Automatic Segmentation of the Spinal Cord and Spinal Canal Coupled With Vertebral Labeling

Neuroimage analysis using artificial intelligence approaches: a systematic review

In the contemporary era, artificial intelligence (AI) has undergone a transformative evolution, exerting a profound influence on neuroimaging data analysis. This development has significantly elevated our comprehension of intricate brain functions. This study investigates the ramifications of employing AI techniques on neuroimaging data, with a specific objective to improve diagnostic capabilities and contribute to the overall progress of the field. A systematic search was conducted in prominent scientific databases, including PubMed, IEEE Xplore, and Scopus, meticulously curating 456 relevant articles on AI-driven neuroimaging analysis spanning from 2013 to 2023. To maintain rigor and credibility, stringent inclusion criteria, quality assessments, and precise data extraction protocols were consistently enforced throughout this review. Following a rigorous selection process, 104 studies were selected for review, focusing on diverse neuroimaging modalities with an emphasis on mental and neurological disorders. Among these, 19.2% addressed mental illness, and 80.7% focused on neurological disorders. It is found that the prevailing clinical tasks are disease classification (58.7%) and lesion segmentation (28.9%), whereas image reconstruction constituted 7.3%, and image regression and prediction tasks represented 9.6%. AI-driven neuroimaging analysis holds tremendous potential, transforming both research and clinical applications. Machine learning and deep learning algorithms outperform traditional methods, reshaping the field significantly.

A deep learning approach for cervical cord injury severity determination through axial and sagittal magnetic resonance imaging segmentation and classification

STUDY DESIGN

Cross-sectional Database Study.

OBJECTIVE

While the American Spinal Injury Association (ASIA) Impairment Scale is the standard for assessing spinal cord injuries (SCI), it has limitations due to subjectivity and impracticality. Advances in machine learning (ML) and image recognition have spurred research into their use for outcome prediction. This study aims to analyze deep learning techniques for identifying and classifying cervical SCI severity from MRI scans.

METHODS

The study included patients with traumatic and nontraumatic cervical SCI admitted from 2019 to 2022. MRI images were labeled by two senior resident physicians. A deep convolutional neural network was trained using axial and sagittal cervical MRI images from the dataset. Model performance was assessed using Dice Score and IoU to measure segmentation accuracy by comparing predicted and ground truth masks. Classification accuracy was evaluated with the F1 Score, balancing false positives and negatives.

RESULT

In the axial spinal cord segmentation, we achieved a Dice score of 0.94 for and IoU score of 0.89. In the sagittal spinal cord segmentation, we obtained Dice score up to 0.9201 and IoU scores up to 0.8541. The model for axial image score classification gave a satisfactory result with an F1 score of 0.72 and AUC of 0.79.

CONCLUSION

Our models successfully identified cervical SCI on T2-weighted MR images with satisfactory performance. Further research is needed to develop more advanced models for predicting patient outcomes in SCI cases.

The contribution of paramagnetic rim and cortical lesions to physical and cognitive disability at multiple sclerosis clinical onset: evaluating the power of MRI and OCT biomarkers

BACKGROUND

In multiple sclerosis (MS), imaging biomarkers play a crucial role in characterizing the disease at the time of diagnosis. MRI and optical coherence tomography (OCT) provide readily available biomarkers that may help to define the patient's clinical profile. However, the evaluation of cortical and paramagnetic rim lesions (CL, PRL), as well as retinal atrophy, is not routinely performed in clinic.

OBJECTIVE

To identify the most significant MRI and OCT biomarkers associated with early clinical disability in MS.

METHODS

Brain, spinal cord (SC) MRI, and OCT scans were acquired from 45 patients at MS diagnosis to obtain: brain PRL and non-PRL, CL, SC lesion volumes and counts, brain volumetric metrics, SC C2-C3 cross-sectional area, and retinal layer thickness. Regression models assessed relationships with physical disability (Expanded Disability Status Scale [EDSS]) and cognitive performance (Brief International Cognitive Assessment for Multiple Sclerosis [BICAMS]).

RESULTS

In a stepwise regression (R2 = 0.526), PRL (β = 0.001, p = 0.023) and SC lesion volumes (β = 0.001, p = 0.017) were the most significant predictors of EDSS, while CL volume and age were strongly associated with BICAMS scores. Moreover, in a model where PRL and non-PRL were pooled, only the contribution of SC lesion volume was retained in EDSS prediction. OCT measures did not show associations with disability at the onset.

CONCLUSION

At MS onset, PRL and SC lesions exhibit the strongest association with physical disability, while CL strongly contribute to cognitive performance. Incorporating the evaluation of PRL and CL into the initial MS patient assessment could help define their clinical profile, thus supporting the treatment choice.

Quantification of cervical spinal stenosis by automated 3D MRI segmentation of spinal cord and cerebrospinal fluid space

DESIGN

Prospective diagnostic study.

OBJECTIVES

Anatomical evaluation and graduation of the severity of spinal stenosis is essential in degenerative cervical spine disease. In clinical practice, this is subjectively categorized on cervical MRI lacking an objective and reliable classification. We implemented a fully-automated quantification of spinal canal compromise through 3D T2-weighted MRI segmentation.

SETTING

Medical Center - University of Freiburg, Germany.

METHODS

Evaluation of 202 participants receiving 3D T2-weighted MRI of the cervical spine. Segments C2/3 to C6/7 were analyzed for spinal cord and cerebrospinal fluid space volume through a fully-automated segmentation based on a trained deep convolutional neural network. Spinal canal narrowing was characterized by relative values, across sever segments as adapted Maximal Canal Compromise (aMCC), and within the index segment as adapted Spinal Cord Occupation Ratio (aSCOR). Additionally, all segments were subjectively categorized by three observers as "no", "relative" or "absolute" stenosis. Computed scores were applied on the subjective categorization.

RESULTS

798 (79.0%) segments were subjectively categorized as "no" stenosis, 85 (8.4%) as "relative" stenosis, and 127 (12.6%) as "absolute" stenosis. The calculated scores revealed significant differences between each category (p ≤ 0.001). Youden's Index analysis of ROC curves revealed optimal cut-offs to distinguish between "no" and "relative" stenosis for aMCC = 1.18 and aSCOR = 36.9%, and between "relative" and "absolute" stenosis for aMCC = 1.54 and aSCOR = 49.3%.

CONCLUSION

The presented fully-automated segmentation algorithm provides high diagnostic accuracy and objective classification of cervical spinal stenosis. The calculated cut-offs can be used for convenient radiological quantification of the severity of spinal canal compromise in clinical routine.

Association of Spinal Cord Atrophy and Brain Paramagnetic Rim Lesions With Progression Independent of Relapse Activity in People With MS

BACKGROUND AND OBJECTIVES

Progression independent of relapse activity (PIRA) is a crucial determinant of overall disability accumulation in multiple sclerosis (MS). Accelerated brain atrophy has been shown in patients experiencing PIRA. In this study, we assessed the relation between PIRA and neurodegenerative processes reflected by (1) longitudinal spinal cord atrophy and (2) brain paramagnetic rim lesions (PRLs). Besides, the same relationship was investigated in progressive MS (PMS). Last, we explored the value of cross-sectional brain and spinal cord volumetric measurements in predicting PIRA.

METHODS

From an ongoing multicentric cohort study, we selected patients with MS with (1) availability of a susceptibility-based MRI scan and (2) regular clinical and conventional MRI follow-up in the 4 years before the susceptibility-based MRI. Comparisons in spinal cord atrophy rates (explored with linear mixed-effect models) and PRL count (explored with negative binomial regression models) were performed between: (1) relapsing-remitting (RRMS) and PMS phenotypes and (2) patients experiencing PIRA and patients without confirmed disability accumulation (CDA) during follow-up (both considering the entire cohort and the subgroup of patients with RRMS). Associations between baseline MRI volumetric measurements and time to PIRA were explored with multivariable Cox regression analyses.

RESULTS

In total, 445 patients with MS (64.9% female; mean [SD] age at baseline 45.0 [11.4] years; 11.2% with PMS) were enrolled. Compared with patients with RRMS, those with PMS had accelerated cervical cord atrophy (mean difference in annual percentage volume change [MD-APC] -1.41; p = 0.004) and higher PRL load (incidence rate ratio [IRR] 1.93; p = 0.005). Increased spinal cord atrophy (MD-APC -1.39; p = 0.0008) and PRL burden (IRR 1.95; p = 0.0008) were measured in patients with PIRA compared with patients without CDA; such differences were also confirmed when restricting the analysis to patients with RRMS. Baseline volumetric measurements of the cervical cord, whole brain, and cerebral cortex significantly predicted time to PIRA (all p ≤ 0.002).

