A priori free spectral unmixing with periodic absorbance changes: application for auto-calibrated intraoperative functional brain mapping.

Spectral unmixing designates techniques that allow to decompose measured spectra into linear or non-linear combination of spectra of all targets (endmembers). This technique was initially developed for satellite applications, but it is now also widely used in biomedical applications. However, several drawbacks limit the use of these techniques with standard optical devices like RGB cameras. The devices need to be calibrated and a a priori on the observed scene is often necessary. We propose a new method for estimating endmembers and their proportion automatically and without calibration of the acquisition device based on near separable non-negative matrix factorization. This method estimates the endmembers on spectra of absorbance changes presenting periodic events. This is very common in in vivo biomedical and medical optical imaging where hemodynamics dominate the absorbance fluctuations. We applied the method for identifying functional brain areas during neurosurgery using four different RGB cameras (an industrial camera, a smartphone and two surgical microscopes). Results obtained with the auto-calibration method were consistent with the intraoperative gold standards. Endmembers estimated with the auto-calibration method were similar to the calibrated endmembers used in the modified Beer-Lambert law. The similarity was particularly strong when both cardiac and respiratory periodic events were considered. This work can allow a widespread use of spectral imaging in the industrial or medical field.

Evaluation of deep learning models for quality control of MR spectra.

Purpose: While 3D MR spectroscopic imaging (MRSI) provides valuable spatial metabolic information, one of the hurdles for clinical translation is its interpretation, with voxel-wise quality control (QC) as an essential and the most time-consuming step. This work evaluates the accuracy of machine learning (ML) models for automated QC filtering of individual spectra from 3D healthy control and patient datasets. Methods: A total of 53 3D MRSI datasets from prior studies (30 neurological diseases, 13 brain tumors, and 10 healthy controls) were included in the study. Three ML models were evaluated: a random forest classifier (RF), a convolutional neural network (CNN), and an inception CNN (ICNN) along with two hybrid models: CNN + RF, ICNN + RF. QC labels used for training were determined manually through consensus of two MRSI experts. Normalized and cropped real-valued spectra was used as input. A cross-validation approach was used to separate datasets into training/validation/testing sets of aggregated voxels. Results: All models achieved a minimum AUC of 0.964 and accuracy of 0.910. In datasets from neurological disease and controls, the CNN model produced the highest AUC (0.982), while the RF model achieved the highest AUC in patients with brain tumors (0.976). Within tumor lesions, which typically exhibit abnormal metabolism, the CNN AUC was 0.973 while that of the RF was 0.969. Data quality inference times were on the order of seconds for an entire 3D dataset, offering drastic time reduction compared to manual labeling. Conclusion: ML methods accurately and rapidly performed automated QC. Results in tumors highlights the applicability to a variety of metabolic conditions.

Intraoperative identification of functional brain areas with RGB imaging using statistical parametric mapping: Simulation and clinical studies.

Complementary technique to preoperative fMRI and electrical brain stimulation (EBS) for glioma resection could improve dramatically the surgical procedure and patient care. Intraoperative RGB optical imaging is a technique for localizing functional areas of the human cerebral cortex that can be used during neurosurgical procedures. However, it still lacks robustness to be used with neurosurgical microscopes as a clinical standard. In particular, a robust quantification of biomarkers of brain functionality is needed to assist neurosurgeons. We propose a methodology to evaluate and optimize intraoperative identification of brain functional areas by RGB imaging. This consist in a numerical 3D brain model based on Monte Carlo simulations to evaluate intraoperative optical setups for identifying functional brain areas. We also adapted fMRI Statistical Parametric Mapping technique to identify functional brain areas in RGB videos acquired for 12 patients. Simulation and experimental results were consistent and showed that the intraoperative identification of functional brain areas is possible with RGB imaging using deoxygenated hemoglobin contrast. Optical functional identifications were consistent with those provided by EBS and preoperative fMRI. We also demonstrated that a halogen lighting may be particularity adapted for functional optical imaging. We showed that an RGB camera combined with a quantitative modeling of brain hemodynamics biomarkers can evaluate in a robust way the functional areas during neurosurgery and serve as a tool of choice to complement EBS and fMRI.

