve tracking of the TGN, in particular on tracking different anatomical regions of the TGN. OBJECTIVE: We compare TGN tracking performance using dMRI data with different b-values, in combination with both single- and multi-tensor tractography methods. Our goal is to assess the advantages and limitations of these different strategies for identifying the anatomical regions of the TGN. METHODS: We proposed seven anatomical rating criteria including true and false positive structures, and we performed an expert rating study of over 1000 TGN visualizations, as follows. We tracked the TGN using high-quality dMRI data from 100 healthy adult subjects from the Human Connectome Project (HCP). TGN tracking performance was compared across dMRI acquisitions with b = 1000 s/mm2, b = 2000 s/mm2 and b = 3000 s/mm2, using single-tensor (1T) and two-tensor (2T) unscented Kalman filter (UKF) tractography. This resulted in a total of six tracking strategies. The TGN was identified using an anatomical region-of-interest (ROI) selection approach. First, in a subset of the dataset we identified ROIs that provided good TGN tracking performance across all tracking strategies. Using these ROIs, the TGN was then tracked in all subjects using the six tracking strategies. An expert rater (GX) visually assessed and scored each TGN based on seven anatomical judgment criteria. These criteria included the presence of multiple expected anatomical segments of the TGN (true positive structures), specifically branch-like structures, cisternal portion, mesencephalic trigeminal tract, and spinal cord tract of the TGN. False positive criteria included the presence of any fibers entering the temporal lobe, the inferior cerebellar peduncle, or the middle cerebellar peduncle. Expert rating scores were analyzed to compare TGN tracking performance across the six tracking strategies. Intra- and inter-rater validation was performed to assess the reliability of the expert TGN rating result. RESULTS: The TGN was selected using two anatomical ROIs (Meckel's Cave and cisternal portion of the TGN). The two-tensor tractography method had significantly better performance on identifying true positive structures, while generating more false positive streamlines in comparison to the single-tensor tractography method. TGN tracking performance was significantly different across the three b-values for almost all structures studied. Tracking performance was reported in terms of the percentage of subjects achieving each anatomical rating criterion. Tracking of the cisternal portion and branching structure of the TGN was generally successful, with the highest performance of over 98% using two-tensor tractography and b = 1000 or b = 2000. However, tracking the smaller mesencephalic and spinal cord tracts of the TGN was quite challenging (highest performance of 37.5% and 57.07%, using two-tensor tractography with b = 1000 and b = 2000, respectively). False positive connections to the temporal lobe (over 38% of subjects for all strategies) and cerebellar peduncles (100% of subjects for all strategies) were prevalent. High joint probability of agreement was obtained in the inter-rater (on average 83%) and intra-rater validation (on average 90%), showing a highly reliable expert rating result. CONCLUSIONS: Overall, the results of the study suggest that researchers and clinicians may benefit from tailoring their acquisition and tracking methodology to the specific anatomical portion of the TGN that is of the greatest interest. For example, tracking of branching structures and TGN-T2 overlap can be best achieved with a two-tensor model and an acquisition using b = 1000 or b = 2000. In general, b = 1000 and b = 2000 acquisitions provided the best-rated tracking results. Further research is needed to improve both sensitivity and specificity of the depiction of the TGN anatomy using dMRI.
Cerebrospinal fluid partial volume effect is a known bias in the estimation of Diffusion Tensor Imaging (DTI) parameters from diffusion MRI data. The Free-Water Imaging model for diffusion MRI data adds a second compartment to the DTI model, which explicitly accounts for the signal contribution of extracellular free-water, such as cerebrospinal fluid. As a result the DTI parameters obtained through the free-water model are corrected for partial volume effects, and thus better represent tissue microstructure. In addition, the model estimates the fractional volume of free-water, and can be used to monitor changes in the extracellular space. Under certain assumptions, the model can be estimated from single-shell diffusion MRI data. However, by using data from multi-shell diffusion acquisitions, these assumptions can be relaxed, and the fit becomes more robust. Nevertheless, fitting the model to multi-shell data requires high computational cost, with a non-linear iterative minimization, which has to be initialized close enough to the global minimum to avoid local minima and to robustly estimate the model parameters. Here we investigate the properties of the main initialization approaches that are currently being used, and suggest new fast approaches to improve the initial estimates of the model parameters. We show that our proposed approaches provide a fast and accurate initial approximation of the model parameters, which is very close to the final solution. We demonstrate that the proposed initializations improve the final outcome of non-linear model fitting.