DISCUSSION

Our results show that PIRA is associated with both increased spinal cord atrophy and PRL burden, and this association is evident also in patients with RRMS. These findings further point to the need to develop targeted treatment strategies for PIRA to prevent irreversible neuroaxonal loss and optimize long-term outcomes of patients with MS.

The CALIPR framework for highly accelerated myelin water imaging with improved precision and sensitivity

Quantitative magnetic resonance imaging (MRI) techniques are powerful tools for the study of human tissue, but, in practice, their utility has been limited by lengthy acquisition times. Here, we introduce the Constrained, Adaptive, Low-dimensional, Intrinsically Precise Reconstruction (CALIPR) framework in the context of myelin water imaging (MWI); a quantitative MRI technique generally regarded as the most rigorous approach for noninvasive, in vivo measurement of myelin content. The CALIPR framework exploits data redundancy to recover high-quality images from a small fraction of an imaging dataset, which allowed MWI to be acquired with a previously unattainable sequence (fully sampled acquisition 2 hours:57 min:20 s) in 7 min:26 s (4.2% of the dataset, acceleration factor 23.9). CALIPR quantitative metrics had excellent precision (myelin water fraction mean coefficient of variation 3.2% for the brain and 3.0% for the spinal cord) and markedly increased sensitivity to demyelinating disease pathology compared to a current, widely used technique. The CALIPR framework facilitates drastically improved MWI and could be similarly transformative for other quantitative MRI applications.

Lumbar Spinal Canal Segmentation in Cases with Lumbar Stenosis Using Deep-U-Net Ensembles

BACKGROUND

Narrowing of the lumbar spinal canal, or lumbar stenosis (LS), may cause debilitating radicular pain or muscle weakness. It is the most frequent indication for spinal surgery in the elderly population. Modern diagnosis relies on magnetic resonance imaging and its inherently subjective interpretation. Diagnostic rigor, accuracy, and speed may be improved by automation. In this work, we aimed to determine whether a deep-U-Net ensemble trained to segment spinal canals on a heterogeneous mix of clinical data is comparable to radiologists' segmentation of these canals in patients with LS.

METHODS

The deep U-nets were trained on spinal canals segmented by physicians on 100 axial T2 lumbar magnetic resonance imaging selected randomly from our institutional database. Test data included a total of 279 elderly patients with LS that were separate from the training set.

RESULTS

Machine-generated segmentations (MA) were qualitatively similar to expert-generated segmentations (ME1, ME2). Machine- and expert-generated segmentations were quantitatively similar, as evidenced by Dice scores (MA vs. ME1: 0.88 ± 0.04, MA vs. ME2: 0.89 ± 0.04), the Hausdorff distance (MA vs. ME1: 11.7 mm ± 13.8, MA vs. ME2: 13.1 mm ± 16.3), and average surface distance (MAvs. ME1: 0.18 mm ± 0.13, MA vs. ME2 0.18 mm ± 0.16) metrics. These metrics are comparable to inter-rater variation (ME1 vs. ME2 Dice scores: 0.94 ± 0.02, the Hausdorff distances: 9.3 mm ± 15.6, average surface distances: 0.08 mm ± 0.09).

CONCLUSION

We conclude that machine learning algorithms can segment lumbar spinal canals in LS patients, and automatic delineations are both qualitatively and quantitatively comparable to expert-generated segmentations.

Automated signal intensity analysis of the spinal cord for detection of degenerative cervical myelopathy - a matched-pair MRI study

PURPOSE

Detection of T2 hyperintensities in suspected degenerative cervical myelopathy (DCM) is done subjectively in clinical practice. To gain objective quantification for dedicated treatment, signal intensity analysis of the spinal cord is purposeful. We investigated fully automated quantification of the T2 signal intensity (T2-SI) of the spinal cord using a high-resolution MRI segmentation.

METHODS

Matched-pair analysis of prospective acquired cervical 3D T2-weighted sequences of 114 symptomatic patients and 88 healthy volunteers. Cervical spinal cord was segmented automatically through a trained convolutional neuronal network with subsequent T2-SI registration slice-by-slice. Received T2-SI curves were subdivided for each cervical level from C2 to C7. Additionally, all levels were subjectively classified concerning a present T2 hyperintensity. For T2-positive levels, corresponding T2-SI curves were compared to curves of age-matched volunteers at the identical level.

RESULTS

Forty-nine patients showed subjective T2 hyperintensities at any level. The corresponding T2-SI curves showed higher signal variabilities reflected by standard deviation (18.51 vs. 7.47 a.u.; p < 0.001) and range (56.09 vs. 24.34 a.u.; p < 0.001) compared to matched controls. Percentage of the range from the mean absolute T2-SI per cervical level, introduced as "T2 myelopathy index" (T2-MI), was correspondingly significantly higher in T2-positive segments (23.99% vs. 10.85%; p < 0.001). ROC analysis indicated excellent differentiation for all three parameters (AUC 0.865-0.920).

CONCLUSION

This fully automated T2-SI quantification of the spinal cord revealed significantly increased signal variability for DCM patients compared to healthy volunteers. This innovative procedure and the applied parameters showed sufficient diagnostic accuracy, potentially diagnosing radiological DCM more objective to optimize treatment recommendation.

TRIAL REGISTRATION

DRKS00012962 (17.01.2018) and DRKS00017351 (28.05.2019).

Man versus machine: Automatic pedicle screw planning using registration-based techniques compared with manual screw planning for thoracolumbar fusion surgeries

OBJECTIVE

This study evaluates the precision of a commercially available spine planning software in automatic spine labelling and screw-trajectory proposal.

METHODS

The software uses automatic segmentation and registration of the vertebra to generate screw proposals. 877 trajectories were compared. Four neurosurgeons assessed suggested trajectories, performed corrections, and manually planned pedicle screws. Additionally, automatic identification/labelling was evaluated.

RESULTS

Automatic labelling was correct in 89% of the cases. 92.9% of automatically planned trajectories were in accordance with G&R grade A + B. Automatic mode reduced the time spent planning screw trajectories by 7 s per screw to 20 s per vertebra. Manual mode yielded differences in screw-length between surgeons (largest distribution peak: 5 mm), automatic in contrast at 0 mm. The size of suggested pedicle screws was significantly smaller (largest peaks in difference between 0.5 and 3 mm) than the surgeon's choice.

CONCLUSION

Automatic identification of vertebrae works in most cases and suggested pedicle screw trajectories are acceptable. So far, it does not substitute for an experienced surgeon's assessment.

Automatic semantic segmentation of the lumbar spine: Clinical applicability in a multi-parametric and multi-center study on magnetic resonance images

Significant difficulties in medical image segmentation include the high variability of images caused by their origin (multi-center), the acquisition protocols (multi-parametric), the variability of human anatomy, illness severity, the effect of age and gender, and notable other factors. This work addresses problems associated with the automatic semantic segmentation of lumbar spine magnetic resonance images using convolutional neural networks. We aimed to assign a class label to each pixel of an image, with classes defined by radiologists corresponding to structural elements such as vertebrae, intervertebral discs, nerves, blood vessels, and other tissues. The proposed network topologies represent variants of the U-Net architecture, and we used several complementary blocks to define the variants: three types of convolutional blocks, spatial attention models, deep supervision, and multilevel feature extractor. Here, we describe the topologies and analyze the results of the neural network designs that obtained the most accurate segmentation. Several proposed designs outperform the standard U-Net used as a baseline, primarily when used in ensembles, where the outputs of multiple neural networks are combined according to different strategies.