A Weakly Supervised Gradient Attribution Constraint for Interpretable Classification and Anomaly Detection.

The lack of interpretability of deep learning reduces understanding of what happens when a network does not work as expected and hinders its use in critical fields like medicine, which require transparency of decisions. For example, a healthy vs pathological classification model should rely on radiological signs and not on some training dataset biases. Several post-hoc models have been proposed to explain the decision of a trained network. However, they are very seldom used to enforce interpretability during training and none in accordance with the classification. In this paper, we propose a new weakly supervised method for both interpretable healthy vs pathological classification and anomaly detection. A new loss function is added to a standard classification model to constrain each voxel of healthy images to drive the network decision towards the healthy class according to gradient-based attributions. This constraint reveals pathological structures for patient images, allowing their unsupervised segmentation. Moreover, we advocate both theoretically and experimentally, that constrained training with the simple Gradient attribution is similar to constraints with the heavier Expected Gradient, consequently reducing the computational cost. We also propose a combination of attributions during the constrained training making the model robust to the attribution choice at inference. Our proposition was evaluated on two brain pathologies: tumors and multiple sclerosis. This new constraint provides a more relevant classification, with a more pathology-driven decision. For anomaly detection, the proposed method outperforms state-of-the-art especially on difficult multiple sclerosis lesions segmentation task with a 15 points Dice improvement.

Intraoperative Resting-State Functional Connectivity Based on RGB Imaging.

RGB optical imaging is a marker-free, contactless, and non-invasive technique that is able to monitor hemodynamic brain response following neuronal activation using task-based and resting-state procedures. Magnetic resonance imaging (fMRI) and functional near infra-red spectroscopy (fNIRS) resting-state procedures cannot be used intraoperatively but RGB imaging provides an ideal solution to identify resting-state networks during a neurosurgical operation. We applied resting-state methodologies to intraoperative RGB imaging and evaluated their ability to identify resting-state networks. We adapted two resting-state methodologies from fMRI for the identification of resting-state networks using intraoperative RGB imaging. Measurements were performed in 3 patients who underwent resection of lesions adjacent to motor sites. The resting-state networks were compared to the identifications provided by RGB task-based imaging and electrical brain stimulation. Intraoperative RGB resting-state networks corresponded to RGB task-based imaging (DICE:0.55±0.29). Resting state procedures showed a strong correspondence between them (DICE:0.66±0.11) and with electrical brain stimulation. RGB imaging is a relevant technique for intraoperative resting-state networks identification. Intraoperative resting-state imaging has several advantages compared to functional task-based analyses: data acquisition is shorter, less complex, and less demanding for the patients, especially for those unable to perform the tasks.

Automatic segmentation and location learning of neonatal cerebral ventricles in 3D ultrasound data combining CNN and CPPN.

Preterm neonates are highly likely to suffer from ventriculomegaly, a dilation of the Cerebral Ventricular System (CVS). This condition can develop into life-threatening hydrocephalus and is correlated with future neuro-developmental impairments. Consequently, it must be detected and monitored by physicians. In clinical routing, manual 2D measurements are performed on 2D ultrasound (US) images to estimate the CVS volume but this practice is imprecise due to the unavailability of 3D information. A way to tackle this problem would be to develop automatic CVS segmentation algorithms for 3D US data. In this paper, we investigate the potential of 2D and 3D Convolutional Neural Networks (CNN) to solve this complex task and propose to use Compositional Pattern Producing Network (CPPN) to enable Fully Convolutional Networks (FCN) to learn CVS location. Our database was composed of 25 3D US volumes collected on 21 preterm nenonates at the age of 35.8±1.6 gestational weeks. We found that the CPPN enables to encode CVS location, which increases the accuracy of the CNNs when they have few layers. Accuracy of the 2D and 3D FCNs reached intraobserver variability (IOV) in the case of dilated ventricles with Dice of 0.893±0.008 and 0.886±0.004 respectively (IOV = 0.898±0.008) and with volume errors of 0.45±0.42 cm3 and 0.36±0.24 cm3 respectively (IOV = 0.41±0.05 cm3). 3D FCNs were more accurate than 2D FCNs in the case of normal ventricles with Dice of 0.797±0.041 against 0.776±0.038 (IOV = 0.816±0.009) and volume errors of 0.35±0.29 cm3 against 0.35±0.24 cm3 (IOV = 0.2±0.11 cm3). The best segmentation time of volumes of size 320×320×320 was obtained by a 2D FCN in 3.5±0.2 s.