Neuroimaging studies typically adopt a common feature space for all data, which may obscure aspects of neuroanatomy only observable in subsets of a population, e.g. cortical folding patterns unique to individuals or shared by close relatives. Here, we propose to model individual variability using a distinctive keypoint signature: a set of unique, localized patterns, detected automatically in each image by a generic saliency operator. The similarity of an image pair is then quantified by the proportion of keypoints they share using a novel Jaccard-like measure of set overlap. Experiments demonstrate the keypoint method to be highly efficient and accurate, using a set of 7536 T1-weighted MRIs pooled from four public neuroimaging repositories, including twins, non-twin siblings, and 3334 unique subjects. All same-subject image pairs are identified by a similarity threshold despite confounds including aging and neurodegenerative disease progression. Outliers reveal previously unknown data labeling inconsistencies, demonstrating the usefulness of the keypoint signature as a computational tool for curating large neuroimage datasets.
Motion is a major confound in diffusion-weighted imaging (DWI) in the body, and it is a common cause of image artefacts. The effects are particularly severe in cardiac applications, due to the nonrigid cyclical deformation of the myocardium. Spin echo-based DWI commonly employs gradient moment-nulling techniques to desensitise the acquisition to velocity and acceleration, ie, nulling gradient moments up to the 2nd order (M2-nulled). However, current M2-nulled DWI scans are limited to encode diffusion along a single direction at a time. We propose a method for designing b-tensors of arbitrary shapes, including planar, spherical, prolate and oblate tensors, while nulling gradient moments up to the 2nd order and beyond. The design strategy comprises initialising the diffusion encoding gradients in two encoding blocks about the refocusing pulse, followed by appropriate scaling and rotation, which further enables nulling undesired effects of concomitant gradients. Proof-of-concept assessment of in vivo mean diffusivity (MD) was performed using linear and spherical tensor encoding (LTE and STE, respectively) in the hearts of five healthy volunteers. The results of the M2-nulled STE showed that (a) the sequence was robust to cardiac motion, and (b) MD was higher than that acquired using standard M2-nulled LTE, where diffusion-weighting was applied in three orthogonal directions, which may be attributed to the presence of restricted diffusion and microscopic diffusion anisotropy. Provided adequate signal-to-noise ratio, STE could significantly shorten estimation of MD compared with the conventional LTE approach. Importantly, our theoretical analysis and the proposed gradient waveform design may be useful in microstructure imaging beyond diffusion tensor imaging where the effects of motion must be suppressed.
Posterior circulation ischemic stroke (PCiS) constitutes 20-30% of ischemic stroke cases. Detailed information about differences between PCiS and anterior circulation ischemic stroke (ACiS) remains scarce. Such information might guide clinical decision making and prevention strategies. We studied risk factors and ischemic stroke subtypes in PCiS vs. ACiS and lesion location on magnetic resonance imaging (MRI) in PCiS.
Out of 3,301 MRIs from 12 sites in the National Institute of Neurological Disorders and Stroke (NINDS) Stroke Genetics Network (SiGN), we included 2,381 cases with acute DWI lesions. The definition of ACiS or PCiS was based on lesion location. We compared the groups using Chi-squared and logistic regression.
PCiS occurred in 718 (30%) patients and ACiS in 1663 (70%). Diabetes and male sex were more common in PCiS vs. ACiS (diabetes 27% vs. 23%, p < 0.05; male sex 68% vs. 58%, p < 0.001). Both were independently associated with PCiS (diabetes, OR = 1.29; 95% CI 1.04-1.61; male sex, OR = 1.46; 95% CI 1.21-1.78). ACiS more commonly had large artery atherosclerosis (25% vs. 20%, p < 0.01) and cardioembolic mechanisms (17% vs. 11%, p < 0.001) compared to PCiS. Small artery occlusion was more common in PCiS vs. ACiS (20% vs. 14%, p < 0.001). Small artery occlusion accounted for 47% of solitary brainstem infarctions.
Ischemic stroke subtypes differ between the two phenotypes. Diabetes and male sex have a stronger association with PCiS than ACiS. Definitive MRI-based PCiS diagnosis aids etiological investigation and contributes additional insights into specific risk factors and mechanisms of injury in PCiS.