Deep Learning-Based Auto-Segmentation of Spinal Cord Internal Structure of Diffusion Tensor Imaging in Cervical Spondylotic Myelopathy

Cervical spondylotic myelopathy (CSM) is a chronic disorder of the spinal cord. ROI-based features on diffusion tensor imaging (DTI) provide additional information about spinal cord status, which would benefit the diagnosis and prognosis of CSM. However, the manual extraction of the DTI-related features on multiple ROIs is time-consuming and laborious. In total, 1159 slices at cervical levels from 89 CSM patients were analyzed, and corresponding fractional anisotropy (FA) maps were calculated. Eight ROIs were drawn, covering both sides of lateral, dorsal, ventral, and gray matter. The UNet model was trained with the proposed heatmap distance loss for auto-segmentation. Mean Dice coefficients on the test dataset for dorsal, lateral, and ventral column and gray matter were 0.69, 0.67, 0.57, 0.54 on the left side and 0.68, 0.67, 0.59, 0.55 on the right side. The ROI-based mean FA value based on segmentation model strongly correlated with the value based on manual drawing. The percentages of the mean absolute error between the two values of multiple ROIs were 0.07, 0.07, 0.11, and 0.08 on the left side and 0.07, 0.1, 0.1, 0.11, and 0.07 on the right side. The proposed segmentation model has the potential to offer a more detailed spinal cord segmentation and would be beneficial for quantifying a more detailed status of the cervical spinal cord.

Magnetic resonance image segmentation of the compressed spinal cord in patients with degenerative cervical myelopathy using convolutional neural networks

PURPOSE

Spinal cord segmentation is the first step in atlas-based spinal cord image analysis, but segmentation of compressed spinal cords from patients with degenerative cervical myelopathy is challenging. We applied convolutional neural network models to segment the spinal cord from T2-weighted axial magnetic resonance images of DCM patients. Furthermore, we assessed the correlation between the cross-sectional area segmented by this network and the neurological symptoms of the patients.

METHODS

The CNN architecture was built using U-Net and DeepLabv3 + and PyTorch. The CNN was trained on 2762 axial slices from 174 patients, and an additional 517 axial slices from 33 patients were held out for validation and 777 axial slices from 46 patients for testing. The performance of the CNN was evaluated on a test dataset with Dice coefficients as the outcome measure. The ratio of CSA at the maximum compression level to CSA at the C2 level, as segmented by the CNN, was calculated. The correlation between the spinal cord CSA ratio and the Japanese Orthopaedic Association score in DCM patients from the test dataset was investigated using Spearman's rank correlation coefficient.

RESULTS

The best Dice coefficient was achieved when U-Net was used as the architecture and EfficientNet-b7 as the model for transfer learning. Spearman's r_s between the spinal cord CSA ratio and the JOA score of DCM patients was 0.38 (p = 0.007), showing a weak correlation.

CONCLUSION

Using deep learning with magnetic resonance images of deformed spinal cords as training data, we were able to segment compressed spinal cords of DCM patients with a high concordance with expert manual segmentation. In addition, the spinal cord CSA ratio was weakly, but significantly, correlated with neurological symptoms. Our study demonstrated the first steps needed to implement automated atlas-based analysis of DCM patients.

Machine learning for image analysis in the cervical spine: Systematic review of the available models and methods

•Neural network approaches show the most potential for automated image analysis of thecervical spine.•Fully automatic convolutional neural network (CNN) models are promising Deep Learning methods for segmentation.•In cervical spine analysis, the biomechanical features are most often studied using finiteelement models.•The application of artificial neural networks and support vector machine models looks promising for classification purposes.•This article provides an overview of the methods for research on computer aided imaging diagnostics of the cervical spine.

Spinal cord segmentation and injury detection using a Crow Search-Rider optimization algorithm

The damage in the spinal cord due to vertebral fractures may result in loss of sensation and muscle function either permanently or temporarily. The neurological condition of the patient can be improved only with the early detection and the treatment of the injury in the spinal cord. This paper proposes a spinal cord segmentation and injury detection system based on the proposed Crow search-Rider Optimization-based DCNN (CS-ROA DCNN) method, which can detect the injury in the spinal cord in an effective manner. Initially, the segmentation of the CT image of the spinal cord is performed using the adaptive thresholding method, followed by which the localization of the disc is performed using the Sparse FCM clustering algorithm (Sparse-FCM). The localized discs are subjected to a feature extraction process, where the features necessary for the classification process are extracted. The classification process is done using DCNN trained using the proposed CS-ROA, which is the integration of the Crow Search Algorithm (CSA) and Rider Optimization Algorithm (ROA). The experimentation is performed using the evaluation metrics, such as accuracy, sensitivity, and specificity. The proposed method achieved the high accuracy, sensitivity, and specificity of 0.874, 0.8961, and 0.8828, respectively that shows the effectiveness of the proposed CS-ROA DCNN method in spinal cord injury detection.

Automatic spinal cord segmentation from axial-view MRI slices using CNN with grayscale regularized active contour propagation

Accurate positioning of the responsible segment for patients with cervical spondylotic myelopathy (CSM) is clinically important not only to the surgery but also to reduce the incidence of surgical trauma and complications. Spinal cord segmentation is a crucial step in the positioning procedure. This study proposed a fully automated approach for spinal cord segmentation from 2D axial-view MRI slices of patients with CSM. The proposed method was trained and tested using clinical data from 20 CSM patients (359 images) acquired by the Peking University Third Hospital, with ground truth labeled by professional radiologists. The accuracy of the proposed method was evaluated using quantitative measures, the reliability metric as well as visual assessment. The proposed method yielded a Dice coefficient of 87.0%, Hausdorff distance of 9.7 mm, root-mean-square error of 5.9 mm. Higher conformance with ground truth was observed for the proposed method in comparison to the state-of-the-art algorithms. The results are also statistically significant with p-values calculated between state-of-the-art methods and the proposed methods.

3D patient-specific spinal cord computational model for SCS management: potential clinical applications

BACKGROUND AND OBJECTIVE

Although Spinal Cord Stimulation (SCS) is an established therapy for treating neuropathic chronic pain, in tonic stimulation, postural changes, electrode migration or badly-positioned electrodes can produce annoying stimulation (intercostal neuralgia) in about 35% of the patients. SCS models are used to study the effect of electrical stimulation to better manage the stimulation parameters and electrode position. The goal of this work was to develop a realistic 3D patient-specific spinal cord model from a real patient and develop a future clinical application that would help physicians to optimize paresthesia coverage in SCS therapy.

METHODS

We developed two 3D patient-specific models from a high-resolution MRI of two patients undergoing SCS treatment. The model consisted of a finite element model of the spinal cord and a sensory myelinated nerve fiber model. The same simulations were performed with a generalized spinal cord model and we compared the results with the clinical data to evaluate the advantages of a patient-specific model. To identify the geometrical parameters that most influence the stimulation predictions, a sensitivity analysis was conducted. We used the patient-specific model to perform a clinical application involving the pre-implantation selection of electrode polarity and study the effect of electrode offset.

RESULTS

The patient-specific model correlated better with clinical data than the generalized model. Electrode-dura mater distance, dorsal CSF thickness, and CSF diameter are the geometrical parameters that caused significant changes in the stimulation predictions. Electrode polarity could be planned and optimized to stimulate the patient's painful dermatomes. The addition of offset in parallel electrodes would not have been beneficial for one of the patients of this study because they reduce neural activation displacement.

CONCLUSIONS

This is the first study to relate the activation area model prediction in dorsal columns with the clinical effect on paresthesia coverage. The outcomes show that 3D patient-specific models would help physicians to choose the best stimulation parameters to optimize neural activation and SCS therapy in tonic stimulation.

Focal cervical spinal stenosis causes mechanical strain on the entire cervical spinal cord tissue - A prospective controlled, matched-pair analysis based on phase-contrast MRI

BACKGROUND

Focally increased spinal cord motion at the level of cervical spinal stenosis has been revealed by phase-contrast MRI (PC-MRI).

OBJECTIVE

To investigate spinal cord motion among patients suffering of degenerative cervical myelopathy (DCM) across the entire cervical spine applying automated segmentation and standardized PC-MRI post-processing protocols.

METHODS

Prospective, matched-pair controlled trial on 29 patients with stenosis at C5/C6. MRI-protocol covering all cervical segments: 3D T2-SPACE, prospectively ECG-triggered sagittal PC-MRI. Segmentation by trained 3D hierarchical deep convolutional neural network and data processing were conducted via in-house software pipeline. Parameters per segment: maximum velocity, peak-to-peak (PTP)-amplitude, total displacement, PTP-amplitude^HB (PTP-amplitude per duration of heartbeat), and, for characterization of intraindividual alterations, the PTP-amplitude index between the cervical segments C3/C4-C7/T1 and C2/C3.