MRI-based radiomics to predict lipomatous soft tissue tumors malignancy: a pilot study.

OBJECTIVES: To develop and validate a MRI-based radiomic method to predict malignancies in lipomatous soft tissue tumors. METHODS: This retrospective study searched in the database of our pathology department, data from patients with lipomatous soft tissue tumors, with histology and gadolinium-contrast enhanced T1w MR images, obtained from 56 centers with non-uniform protocols. For each tumor, 87 radiomic features were extracted by two independent observers to evaluate the inter-observer reproducibility. A reduction of learning base dimension was performed from reproducibility and relevancy criteria. A model was subsequently prototyped using a linear support vector machine to predict malignant lesions. RESULTS: Eighty-one subjects with lipomatous soft tissue tumors including 40 lipomas and 41 atypical lipomatous tumors or well-differentiated liposarcomas with fat-suppressed T1w contrast enhanced MR images available were retrospectively enrolled. Based on a Pearson's correlation coefficient threshold at 0.8, 55 out of 87 (63.2%) radiomic features were considered reproducible. Further introduction of relevancy finally selected 35 radiomic features to be integrated in the model. To predict malignant tumors, model diagnostic performances were as follow: AUROC = 0.96; sensitivity = 100%; specificity = 90%; positive predictive value = 90.9%; negative predictive value = 100% and overall accuracy = 95.0%. CONCLUSION: This work demonstrates that radiomics allows to predict malignancy in soft tissue lipomatous tumors with routinely used MR acquisition in clinical oncology. These encouraging results need to be further confirmed in an external validation population.

Removing segmentation inconsistencies with semi-supervised non-adjacency constraint.

The advent of deep learning has pushed medical image analysis to new levels, rapidly replacing more traditional machine learning and computer vision pipelines. However segmenting and labelling anatomical regions remains challenging owing to appearance variations, imaging artifacts, the paucity and variability of annotated data, and the difficulty of fully exploiting domain constraints such as anatomical knowledge about inter-region relationships. We address the last point, improving the network's region-labeling consistency by introducing NonAdjLoss, an adjacency-graph based auxiliary training loss that penalizes outputs containing regions with anatomically-incorrect adjacency relationships. NonAdjLoss supports both fully-supervised training and a semi-supervised extension in which it is applied to unlabeled supplementary training data. The approach substantially reduces segmentation anomalies on the MICCAI-2012, IBSRv2 brain MRI datasets and the Anatomy3 whole body CT dataset, especially when semi-supervised training is included.

Automatic myocardial ischemic lesion detection on magnetic resonance perfusion weighted imaging prior perfusion quantification: A pre-modeling strategy.

Even if cardiovascular magnetic resonance (CMR) perfusion imaging has proven its relevance for visual detection of ischemia, myocardial blood flow (MBF) quantification at the voxel observation scale remains challenging. Integration of an automated segmentation step, prior to perfusion index estimation, might be a significant reconstruction component that could allow sustainable assumptions and constraint enlargement prior to advanced modeling. Current clustering techniques, such as bullseye representation or manual delineation, are not designed to discriminate voxels belonging to the lesion from healthy areas. Hence, the resulting average time-intensity curve, which is assumed to represent the dynamic contrast enhancement inside of a lesion, might be contaminated by voxels with perfectly healthy microcirculation. This study introduces a hierarchical lesion segmentation approach based on time-intensity curve features that considers the spatial particularities of CMR myocardial perfusion. A first k-means clustering approach enables this method to perform coarse clustering, which is refined by a novel spatiotemporal region-growing (STRG) segmentation, thus ensuring spatial and time-intensity curve homogeneity. Over a cohort of 30 patients, myocardial blood flow (MBF) measured in voxels of lesion regions detected with STRG was significantly lower than in regions drawn manually (mean difference = 0.14, 95% CI [0.07, 0.2]) and defined with the bullseye template (mean difference = 0.25, 95% CI [0.17, 0.36]). Over the 90 analyzed slices, the median Dice score calculated against the ground truth ranged between 0.62 and 0.67, the inclusion coefficients ranged between 0.62 and 0.76 and the centroid distances ranged between 0.97 and 3.88 mm. Therefore, though these metrics highlight spatial differences, they could not be used as an index to evaluate the accuracy and performance of the method, which can only be attested by the variability of the MBF clinical index.