BACKGROUND: Auditory verbal hallucinations (AVH) are a cardinal feature of schizophrenia, but they can also appear in otherwise healthy individuals. Imaging studies implicate language networks in the generation of AVH; however, it remains unclear if alterations reflect biologic substrates of AVH, irrespective of diagnostic status, age, or illness-related factors. We applied multimodal imaging to identify AVH-specific pathology, evidenced by overlapping gray or white matter deficits between schizophrenia patients and healthy voice-hearers. METHODS: Diffusion-weighted and T1-weighted magnetic resonance images were acquired in 35 schizophrenia patients with AVH (SCZ-AVH), 32 healthy voice-hearers (H-AVH), and 40 age- and sex-matched controls without AVH. White matter fractional anisotropy (FA) and gray matter thickness (GMT) were computed for each region comprising ICBM-DTI and Desikan-Killiany atlases, respectively. Regions were tested for significant alterations affecting both SCZ-AVH and H-AVH groups, relative to controls. RESULTS: Compared with controls, the SCZ-AVH showed widespread FA and GMT reductions; but no significant differences emerged between H-AVH and control groups. While no overlapping pathology appeared in the overall study groups, younger (<40 years) H-AVH and SCZ-AVH subjects displayed overlapping FA deficits across four regions (p < 0.05): the genu and splenium of the corpus callosum, as well as the anterior limbs of the internal capsule. Analyzing these regions with free-water imaging ascribed overlapping FA abnormalities to tissue-specific anisotropy changes. CONCLUSIONS: We identified white matter pathology associated with the presence of AVH, independent of diagnostic status. However, commonalities were constrained to younger and more homogenous groups, after reducing pathologic variance associated with advancing age and chronicity effects.
We introduce an approach for image segmentation based on sparse correspondences between keypoints in testing and training images. Keypoints represent automatically identified distinctive image locations, where each keypoint correspondence suggests a transformation between images. We use these correspondences to transfer label maps of entire organs from the training images to the test image. The keypoint transfer algorithm includes three steps: (i) keypoint matching, (ii) voting-based keypoint labeling, and (iii) keypoint-based probabilistic transfer of organ segmentations. We report segmentation results for abdominal organs in whole-body CT and MRI, as well as in contrast-enhanced CT and MRI. Our method offers a speed-up of about three orders of magnitude in comparison to common multi-atlas segmentation, while achieving an accuracy that compares favorably. Moreover, keypoint transfer does not require the registration to an atlas or a training phase. Finally, the method allows for the segmentation of scans with highly variable field-of-view.
PURPOSE: Multimodal imaging plays a key role in patient assessment and treatment planning in liver radioembolization. It will reach its full potential for convenient use in combination with deformable image registration methods. A registration framework is proposed for multimodal liver image registration of multi-phase CT, contrast-enhanced late-phase T1, T2, and DWI MRI sequences. METHODS: A chain of four pair-wise image registrations based on a variational registration framework using normalized gradient fields as distance measure and curvature regularization is introduced. A total of 103 cases of 35 patients was evaluated based on anatomical landmarks and deformation characteristics. RESULTS: Good anatomical correspondence and physical plausibility of the deformation fields were attained. The global mean landmark errors vary from 3.20 to 5.36 mm, strongly influenced by low resolved images in z-direction. Moderate volume changes are indicated by mean minimum and maximum Jacobian determinants of 0.44 up to 1.88. No deformation foldings were detected. The mean average divergence of the deformation fields range from 0.08 to 0.16 and the mean harmonic energies vary from 0.08 to 0.58. CONCLUSION: The proposed registration solutions enable the combined use of information from multimodal imaging and provide an excellent basis for patient assessment and primary planning for liver radioembolization.
We introduce Disease Knowledge Transfer (DKT), a novel technique for transferring biomarker information between related neurodegenerative diseases. DKT infers robust multimodal biomarker trajectories in rare neurodegenerative diseases even when only limited, unimodal data is available, by transferring information from larger multimodal datasets from common neurodegenerative diseases. DKT is a joint-disease generative model of biomarker progressions, which exploits biomarker relationships that are shared across diseases. Our proposed method allows, for the first time, the estimation of plausible biomarker trajectories in Posterior Cortical Atrophy (PCA), a rare neurodegenerative disease where only unimodal MRI data is available. For this we train DKT on a combined dataset containing subjects with two distinct diseases and sizes of data available: 1) a larger, multimodal typical AD (tAD) dataset from the TADPOLE Challenge, and 2) a smaller unimodal Posterior Cortical Atrophy (PCA) dataset from the Dementia Research Centre (DRC), for which only a limited number of Magnetic Resonance Imaging (MRI) scans are available. Although validation is challenging due to lack of data in PCA, we validate DKT on synthetic data and two patient datasets (TADPOLE and PCA cohorts), showing it can estimate the ground truth parameters in the simulation and predict unseen biomarkers on the two patient datasets. While we demonstrated DKT on Alzheimer's variants, we note DKT is generalisable to other forms of related neurodegenerative diseases. Source code for DKT is available online: https://github.com/mrazvan22/dkt.