RESULTS

Spinal cord motion was increased at C4/C5, C5/C6 and C6/C7 among patients (all parameters, p < 0.001-0.025). The PTP-amplitude index revealed an increase from C3/C4 to C4/C5 (p = 0.002), C4/C5 to C5/C6 (p = 0.037) and a decrease from C5/C6 to C6/C7 and C6/C7 to C7/T1 (p < 0.001, each). This implied an up-building stretch on spinal cord tissue cranial and a mechanical compression caudal of the stenotic level. Furthermore, significant far range effects across the entire cervical spinal cord were observed (e.g. PTP-amplitude C2/C3 vs. C6/C7, p = 0.026) in contrast to controls (p = 1.00).

CONCLUSION

This study revealed the nature and extends of mechanical stress on the entire cervical spinal cord tissue due to focal stenosis. These pathophysiological alterations of spinal cord motion can be expected to be clinically relevant.

Mapping the rest of the human connectome: Atlasing the spinal cord and peripheral nervous system

The emergence of diffusion, structural, and functional neuroimaging methods has enabled major multi-site efforts to map the human connectome, which has heretofore been defined as containing all neural connections in the central nervous system (CNS). However, these efforts are not structured to examine the richness and complexity of the peripheral nervous system (PNS), which arguably forms the (neglected) rest of the connectome. Despite increasing interest in an atlas of the spinal cord (SC) and PNS which is simultaneously stereotactic, interactive, electronically dissectible, scalable, population-based and deformable, little attention has thus far been devoted to this task of critical importance. Nevertheless, the atlasing of these complete neural structures is essential for neurosurgical planning, neurological localization, and for mapping those components of the human connectome located outside of the CNS. Here we recommend a modification to the definition of the human connectome to include the SC and PNS, and argue for the creation of an inclusive atlas to complement current efforts to map the brain's human connectome, to enhance clinical education, and to assist progress in neuroscience research. In addition to providing a critical overview of existing neuroimaging techniques, image processing methodologies and algorithmic advances which can be combined for the creation of a full connectome atlas, we outline a blueprint for ultimately mapping the entire human nervous system and, thereby, for filling a critical gap in our scientific knowledge of neural connectivity.

Comparison of multi echo T2 relaxation and steady state approaches for myelin imaging in the central nervous system

The traditional approach for measuring myelin-associated water with quantitative magnetic resonance imaging (MRI) uses multi-echo T₂ relaxation data to calculate the myelin water fraction (MWF). A fundamentally different approach, abbreviated “mcDESPOT”, uses a more efficient steady-state acquisition to generate an equivalent metric (f_M). Although previous studies have demonstrated inherent instability and bias in the complex mcDESPOT analysis procedure, f_M has often been used as a surrogate for MWF. We produced and compared multivariate atlases of MWF and f_M in healthy human brain and cervical spinal cord (available online) and compared their ability to detect multiple sclerosis pathology. A significant bias was found in all regions (p < 10^–5), albeit reversed for spinal cord (f_M-MWF =  − 3.4%) compared to brain (+ 6.2%). MWF and f_M followed an approximately linear relationship for regions with MWF <  ~ 10%. For MWF >  ~ 10%, the relationship broke down and f_M no longer increased in tandem with MWF. For multiple sclerosis patients, MWF and f_M Z score maps showed overlapping areas of low Z score and similar trends between patients and brain regions, although those of f_M generally had greater spatial extent and magnitude of severity. These results will guide future choice of myelin-sensitive quantitative MRI and improve interpretation of studies using either myelin imaging approach.

Injury classification and level detection of the spinal cord based on the optimized recurrent neural network

Objectives

Spinal cord damage is one of the traumatic situations in persons that may cause the loss of sensation and proper functioning of the muscles either temporarily or permanently. Hence, steps to assure the recovery through the early functioning and precaution could safe-guard a proper interceptive. To ensure the recovery of spinal cord damage through optimized recurrent neural network.

Methods

The research on the spinal cord injury classification and level detection is done using the CT images, which is initially given to the segmentation that is done using the adaptive thresholding methodology. Once the segments are formed, the disc is localized using the sparse fuzzy C-means clustering approach. In the next step, the features are extracted from the localized disc and the features include the connectivity features, statistical features, image-level features, grid-level features, Histogram of Oriented Gradients (HOG), and Linear Gradient Pattern (LGP). Then, the injury detection is done based on the Crow search Rider Optimization algorithm-based Deep Convolutional Neural Network (CS-ROA-based DCNN). Once the result regarding the presence of the injury is obtained, the injury-level classification is done based on the proposed Deep Recurrent Neural Network (Deep RNN), and in case of the absence of injury, the process is terminated. Therefore, the injury detection classifier derives the level of the injury, such as normal, wedge, biconcavity, and crush.

Results

The experimentation is carried out using an Osteoporotic vertebral fractures database. The performance of the injury level detection based on the proposed model is evaluated based on accuracy, sensitivity, and specificity. The proposed model achieves the maximal accuracy of 0.895, maximal sensitivity of 0.871, and the maximal specificity of 0.933 with respect to K-Fold.

Conclusions

The experimental results show that the proposed model is better than the existing models in terms of accuracy, sensitivity, and specificity.

HARDI-ZOOMit protocol improves specificity to microstructural changes in presymptomatic myelopathy

Diffusion magnetic resonance imaging (dMRI) proved promising in patients with non-myelopathic degenerative cervical cord compression (NMDCCC), i.e., without clinically manifested myelopathy. Aim of the study is to present a fast multi-shell HARDI-ZOOMit dMRI protocol and validate its usability to detect microstructural myelopathy in NMDCCC patients. In 7 young healthy volunteers, 13 age-comparable healthy controls, 18 patients with mild NMDCCC and 15 patients with severe NMDCCC, the protocol provided higher signal-to-noise ratio, enhanced visualization of white/gray matter structures in microstructural maps, improved dMRI metric reproducibility, preserved sensitivity (SE = 87.88%) and increased specificity (SP = 92.31%) of control-patient group differences when compared to DTI-RESOLVE protocol (SE = 87.88%, SP = 76.92%). Of the 56 tested microstructural parameters, HARDI-ZOOMit yielded significant patient-control differences in 19 parameters, whereas in DTI-RESOLVE data, differences were observed in 10 parameters, with mostly lower robustness. Novel marker the white-gray matter diffusivity gradient demonstrated the highest separation. HARDI-ZOOMit protocol detected larger number of crossing fibers (5–15% of voxels) with physiologically plausible orientations than DTI-RESOLVE protocol (0–8% of voxels). Crossings were detected in areas of dorsal horns and anterior white commissure. HARDI-ZOOMit protocol proved to be a sensitive and practical tool for clinical quantitative spinal cord imaging.

Spinal cord atrophy as a measure of severity of myelopathy in adrenoleukodystrophy

All men and most women with X-linked adrenoleukodystrophy (ALD) develop myelopathy in adulthood. As clinical trials with new potential disease-modifying therapies are emerging, sensitive outcome measures for quantifying myelopathy are needed. This prospective cohort study evaluated spinal cord size (cross-sectional area - CSA) and shape (eccentricity) as potential new quantitative outcome measures for myelopathy in ALD. Seventy-four baseline magnetic resonance imaging (MRI) scans, acquired in 42 male ALD patients and 32 age-matched healthy controls, and 26 follow-up scans of ALD patients were included in the study. We used routine T1 -weighted MRI sequences to measure mean CSA, eccentricity, right-left and anteroposterior diameters in the cervical spinal cord. We compared MRI measurements between groups and correlated CSA with clinical outcome measures of disease severity. Longitudinally, we compared MRI measurements between baseline and 1-year follow-up. CSA was significantly smaller in patients compared to controls on all measured spinal cord levels (P < .001). The difference was completely explained by the effect of the symptomatic subgroup. Furthermore, the spinal cord showed flattening (higher eccentricity and smaller anteroposterior diameters) in patients. CSA correlated strongly with all clinical measures of severity of myelopathy. There was no detectable change in CSA after 1-year follow-up. The cervical spinal cord in symptomatic ALD patients is smaller and flattened compared to controls, possibly due to atrophy of the dorsal columns. CSA is a reliable marker of disease severity and can be a valuable outcome measure in long-term follow-up studies in ALD. SYNOPSIS: A prospective cohort study in 42 adrenoleukodystrophy (ALD) patients and 32 controls demonstrated that the spinal cord cross-sectional area of patients is smaller compared to healthy controls and correlates with severity of myelopathy in patients, hence it could be valuable as a much needed surrogate outcome measure.