Repetitive motion compensation for real time intraoperative video processing.

In this paper, we present a motion compensation algorithm dedicated to video processing during neurosurgery. After craniotomy, the brain surface undergoes a repetitive motion due to the cardiac pulsation. This motion as well as potential video camera motion prevent accurate video analysis. We propose a dedicated motion model where the brain deformation is described using a linear basis learned from a few initial frames of the video. As opposed to other works using linear basis for the flow, the camera motion is explicitly accounted in the transformation model. Despite the nonlinear nature of our model, all the motion parameters are robustly estimated all at once, using only one singular value decomposition (SVD), making our procedure computationally efficient. A Lagrangian specification of the flow field ensures the stability of the method. Experiments on in vivo data are presented to evaluate the capacity of the method to cope with occlusion or camera motion. The method we propose satisfies the intraoperative constraints: it is robust to surgical tools occlusions, it works in real time, and it is able to handle large camera viewpoint changes.

Intraoperative quantitative functional brain mapping using an RGB camera.

Intraoperative optical imaging is a localization technique for the functional areas of the human brain cortex during neurosurgical procedures. However, it still lacks robustness to be used as a clinical standard. In particular, new biomarkers of brain functionality with improved sensitivity and specificity are needed. We present a method for the computation of hemodynamics-based functional brain maps using an RGB camera and a white light source. We measure the quantitative oxy and deoxyhemoglobin concentration changes in the human brain cortex with the modified Beer-Lambert law and Monte Carlo simulations. A functional model has been implemented to evaluate the functional brain areas following neuronal activation by physiological stimuli. The results show a good correlation between the computed quantitative functional maps and the brain areas localized by electrical brain stimulation (EBS). We demonstrate that an RGB camera combined with a quantitative modeling of brain hemodynamics biomarkers can evaluate in a robust way the functional areas during neurosurgery and serve as a tool of choice to complement EBS.

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.

Reproducibility of in vivo magnetic resonance imaging T1 rho and T2 relaxation time measurements of hip cartilage at 3.0T in healthy volunteers.

PURPOSE: To assess the T1 ρ and T2 values in the hip cartilage of healthy volunteers and to evaluate the reproducibility of these measurements. MATERIALS AND METHODS: The right hip joint of 30 asymptomatic volunteers was explored with 3T magnetic resonance imaging (MRI). Quantitative 3D T1 ρ- and T2 -maps sequences were repeated twice with a 30-minute delay (immediate reproducibility). The same protocol was repeated 14 days later (short-term reproducibility). Immediate and short-term reproducibility were estimated using coefficients of variation and correlation concordance coefficients (CCC). The precisions of the measurements were estimated by the ratio of the standard deviations. A mixed linear model was used to analyze the effect of patient's characteristics on T1 ρ and T2 values. RESULTS: Immediate reproducibility was significantly better than short-term reproducibility for T1 ρ (CCC of 0.75 versus 0.55; P = 0.007) and T2 (CCC 0.65 versus 0.32; P < 0.001). The precisions of the measurements were estimated between 5.5% and 9.1%. Median T1 ρ values were 6.0 msec higher in women than in men (P = 0.006), with no significant influence of age, body mass index (BMI), or sports activity. Median T2 values were not significantly different between men and women (0.4 msec lower in women; P = 0.76). There was no significant influence of age, BMI, or sports activity. T1 ρ and T2 values were lower in lateral regions than in medial regions (4.9 msec and 2.5 msec lower respectively; P < 0.0001). CONCLUSION: Immediate reproducibility of T1 ρ and T2 values is better than short-term, with limited effect of 30 minutes decubitus. T1 ρ values are significantly higher in women. LEVEL OF EVIDENCE: 2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2018;47:1022-1033.