The performance and diagnostic utility of magnetic resonance imaging (MRI) in pregnancy is fundamentally constrained by fetal motion. Motion of the fetus, which is unpredictable and rapid on the scale of conventional imaging times, limits the set of viable acquisition techniques to single-shot imaging with severe compromises in signal-to-noise ratio and diagnostic contrast, and frequently results in unacceptable image quality. Surprisingly little is known about the characteristics of fetal motion during MRI and here we propose and demonstrate methods that exploit a growing repository of MRI observations of the gravid abdomen that are acquired at low spatial resolution but relatively high temporal resolution and over long durations (10-30 minutes). We estimate fetal pose per frame in MRI volumes of the pregnant abdomen via deep learning algorithms that detect key fetal landmarks. Evaluation of the proposed method shows that our framework achieves quantitatively an average error of 4.47 mm and 96.4% accuracy (with error less than 10 mm). Fetal pose estimation in MRI time series yields novel means of quantifying fetal movements in health and disease, and enables the learning of kinematic models that may enhance prospective mitigation of fetal motion artifacts during MRI acquisition.
We propose and demonstrate a joint model of anatomical shapes, image features and clinical indicators for statistical shape modeling and medical image analysis. The key idea is to employ a copula model to separate the joint dependency structure from the marginal distributions of variables of interest. This separation provides flexibility on the assumptions made during the modeling process. The proposed method can handle binary, discrete, ordinal and continuous variables. We demonstrate a simple and efficient way to include binary, discrete and ordinal variables into the modeling. We build Bayesian conditional models based on observed partial clinical indicators, features or shape based on Gaussian processes capturing the dependency structure. We apply the proposed method on a stroke dataset to jointly model the shape of the lateral ventricles, the spatial distribution of the white matter hyperintensity associated with periventricular white matter disease, and clinical indicators. The proposed method yields interpretable joint models for data exploration and patient-specific statistical shape models for medical image analysis.
We present a volumetric mesh-based algorithm for flattening the placenta to a canonical template to enable effective visualization of local anatomy and function. Monitoring placental function promises to support pregnancy assessment and to improve care outcomes. We aim to alleviate visualization and interpretation challenges presented by the shape of the placenta when it is attached to the curved uterine wall. To do so, we flatten the volumetric mesh that captures placental shape to resemble the well-studied shape. We formulate our method as a map from the shape to a flattened template that minimizes the symmetric Dirichlet energy to control distortion throughout the volume. Local injectivity is enforced via constrained line search during gradient descent. We evaluate the proposed method on 28 placenta shapes extracted from MRI images in a clinical study of placental function. We achieve sub-voxel accuracy in mapping the boundary of the placenta to the template while successfully controlling distortion throughout the volume. We illustrate how the resulting mapping of the placenta enhances visualization of placental anatomy and function. Our implementation is freely available at https://github.com/mabulnaga/placenta-flattening.