A scalable approach to T2-MRI colon segmentation

The study of the colonic volume is a procedure with strong relevance to gastroenterologists. Depending on the clinical protocols, the volume analysis has to be performed on MRI of the unprepared colon without contrast administration. In such circumstances, existing measurement procedures are cumbersome and time-consuming for the specialists. The algorithm presented in this paper permits a quasi-automatic segmentation of the unprepared colon on T2-weighted MRI scans. The segmentation algorithm is organized as a three-stage pipeline. In the first stage, a custom tubularity filter is run to detect colon candidate areas. The specialists provide a list of points along the colon trajectory, which are combined with tubularity information to calculate an estimation of the colon medial path. In the second stage, we delimit the region of interest by applying custom segmentation algorithms to detect colon neighboring regions and the fat capsule containing abdominal organs. Finally, within the reduced search space, segmentation is performed via 3D graph-cuts in a three-stage multigrid approach. Our algorithm was tested on MRI abdominal scans, including different acquisition resolutions, and its results were compared to the colon ground truth segmentations provided by the specialists. The experiments proved the accuracy, efficiency, and usability of the algorithm, while the variability of the scan resolutions contributed to demonstrate the computational scalability of the multigrid architecture. The system is fully applicable to the colon measurement clinical routine, being a substantial step towards a fully automated segmentation.

Heritability of cervical spinal cord structure

OBJECTIVE

Measures of spinal cord structure can be a useful phenotype to track disease severity and development; this observational study measures the hereditability of cervical spinal cord anatomy and its correlates in healthy human beings.

METHODS

Twin data from the Human Connectome Project were analyzed with semiautomated spinal cord segmentation, evaluating test-retest reliability and broad-sense heritability with an AE model. Relationships between spinal cord metrics, general physical measures, regional brain structural measures, and motor function were assessed.

RESULTS

We found that the spinal cord C2 cross-sectional area (CSA), left-right width (LRW), and anterior-posterior width (APW) are highly heritable (85%-91%). All measures were highly correlated with the brain volume, and CSA only was positively correlated with thalamic volumes (p = 0.005) but negatively correlated with the occipital cortex area (p = 0.001). LRW was correlated with the participant's height (p = 0.00027). The subjects' sex significantly influenced these metrics. Analyses of a test-retest data set confirmed validity of the approach.

CONCLUSIONS

This study provides the evidence of genetic influence on spinal cord structure. MRI metrics of cervical spinal cord anatomy are robust and not easily influenced by nonpathological environmental factors, providing a useful metric for monitoring normal development and progression of neurodegenerative disorders affecting the spinal cord, including-but not limited to-spinal cord injury and MS.

Fully automatic 3D segmentation of the thoracolumbar spinal cord and the vertebral canal from T2-weighted MRI using K-means clustering algorithm

STUDY DESIGN

Method development.

OBJECTIVES

To develop a reliable protocol for automatic segmentation of Thoracolumbar spinal cord using MRI based on K-means clustering algorithm in 3D images.

SETTING

University-based laboratory, Tehran, Iran.

METHODS

T2 structural volumes acquired from the spinal cord of 20 uninjured volunteers on a 3T MR scanner. We proposed an automatic method for spinal cord segmentation based on the K-means clustering algorithm in 3D images and compare our results with two available segmentation methods (PropSeg, DeepSeg) implemented in the Spinal Cord Toolbox. Dice and Hausdorff were used to compare the results of our method (K-Seg) with the manual segmentation, PropSeg, and DeepSeg.

RESULTS

The accuracy of our automatic segmentation method for T2-weighted images was significantly better or similar to the SCT methods, in terms of 3D DC (p < 0.001). The 3D DCs were respectively (0.81 ± 0.04) and Hausdorff Distance (12.3 ± 2.48) by the K-Seg method in contrary to other SCT methods for T2-weighted images.

CONCLUSIONS

The output with similar protocols showed that K-Seg results match the manual segmentation better than the other methods especially on the thoracolumbar levels in the spinal cord due to the low image contrast as a result of poor SNR in these areas.

Intersubject Variability and Normalization Strategies for Spinal Cord Total Cross-Sectional and Gray Matter Areas

BACKGROUND AND PURPOSE

The quantification of spinal cord (SC) atrophy by MRI has assumed an important role in assessment of neuroinflammatory/neurodegenerative diseases and traumatic SC injury. Recent technical advances make possible the quantification of gray matter (GM) and white matter tissues in clinical settings. However, the goal of a reliable diagnostic, prognostic or predictive marker is still elusive, in part due to large intersubject variability of SC areas. Here, we investigated the sources of this variability and explored effective strategies to reduce it.

METHODS

One hundred twenty-nine healthy subjects (mean age: 41.0 ± 15.9) underwent MRI on a Siemens 3T Skyra scanner. Two-dimensional PSIR at the C2-C3 vertebral level and a sagittal 1 mm³ 3D T1-weighted brain acquisition extended to the upper cervical cord were acquired. Total cross-sectional area and GM area were measured at C2-C3, as well as measures of the vertebra, spinal canal and the skull. Correlations between the different metrics were explored using Pearson product-moment coefficients. The most promising metrics were used to normalize cord areas using multiple regression analyses.

RESULTS

The most effective normalization metrics were the V-scale (from SienaX) and the product of the C2-C3 spinal canal diameters. Normalization methods based on these metrics reduced the intersubject variability of cord areas of up to 17.74%. The measured cord areas had a statistically significant sex difference, while the effect of age was moderate.

CONCLUSIONS

The present work explored in a large cohort of healthy subjects the source of intersubject variability of SC areas and proposes effective normalization methods for its reduction.

Machine learning for automated 3-dimensional segmentation of the spine and suggested placement of pedicle screws based on intraoperative cone-beam computer tomography

OBJECTIVE

The goal of this study was to develop and validate a system for automatic segmentation of the spine, pedicle identification, and screw path suggestion for use with an intraoperative 3D surgical navigation system.

METHODS

Cone-beam CT (CBCT) images of the spines of 21 cadavers were obtained. An automated model-based approach was used for segmentation. Using machine learning methodology, the algorithm was trained and validated on the image data sets. For measuring accuracy, surface area errors of the automatic segmentation were compared to the manually outlined reference surface on CBCT. To further test both technical and clinical accuracy, the algorithm was applied to a set of 20 clinical cases. The authors evaluated the system's accuracy in pedicle identification by measuring the distance between the user-defined midpoint of each pedicle and the automatically segmented midpoint. Finally, 2 independent surgeons performed a qualitative evaluation of the segmentation to judge whether it was adequate to guide surgical navigation and whether it would have resulted in a clinically acceptable pedicle screw placement.

RESULTS

The clinically relevant pedicle identification and automatic pedicle screw planning accuracy was 86.1%. By excluding patients with severe spinal deformities (i.e., Cobb angle > 75° and severe spinal degeneration) and previous surgeries, a success rate of 95.4% was achieved. The mean time (± SD) for automatic segmentation and screw planning in 5 vertebrae was 11 ± 4 seconds.

CONCLUSIONS

The technology investigated has the potential to aid surgeons in navigational planning and improve surgical navigation workflow while maintaining patient safety.

Convolutional Neural Network-Based Automated Segmentation of the Spinal Cord and Contusion Injury: Deep Learning Biomarker Correlates of Motor Impairment in Acute Spinal Cord Injury

BACKGROUND AND PURPOSE

Our aim was to use 2D convolutional neural networks for automatic segmentation of the spinal cord and traumatic contusion injury from axial T2-weighted MR imaging in a cohort of patients with acute spinal cord injury.

MATERIALS AND METHODS

Forty-seven patients who underwent 3T MR imaging within 24 hours of spinal cord injury were included. We developed an image-analysis pipeline integrating 2D convolutional neural networks for whole spinal cord and intramedullary spinal cord lesion segmentation. Linear mixed modeling was used to compare test segmentation results between our spinal cord injury convolutional neural network (Brain and Spinal Cord Injury Center segmentation) and current state-of-the-art methods. Volumes of segmented lesions were then used in a linear regression analysis to determine associations with motor scores.