Individual muscle segmentation in MR images: A 3D propagation through 2D non-linear registration approaches.

Manual and automated segmentation of individual muscles in magnetic resonance images have been recognized as challenging given the high variability of shapes between muscles and subjects and the discontinuity or lack of visible boundaries between muscles. In the present study, we proposed an original algorithm allowing a semi-automatic transversal propagation of manually-drawn masks. Our strategy was based on several ascending and descending non-linear registration approaches which is similar to the estimation of a Lagrangian trajectory applied to manual masks. Using several manually-segmented slices, we have evaluated our algorithm on the four muscles of the quadriceps femoris group. We mainly showed that our 3D propagated segmentation was very accurate with an averaged Dice similarity coefficient value higher than 0.91 for the minimal manual input of only two manually-segmented slices.

Multi-atlas-based fully automatic segmentation of individual muscles in rat leg.

OBJECTIVE: To quantify individual muscle volume in rat leg MR images using a fully automatic multi-atlas-based segmentation method. MATERIALS AND METHODS: We optimized a multi-atlas-based segmentation method to take into account the voxel anisotropy of numbers of MRI acquisition protocols. We mainly tested an image upsampling process along Z and a constraint on the nonlinear deformation in the XY plane. We also evaluated a weighted vote procedure and an original implementation of an artificial atlas addition. Using this approach, we measured gastrocnemius and plantaris muscle volumes and compared the results with manual segmentation. The method reliability for volume quantification was evaluated using the relative overlap index. RESULTS: The most accurate segmentation was obtained using a nonlinear registration constrained in the XY plane by zeroing the Z component of the displacement and a weighted vote procedure for both muscles regardless of the number of atlases. The performance of the automatic segmentation and the corresponding volume quantification outperformed the interoperator variability using a minimum of three original atlases. CONCLUSION: We demonstrated the reliability of a multi-atlas segmentation approach for the automatic segmentation and volume quantification of individual muscles in rat leg and found that constraining the registration in plane significantly improved the results.

Enhancing atlas based segmentation with multiclass linear classifiers.

PURPOSE: To present a method to enrich atlases for atlas based segmentation. Such enriched atlases can then be used as a single atlas or within a multiatlas framework. METHODS: In this paper, machine learning techniques have been used to enhance the atlas based segmentation approach. The enhanced atlas defined in this work is a pair composed of a gray level image alongside an image of multiclass classifiers with one classifier per voxel. Each classifier embeds local information from the whole training dataset that allows for the correction of some systematic errors in the segmentation and accounts for the possible local registration errors. The authors also propose to use these images of classifiers within a multiatlas framework: results produced by a set of such local classifier atlases can be combined using a label fusion method. RESULTS: Experiments have been made on the in vivo images of the IBSR dataset and a comparison has been made with several state-of-the-art methods such as FreeSurfer and the multiatlas nonlocal patch based method of Coupé or Rousseau. These experiments show that their method is competitive with state-of-the-art methods while having a low computational cost. Further enhancement has also been obtained with a multiatlas version of their method. It is also shown that, in this case, nonlocal fusion is unnecessary. The multiatlas fusion can therefore be done efficiently. CONCLUSIONS: The single atlas version has similar quality as state-of-the-arts multiatlas methods but with the computational cost of a naive single atlas segmentation. The multiatlas version offers a improvement in quality and can be done efficiently without a nonlocal strategy.

A reliable spatially normalized template of the human spinal cord--Applications to automated white matter/gray matter segmentation and tensor-based morphometry (TBM) mapping of gray matter alterations occurring with age.