The Alzheimer's Disease Prediction Of Longitudinal Evolution (TADPOLE) Challenge compares the performance of algorithms at predicting the future evolution of individuals at risk of Alzheimer's disease. TADPOLE Challenge participants train their models and algorithms on historical data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) study. Participants are then required to make forecasts of three key outcomes for ADNI-3 rollover participants: clinical diagnosis, Alzheimer's Disease Assessment Scale Cognitive Subdomain (ADAS-Cog 13), and total volume of the ventricles - which are then compared with future measurements. Strong points of the challenge are that the test data did not exist at the time of forecasting (it was acquired afterwards), and that it focuses on the challenging problem of cohort selection for clinical trials by identifying fast progressors. The submission phase of TADPOLE was open until 15 November 2017; since then data has been acquired until April 2019 from 219 subjects with 223 clinical visits and 150 Magnetic Resonance Imaging (MRI) scans, which was used for the evaluation of the participants' predictions. Thirty-three teams participated with a total of 92 submissions. No single submission was best at predicting all three outcomes. For diagnosis prediction, the best forecast (team Frog), which was based on gradient boosting, obtained a multiclass area under the receiver-operating curve (MAUC) of 0.931, while for ventricle prediction the best forecast (team ), which was based on disease progression modelling and spline regression, obtained mean absolute error of 0.41% of total intracranial volume (ICV). For ADAS-Cog 13, no forecast was considerably better than the benchmark mixed effects model ( ), provided to participants before the submission deadline. Further analysis can help understand which input features and algorithms are most suitable for Alzheimer's disease prediction and for aiding patient stratification in clinical trials. The submission system remains open via the website: https://tadpole.grand-challenge.org/.
Probabilistic atlas priors have been commonly used to derive adaptive and robust brain MRI segmentation algorithms. Widely-used neuroimage analysis pipelines rely heavily on these techniques, which are often computationally expensive. In contrast, there has been a recent surge of approaches that leverage deep learning to implement segmentation tools that are computationally efficient at test time. However, most of these strategies rely on learning from manually annotated images. These supervised deep learning methods are therefore sensitive to the intensity profiles in the training dataset. To develop a deep learning-based segmentation model for a new image dataset (e.g., of different contrast), one usually needs to create a new labeled training dataset, which can be prohibitively expensive, or rely on suboptimal adaptation or augmentation approaches. In this paper, we propose an alternative strategy that combines a conventional probabilistic atlas-based segmentation with deep learning, enabling one to train a segmentation model for new MRI scans without the need for any manually segmented images. Our experiments include thousands of brain MRI scans and demonstrate that the proposed method achieves good accuracy for a brain MRI segmentation task for different MRI contrasts, requiring only approximately 15 seconds at test time on a GPU.
We present a deep learning tractography segmentation method that allows fast and consistent white matter fiber tract identification across healthy and disease populations and across multiple diffusion MRI (dMRI) acquisitions. We create a large-scale training tractography dataset of 1 million labeled fiber samples (54 anatomical tracts are included). To discriminate between fibers from different tracts, we propose a novel 2D multi-channel feature descriptor (FiberMap) that encodes spatial coordinates of points along each fiber. We learn a CNN tract classification model based on FiberMap and obtain a high tract classification accuracy of 90.99%. The method is evaluated on a test dataset of 374 dMRI scans from three independently acquired populations across health conditions (healthy control, neuropsychiatric disorders, and brain tumor patients). We perform comparisons with two state-of-the-art white matter tract segmentation methods. Experimental results show that our method obtains a highly consistent segmentation result, where over 99% of the fiber tracts are successfully detected across all subjects under study, most importantly, including patients with space occupying brain tumors. The proposed method leverages deep learning techniques and provides a much faster and more efficient tool for large data analysis than methods using traditional machine learning techniques.
Cerebral microbleeds (CMBs), a common manifestation of mild traumatic brain injury (mTBI), have been sporadically implicated in the neurocognitive deficits of mTBI victims but their clinical significance has not been established adequately. Here we investigate the longitudinal effects of post-mTBI CMBs upon the fractional anisotropy (FA) of white matter (WM) in 21 older mTBI patients across the first ~6 months post-injury. CMBs were segmented automatically from susceptibility-weighted imaging (SWI) by leveraging the intensity gradient properties of SWI to identify CMB-related hypointensities using gradient-based edge detection. A detailed diffusion magnetic resonance imaging (dMRI) atlas of WM was used to segment and cluster tractography streamlines whose prototypes were then identified. The correlation coefficient was calculated between (A) FA values at vertices along streamline prototypes and (B) topological (along-streamline) distances between these vertices and the nearest CMB. Across subjects, the CMB identification approach achieved a sensitivity of 97.1% ± 4.7% and a precision of 72.4% ± 11.0% across subjects. The correlation coefficient was found to be negative and, additionally, statistically significant for 12.3% ± 3.5% of WM clusters (p <; 0.05, corrected), whose FA was found to decrease, on average, by 11.8% ± 5.3% across the first 6 months post-injury. These results suggest that CMBs can be associated with deleterious effects upon peri-lesional WM and highlight the vulnerability of older mTBI patients to neurovascular injury.