RESULTS

Compared with manual labeling, the average test set Dice coefficient for the Brain and Spinal Cord Injury Center segmentation model was 0.93 for spinal cord segmentation versus 0.80 for PropSeg and 0.90 for DeepSeg (both components of the Spinal Cord Toolbox). Linear mixed modeling showed a significant difference between Brain and Spinal Cord Injury Center segmentation compared with PropSeg (P < .001) and DeepSeg (P < .05). Brain and Spinal Cord Injury Center segmentation showed significantly better adaptability to damaged areas compared with PropSeg (P < .001) and DeepSeg (P < .02). The contusion injury volumes based on automated segmentation were significantly associated with motor scores at admission (P = .002) and discharge (P = .009).

CONCLUSIONS

Brain and Spinal Cord Injury Center segmentation of the spinal cord compares favorably with available segmentation tools in a population with acute spinal cord injury. Volumes of injury derived from automated lesion segmentation with Brain and Spinal Cord Injury Center segmentation correlate with measures of motor impairment in the acute phase. Targeted convolutional neural network training in acute spinal cord injury enhances algorithm performance for this patient population and provides clinically relevant metrics of cord injury.

Spinal cord detection in planning CT for radiotherapy through adaptive template matching, IMSLIC and convolutional neural networks

BACKGROUND AND OBJECTIVE

The spinal cord is a very important organ that must be protected in treatments of radiotherapy (RT), considered an organ at risk (OAR). Excess rays associated with the spinal cord can cause irreversible diseases in patients who are undergoing radiotherapy. For the planning of treatments with RT, computed tomography (CT) scans are commonly used to delimit the OARs and to analyze the impact of rays in these organs. Delimiting these OARs take a lot of time from medical specialists, plus the fact that involves a large team of professionals. Moreover, this task made slice-by-slice becomes an exhaustive and consequently subject to errors, especially in organs such as the spinal cord, which extend through several slices of the CT and requires precise segmentation. Thus, we propose, in this work, a computational methodology capable of detecting spinal cord in planning CT images.

METHODS

The techniques highlighted in this methodology are adaptive template matching for initial segmentation, intrinsic manifold simple linear iterative clustering (IMSLIC) for candidate segmentation and convolutional neural networks (CNN) for candidate classification, that consists of four steps: (1) images acquisition, (2) initial segmentation, (3) candidates segmentation and (4) candidates classification.

RESULTS

The methodology was applied on 36 planning CT images provided by The Cancer Imaging Archive, and achieved an accuracy of 92.55%, specificity of 92.87% and sensitivity of 89.23% with 0.065 of false positives per images, without any false positives reduction technique, in detection of spinal cord.

CONCLUSIONS

It is demonstrated the feasibility of the analysis of planning CT images using IMSLIC and convolutional neural network techniques to achieve success in detection of spinal cord regions.

Automatic segmentation of the spinal cord and intramedullary multiple sclerosis lesions with convolutional neural networks

The spinal cord is frequently affected by atrophy and/or lesions in multiple sclerosis (MS) patients. Segmentation of the spinal cord and lesions from MRI data provides measures of damage, which are key criteria for the diagnosis, prognosis, and longitudinal monitoring in MS. Automating this operation eliminates inter-rater variability and increases the efficiency of large-throughput analysis pipelines. Robust and reliable segmentation across multi-site spinal cord data is challenging because of the large variability related to acquisition parameters and image artifacts. In particular, a precise delineation of lesions is hindered by a broad heterogeneity of lesion contrast, size, location, and shape. The goal of this study was to develop a fully-automatic framework - robust to variability in both image parameters and clinical condition - for segmentation of the spinal cord and intramedullary MS lesions from conventional MRI data of MS and non-MS cases. Scans of 1042 subjects (459 healthy controls, 471 MS patients, and 112 with other spinal pathologies) were included in this multi-site study (n = 30). Data spanned three contrasts (T1-, T2-, and T2∗-weighted) for a total of 1943 vol and featured large heterogeneity in terms of resolution, orientation, coverage, and clinical conditions. The proposed cord and lesion automatic segmentation approach is based on a sequence of two Convolutional Neural Networks (CNNs). To deal with the very small proportion of spinal cord and/or lesion voxels compared to the rest of the volume, a first CNN with 2D dilated convolutions detects the spinal cord centerline, followed by a second CNN with 3D convolutions that segments the spinal cord and/or lesions. CNNs were trained independently with the Dice loss. When compared against manual segmentation, our CNN-based approach showed a median Dice of 95% vs. 88% for PropSeg (p ≤ 0.05), a state-of-the-art spinal cord segmentation method. Regarding lesion segmentation on MS data, our framework provided a Dice of 60%, a relative volume difference of -15%, and a lesion-wise detection sensitivity and precision of 83% and 77%, respectively. In this study, we introduce a robust method to segment the spinal cord and intramedullary MS lesions on a variety of MRI contrasts. The proposed framework is open-source and readily available in the Spinal Cord Toolbox.

Reliable and fast volumetry of the lumbar spinal cord using cord image analyser (Cordial)

OBJECTIVE

To validate the precision and accuracy of the semi-automated cord image analyser (Cordial) for lumbar spinal cord (SC) volumetry in 3D T1w MRI data of healthy controls (HC).

MATERIALS AND METHODS

40 3D T1w images of 10 HC (w/m: 6/4; age range: 18-41 years) were acquired at one 3T-scanner in two MRI sessions (time interval 14.9±6.1 days). Each subject was scanned twice per session, allowing determination of test-retest reliability both in back-to-back (intra-session) and scan-rescan images (inter-session). Cordial was applied for lumbar cord segmentation twice per image by two raters, allowing for assessment of intra- and inter-rater reliability, and compared to a manual gold standard.

RESULTS

While manually segmented volumes were larger (mean: 2028±245 mm³ vs. Cordial: 1636±300 mm³, p<0.001), accuracy assessments between manually and semi-automatically segmented images showed a mean Dice-coefficient of 0.88±0.05. Calculation of within-subject coefficients of variation (COV) demonstrated high intra-session (1.22-1.86%), inter-session (1.26-1.84%), as well as intra-rater (1.73-1.83%) reproducibility. No significant difference was shown between intra- and inter-session reproducibility or between intra-rater reliabilities. Although inter-rater reproducibility (COV: 2.87%) was slightly lower compared to all other reproducibility measures, between rater consistency was very strong (intraclass correlation coefficient: 0.974).

CONCLUSION

While under-estimating the lumbar SCV, Cordial still provides excellent inter- and intra-session reproducibility showing high potential for application in longitudinal trials.

KEY POINTS

• Lumbar spinal cord segmentation using the semi-automated cord image analyser (Cordial) is feasible. • Lumbar spinal cord is 40-mm cord segment 60 mm above conus medullaris. • Cordial provides excellent inter- and intra-session reproducibility in lumbar spinal cord region. • Cordial shows high potential for application in longitudinal trials.

Automatic spinal cord localization, robust to MRI contrasts using global curve optimization

During the last two decades, MRI has been increasingly used for providing valuable quantitative information about spinal cord morphometry, such as quantification of the spinal cord atrophy in various diseases. However, despite the significant improvement of MR sequences adapted to the spinal cord, automatic image processing tools for spinal cord MRI data are not yet as developed as for the brain. There is nonetheless great interest in fully automatic and fast processing methods to be able to propose quantitative analysis pipelines on large datasets without user bias. The first step of most of these analysis pipelines is to detect the spinal cord, which is challenging to achieve automatically across the broad range of MRI contrasts, field of view, resolutions and pathologies. In this paper, a fully automated, robust and fast method for detecting the spinal cord centerline on MRI volumes is introduced. The algorithm uses a global optimization scheme that attempts to strike a balance between a probabilistic localization map of the spinal cord center point and the overall spatial consistency of the spinal cord centerline (i.e. the rostro-caudal continuity of the spinal cord). Additionally, a new post-processing feature, which aims to automatically split brain and spine regions is introduced, to be able to detect a consistent spinal cord centerline, independently from the field of view. We present data on the validation of the proposed algorithm, known as "OptiC", from a large dataset involving 20 centers, 4 contrasts (T₂-weighted n = 287, T₁-weighted n = 120, T₂^∗-weighted n = 307, diffusion-weighted n = 90), 501 subjects including 173 patients with a variety of neurologic diseases. Validation involved the gold-standard centerline coverage, the mean square error between the true and predicted centerlines and the ability to accurately separate brain and spine regions. Overall, OptiC was able to cover 98.77% of the gold-standard centerline, with a mean square error of 1.02 mm. OptiC achieved superior results compared to a state-of-the-art spinal cord localization technique based on the Hough transform, especially on pathological cases with an averaged mean square error of 1.08 mm vs. 13.16 mm (Wilcoxon signed-rank test p-value < .01). Images containing brain regions were identified with a 99% precision, on which brain and spine regions were separated with a distance error of 9.37 mm compared to ground-truth. Validation results on a challenging dataset suggest that OptiC could reliably be used for subsequent quantitative analyses tasks, opening the door to more robust analysis on pathological cases.