Recently, a T2*-weighted template and probabilistic atlas of the white and gray matter (WM, GM) of the spinal cord (SC) have been reported. Such template can be used as tissue-priors for automated WM/GM segmentation but can also provide a common reference and normalized space for group studies. Here, a new template has been created (AMU40), and accuracy of automatic template-based WM/GM segmentation was quantified. The feasibility of tensor-based morphometry (TBM) for studying voxel-wise morphological differences of SC between young and elderly healthy volunteers was also investigated. Sixty-five healthy subjects were divided into young (n=40, age<40years old, mean age 28±5years old) and elderly (n=25, age>50years old, mean age 57±5years old) groups and scanned at 3T using an axial high-resolution T2*-weighted sequence. Inhomogeneity correction and affine intensity normalization of the SC and cerebrospinal fluid (CSF) signal intensities across slices were performed prior to both construction of the AMU40 template and WM/GM template-based segmentation. The segmentation was achieved using non-linear spatial normalization of T2*-w MR images to the AMU40 template. Validation of WM/GM segmentations was performed with a leave-one-out procedure by calculating DICE similarity coefficients between manual and automated WM/GM masks. SC morphological differences between young and elderly healthy volunteers were assessed using the same non-linear spatial normalization of the subjects' MRI to a common template, derivation of the Jacobian determinant maps from the warping fields, and a TBM analysis. Results demonstrated robust WM/GM automated segmentation, with mean DICE values greater than 0.8. Concerning the TBM analysis, an anterior GM atrophy was highlighted in elderly volunteers, demonstrating thereby, for the first time, the feasibility of studying local structural alterations in the SC using tensor-based morphometry. This holds great promise for studies of morphological impairment occurring in several central nervous system pathologies.

Fast multidimensional NMR spectroscopy for sparse spectra.

Multidimensional NMR spectroscopy is widely used for studies of molecular and biomolecular structure. A major disadvantage of multidimensional NMR is the long acquisition time which, regardless of sensitivity considerations, may be needed to obtain the final multidimensional frequency domain coefficients. In this article, a method for under-sampling multidimensional NMR acquisition of sparse spectra is presented. The approach is presented in the case of two-dimensional NMR acquisitions. It relies on prior knowledge about the support in the two-dimensional frequency domain to recover an over-determined system from the under-determined system induced in the linear acquisition model when under-sampled acquisitions are performed. This over-determined system can then be solved with linear least squares. The prior knowledge is obtained efficiently at a low cost from the one-dimensional NMR acquisition, which is generally acquired as a first step in multidimensional NMR. If this one-dimensional acquisition is intrinsically sparse, it is possible to reconstruct the corresponding two-dimensional acquisition from far fewer observations than those imposed by the Nyquist criterion, and subsequently to reduce the acquisition time. Further improvements are obtained by optimizing the sampling procedure for the least-squares reconstruction using the sequential backward selection algorithm. Theoretical and experimental results are given in the case of a traditional acquisition scheme, which demonstrate reliable and fast reconstructions with acceleration factors in the range 3-6. The proposed method outperforms the CS methods (OMP, L1) in terms of the reconstruction performance, implementation and computation time. The approach can be easily extended to higher dimensions and spectroscopic imaging.

Construction of an in vivo human spinal cord atlas based on high-resolution MR images at cervical and thoracic levels: preliminary results.

OBJECT: Our goal was to build a probabilistic atlas and anatomical template of the human cervical and thoracic spinal cord (SC) that could be used for segmentation algorithm improvement, parametric group studies, and enrichment of biomechanical modelling. MATERIALS AND METHODS: High-resolution axial T2*-weighted images were acquired at 3T on 15 healthy volunteers using a multi-echo-gradient-echo sequence (1 slice per vertebral level from C1 to L2). After manual segmentation, linear and affine co-registrations were performed providing either inter-individual morphometric variability maps, or substructure probabilistic maps [CSF, white and grey matter (WM/GM)] and anatomical SC template. RESULTS: The larger inter-individual morphometric variations were observed at the thoraco-lumbar levels and in the posterior GM. Mean SC diameters were in agreement with the literature and higher than post-mortem measurements. A representative SC MR template was generated and values up to 90 and 100% were observed on GM and WM-probability maps. CONCLUSION: This work provides a probabilistic SC atlas and a template that could offer great potentialities for parametrical MRI analysis (DTI/MTR/fMRI) and group studies, similar to what has already been performed using a brain atlas. It also offers great perspective for biomechanical models usually based on post-mortem or generic data. Further work will consider integration into an automated SC segmentation pipeline.