Spine labeling in MRI via regularized distribution matching

PURPOSE

This study investigates an efficient (nearly real-time) two-stage spine labeling algorithm that removes the need for an external training while being applicable to different types of MRI data and acquisition protocols.

METHODS

Based solely on the image being labeled (i.e., we do not use training data), the first stage aims at detecting potential vertebra candidates following the optimization of a functional containing two terms: (i) a distribution-matching term that encodes contextual information about the vertebrae via a density model learned from a very simple user input, which amounts to a point (mouse click) on a predefined vertebra; and (ii) a regularization constraint, which penalizes isolated candidates in the solution. The second stage removes false positives and identifies all vertebrae and discs by optimizing a geometric constraint, which embeds generic anatomical information on the interconnections between neighboring structures. Based on generic knowledge, our geometric constraint does not require external training.

RESULTS

We performed quantitative evaluations of the algorithm over a data set of 90 mid-sagittal MRI images of the lumbar spine acquired from 45 different subjects. To assess the flexibility of the algorithm, we used both T1- and T2-weighted images for each subject. A total of 990 structures were automatically detected/labeled and compared to ground-truth annotations by an expert. On the T2-weighted data, we obtained an accuracy of 91.6% for the vertebrae and 89.2% for the discs. On the T1-weighted data, we obtained an accuracy of 90.7% for the vertebrae and 88.1% for the discs.

CONCLUSION

Our algorithm removes the need for external training while being applicable to different types of MRI data and acquisition protocols. Based on the current testing data, a subject-specific model density and generic anatomical information, our method can achieve competitive performances when applied to T1- and T2-weighted MRI images.

Geometric modeling of hepatic arteries in 3D ultrasound with unsupervised MRA fusion during liver interventions

PURPOSE

Modulating the chemotherapy injection rate with regard to blood flow velocities in the tumor-feeding arteries during intra-arterial therapies may help improve liver tumor targeting while decreasing systemic exposure. These velocities can be obtained noninvasively using Doppler ultrasound (US). However, small vessels situated in the liver are difficult to identify and follow in US. We propose a multimodal fusion approach that non-rigidly registers a 3D geometric mesh model of the hepatic arteries obtained from preoperative MR angiography (MRA) acquisitions with intra-operative 3D US imaging.

METHODS

The proposed fusion tool integrates 3 imaging modalities: an arterial MRA, a portal phase MRA and an intra-operative 3D US. Preoperatively, the arterial phase MRA is used to generate a 3D model of the hepatic arteries, which is then non-rigidly co-registered with the portal phase MRA. Once the intra-operative 3D US is acquired, we register it with the portal MRA using a vessel-based rigid initialization followed by a non-rigid registration using an image-based metric based on linear correlation of linear combination. Using the combined non-rigid transformation matrices, the 3D mesh model is fused with the 3D US.

RESULTS

3D US and multi-phase MRA images acquired from 10 porcine models were used to test the performance of the proposed fusion tool. Unimodal registration of the MRA phases yielded a target registration error (TRE) of [Formula: see text] mm. Initial rigid alignment of the portal MRA and 3D US yielded a mean TRE of [Formula: see text] mm, which was significantly reduced to [Formula: see text] mm ([Formula: see text]) after affine image-based registration. The following deformable registration step allowed for further decrease of the mean TRE to [Formula: see text] mm.

CONCLUSION

The proposed tool could facilitate visualization and localization of these vessels when using 3D US intra-operatively for either intravascular or percutaneous interventions to avoid vessel perforation.

Topologically preserving straightening of spinal cord MRI

PURPOSE

To propose a robust and accurate method for straightening magnetic resonance (MR) images of the spinal cord, based on spinal cord segmentation, that preserves spinal cord topology and that works for any MRI contrast, in a context of spinal cord template-based analysis.

MATERIALS AND METHODS

The spinal cord curvature was computed using an iterative Non-Uniform Rational B-Spline (NURBS) approximation. Forward and inverse deformation fields for straightening were computed by solving analytically the straightening equations for each image voxel. Computational speed-up was accomplished by solving all voxel equation systems as one single system. Straightening accuracy (mean and maximum distance from straight line), computational time, and robustness to spinal cord length was evaluated using the proposed and the standard straightening method (label-based spline deformation) on 3T T₂ - and T₁ -weighted images from 57 healthy subjects and 33 patients with spinal cord compression due to degenerative cervical myelopathy (DCM).

RESULTS

The proposed algorithm was more accurate, more robust, and faster than the standard method (mean distance = 0.80 vs. 0.83 mm, maximum distance = 1.49 vs. 1.78 mm, time = 71 vs. 174 sec for the healthy population and mean distance = 0.65 vs. 0.68 mm, maximum distance = 1.28 vs. 1.55 mm, time = 32 vs. 60 sec for the DCM population).

CONCLUSION

A novel image straightening method that enables template-based analysis of quantitative spinal cord MRI data is introduced. This algorithm works for any MRI contrast and was validated on healthy and patient populations. The presented method is implemented in the Spinal Cord Toolbox, an open-source software for processing spinal cord MRI data.

LEVEL OF EVIDENCE

1 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2017;46:1209-1219.

SCT: Spinal Cord Toolbox, an open-source software for processing spinal cord MRI data

For the past 25 years, the field of neuroimaging has witnessed the development of several software packages for processing multi-parametric magnetic resonance imaging (mpMRI) to study the brain. These software packages are now routinely used by researchers and clinicians, and have contributed to important breakthroughs for the understanding of brain anatomy and function. However, no software package exists to process mpMRI data of the spinal cord. Despite the numerous clinical needs for such advanced mpMRI protocols (multiple sclerosis, spinal cord injury, cervical spondylotic myelopathy, etc.), researchers have been developing specific tools that, while necessary, do not provide an integrative framework that is compatible with most usages and that is capable of reaching the community at large. This hinders cross-validation and the possibility to perform multi-center studies. In this study we introduce the Spinal Cord Toolbox (SCT), a comprehensive software dedicated to the processing of spinal cord MRI data. SCT builds on previously-validated methods and includes state-of-the-art MRI templates and atlases of the spinal cord, algorithms to segment and register new data to the templates, and motion correction methods for diffusion and functional time series. SCT is tailored towards standardization and automation of the processing pipeline, versatility, modularity, and it follows guidelines of software development and distribution. Preliminary applications of SCT cover a variety of studies, from cross-sectional area measures in large databases of patients, to the precise quantification of mpMRI metrics in specific spinal pathways. We anticipate that SCT will bring together the spinal cord neuroimaging community by establishing standard templates and analysis procedures.

Computer-Assisted Screw Size and Insertion Trajectory Planning for Pedicle Screw Placement Surgery

Pathological conditions that cause instability of the spine are commonly treated by vertebral fixation involving pedicle screw placement surgery. However, existing methods for preoperative planning are based only on geometrical properties of vertebral structures (i.e., shape) without taking into account their structural properties (i.e., appearance). We propose a novel automated method for computer-assisted preoperative planning of the thoracic pedicle screw size and insertion trajectory. The proposed method extracts geometrical properties of vertebral structures by parametric modeling of vertebral bodies and pedicles in three dimensions (3D), and combines them with structural properties, evaluated through underlying image intensities in computed tomography (CT) images while considering the guidelines for pedicle screw design. The method was evaluated on 81 pedicles, obtained from 3D CT images of 11 patients that were appointed for pedicle screw placement surgery. In terms of mean absolute difference (MAD) and corresponding standard deviation (SD), the resulting high modeling accuracy of 0.39±0.31 mm for 3D vertebral body models and 0.31±0.25 mm for 3D pedicle models created an adequate anatomical frame for 3D pedicle screw models. When comparing the automatically obtained and manually defined plans for pedicle screw placement, a relatively high agreement was observed, with MAD ±SD of 0.4±0.4 mm for the screw diameter, 5.8±4.2 mm for the screw length, 2.0±1.4 mm for the pedicle crossing point and 7.6±5.8(°) for screw insertion angles. However, a statistically significant increase of 48±26% in the screw fastening strength in favor of the proposed automated method was observed in 99% of the cases.

Vessel-based registration of an optical shape sensing catheter for MR navigation

PURPOSE

Magnetic resonance navigation (MRN), achieved with an upgraded MRI scanner, aims to guide therapeutic nanoparticles from their release in the hepatic vascular network to embolize highly vascularized liver tumors. Visualizing the catheter in real-time within the arterial network is important for selective embolization within the MR gantry. To achieve this, a new MR-compatible catheter tracking technology based on optical shape sensing is used.

METHODS

This paper proposes a vessel-based registration pipeline to co-align this novel catheter tracking technology to the patient's diagnostic MR angiography (MRA) with 3D roadmapping. The method first extracts the 3D hepatic arteries from a diagnostic MRA based on concurrent deformable models, creating a detailed representation of the patient's internal anatomy. Once the optical shape sensing fibers, inserted in a double-lumen catheter, is guided into the hepatic arteries, the 3D centerline of the catheter is inferred and updated in real-time using strain measurements derived from fiber Bragg gratings sensors. Using both centerlines, a diffeomorphic registration based on a spectral representation of the high-level geometrical primitives is applied.

RESULTS

Results show promise in registration accuracy in five phantom models created from stereolithography of patient-specific vascular anatomies, with maximum target registration errors below 2 mm. Furthermore, registration accuracy with the shape sensing tracking technology remains insensitive to the magnetic field of the MR magnet.

CONCLUSIONS

This study demonstrates that an accurate registration procedure of a shape sensing catheter with diagnostic imaging is feasible.

Segmentation of the human spinal cord

Segmenting the spinal cord contour is a necessary step for quantifying spinal cord atrophy in various diseases. Delineating gray matter (GM) and white matter (WM) is also useful for quantifying GM atrophy or for extracting multiparametric MRI metrics into specific WM tracts. Spinal cord segmentation in clinical research is not as developed as brain segmentation, however with the substantial improvement of MR sequences adapted to spinal cord MR investigations, the field of spinal cord MR segmentation has advanced greatly within the last decade. Segmentation techniques with variable accuracy and degree of complexity have been developed and reported in the literature. In this paper, we review some of the existing methods for cord and WM/GM segmentation, including intensity-based, surface-based, and image-based methods. We also provide recommendations for validating spinal cord segmentation techniques, as it is important to understand the intrinsic characteristics of the methods and to evaluate their performance and limitations. Lastly, we illustrate some applications in the healthy and pathological spinal cord. One conclusion of this review is that robust and automatic segmentation is clinically relevant, as it would allow for longitudinal and group studies free from user bias as well as reproducible multicentric studies in large populations, thereby helping to further our understanding of the spinal cord pathophysiology and to develop new criteria for early detection of subclinical evolution for prognosis prediction and for patient management. Another conclusion is that at the present time, no single method adequately segments the cord and its substructure in all the cases encountered (abnormal intensities, loss of contrast, deformation of the cord, etc.). A combination of different approaches is thus advised for future developments, along with the introduction of probabilistic shape models. Maturation of standardized frameworks, multiplatform availability, inclusion in large suite and data sharing would also ultimately benefit to the community.

Translating state-of-the-art spinal cord MRI techniques to clinical use: A systematic review of clinical studies utilizing DTI, MT, MWF, MRS, and fMRI

BACKGROUND

A recent meeting of international imaging experts sponsored by the International Spinal Research Trust (ISRT) and the Wings for Life Foundation identified 5 state-of-the-art MRI techniques with potential to transform the field of spinal cord imaging by elucidating elements of the microstructure and function: diffusion tensor imaging (DTI), magnetization transfer (MT), myelin water fraction (MWF), MR spectroscopy (MRS), and functional MRI (fMRI). However, the progress toward clinical translation of these techniques has not been established.

METHODS

A systematic review of the English literature was conducted using MEDLINE, MEDLINE-in-Progress, Embase, and Cochrane databases to identify all human studies that investigated utility, in terms of diagnosis, correlation with disability, and prediction of outcomes, of these promising techniques in pathologies affecting the spinal cord. Data regarding study design, subject characteristics, MRI methods, clinical measures of impairment, and analysis techniques were extracted and tabulated to identify trends and commonalities. The studies were assessed for risk of bias, and the overall quality of evidence was assessed for each specific finding using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework.

RESULTS

A total of 6597 unique citations were identified in the database search, and after full-text review of 274 articles, a total of 104 relevant studies were identified for final inclusion (97% from the initial database search). Among these, 69 studies utilized DTI and 25 used MT, with both techniques showing an increased number of publications in recent years. The review also identified 1 MWF study, 11 MRS studies, and 8 fMRI studies. Most of the studies were exploratory in nature, lacking a priori hypotheses and showing a high (72%) or moderately high (20%) risk of bias, due to issues with study design, acquisition techniques, and analysis methods. The acquisitions for each technique varied widely across studies, rendering direct comparisons of metrics invalid. The DTI metric fractional anisotropy (FA) had the strongest evidence of utility, with moderate quality evidence for its use as a biomarker showing correlation with disability in several clinical pathologies, and a low level of evidence that it identifies tissue injury (in terms of group differences) compared with healthy controls. However, insufficient evidence exists to determine its utility as a sensitive and specific diagnostic test or as a tool to predict clinical outcomes. Very low quality evidence suggests that other metrics also show group differences compared with controls, including DTI metrics mean diffusivity (MD) and radial diffusivity (RD), the diffusional kurtosis imaging (DKI) metric mean kurtosis (MK), MT metrics MT ratio (MTR) and MT cerebrospinal fluid ratio (MTCSF), and the MRS metric of N-acetylaspartate (NAA) concentration, although these results were somewhat inconsistent.

CONCLUSIONS

State-of-the-art spinal cord MRI techniques are emerging with great potential to improve the diagnosis and management of various spinal pathologies, but the current body of evidence has only showed limited clinical utility to date. Among these imaging tools DTI is the most mature, but further work is necessary to standardize and validate its use before it will be adopted in the clinical realm. Large, well-designed studies with a priori hypotheses, standardized acquisition methods, detailed clinical data collection, and robust automated analysis techniques are needed to fully demonstrate the potential of these rapidly evolving techniques.

Fully automated segmentation of the cervical cord from T1-weighted MRI using PropSeg: Application to multiple sclerosis

Spinal cord (SC) atrophy, i.e. a reduction in the SC cross-sectional area (CSA) over time, can be measured by means of image segmentation using magnetic resonance imaging (MRI). However, segmentation methods have been limited by factors relating to reproducibility or sensitivity to change. The purpose of this study was to evaluate a fully automated SC segmentation method (PropSeg), and compare this to a semi-automated active surface (AS) method, in healthy controls (HC) and people with multiple sclerosis (MS). MRI data from 120 people were retrospectively analysed; 26 HC, 21 with clinically isolated syndrome, 26 relapsing remitting MS, 26 primary and 21 secondary progressive MS. MRI data from 40 people returning after one year were also analysed. CSA measurements were obtained within the cervical SC. Reproducibility of the measurements was assessed using the intraclass correlation coefficient (ICC). A comparison between mean CSA changes obtained with the two methods over time was performed using multivariate structural equation regression models. Associations between CSA measures and clinical scores were investigated using linear regression models. Compared to the AS method, the reproducibility of CSA measurements obtained with PropSeg was high, both in patients and in HC, with ICC > 0.98 in all cases. There was no significant difference between PropSeg and AS in terms of detecting change over time. Furthermore, PropSeg provided measures that correlated with physical disability, similar to the AS method. PropSeg is a time-efficient and reliable segmentation method, which requires no manual intervention, and may facilitate large multi-centre neuroprotective trials in progressive MS.

•

PropSeg is a fully automated segmentation method to measure cord atrophy in MS.

•

PropSeg offers comparable results to a widely used semi-automated method.

•

PropSeg is available for immediate clinical utility.