Texture Analysis for Automatic Segmentation of Intervertebral Disks of Scoliotic Spines From MR Images

Uncertainty Estimation in Unsupervised MR-CT Synthesis of Scoliotic Spines

Uncertainty estimations through approximate Bayesian inference provide interesting insights to deep neural networks' behavior. In unsupervised learning tasks, where expert labels are unavailable, it becomes ever more important to critique the model through uncertainties. This paper presents a proof-of-concept for generalizing the aleatoric and epistemic uncertainties in unsupervised MR-CT synthesis of scoliotic spines. A novel adaptation of the cycle-consistency constraint in CycleGAN is proposed such that the model predicts the aleatoric uncertainty maps in addition to the standard volume-to-volume translation between Magnetic Resonance (MR) and Computed Tomography (CT) data. Ablation experiments were performed to understand uncertainty estimation as an implicit regularizer and a measure of the model's confidence. The aleatoric uncertainty helps in distinguishing between the bone and soft-tissue regions in CT and MR data during translation, while the epistemic uncertainty provides interpretable information to the user for downstream tasks.

Lumbar intervertebral disc segmentation for computer modeling and simulation

BACKGROUND AND OBJECTIVE

The present work had as its main objective the development of a method for localizing and automatically segmenting lumbar intervertebral discs (IVD) in 3D from magnetic resonance imaging (MRI), with the goal of supporting the generation of finite element (FE) models from actual lumbar spine anatomy, by providing accurate and personalized information on the shape of the patient's IVD. The extension of the method to allow performing separate segmentations of the IVD's two main structures - annulus fibrosus (AF) and nucleus pulposus (NP) - as well as automatically detecting degenerated IVD where this distinction is no longer possible was also an objective of the work.

METHODS

The method presented here evolves from 2D segmentations in the sagittal profile using Gabor filters towards 3D segmentations. It works by detecting the spine curves and intensity regions corresponding to IVD. As so, the 2D method from Zhu et al. (2013) was partially implemented, modified and adapted to 3D use, and then tested with eight spines from two separated online datasets. The 3D adaptation was achieved by using vertebral body segmentation masks to approximate the shape of the vertebrae and to adjust the spine curves accordingly.

RESULTS

The method showed average values of 85%, 83% and 96% for the Dice coefficient, sensitivity and specificity, respectively. The method correctly identified 65 of 68 (96%) IVD as either healthy or degenerated. The method's Dice coefficient is within the range of existing 3D IVD segmentation methods in the literature (81-92%). The method took on average 6-7 s to perform a full 3D segmentation, which is well within the range of the existing methods (2 s - 19 min).

CONCLUSIONS

The developed method can be used to generate accurate 3D models of the IVD based on MRI, with AF/NP distinction and detection of marked degeneration by comparing each IVD with the remaining spine levels. Further work shall improve the method towards distinguishing between specific levels of degeneration for clinically oriented FE modeling.

Automatic Segmentation of Lumbar Spine MRI Images Based on Improved Attention U-Net

Lumbar spine segmentation is important to help doctors diagnose lumbar disc herniation (LDH) and patients' rehabilitation treatment. In order to accurately segment the lumbar spine, a lumbar spine image segmentation algorithm based on improved Attention U-Net is proposed. The algorithm is based on Attention U-Net, the attention module based on multilevel feature map fusion is adopted, two residual modules are introduced instead of the original convolution blocks. a hybrid loss function is used for prediction during the training process, and finally, the image superposition process is realized. In this experiment, we expanded 420 lumbar MRI images of 180 patients to 1000 images and trained them by different algorithms, respectively, and accuracy, recall, and Dice similarity coefficient metrics were used to analyze these algorithms. The results show that compared with SVM, FCN, R-CNN, U-Net, and Attention U-Net models, the improved model achieved better results in all three evaluations, with 95.50%, 94.53%, and 95.01%, respectively, which proves the better performance of the proposed method for segmentation in lumbar disc and caudal vertebrae.

Use of machine learning to select texture features in investigating the effects of axial loading on T2-maps from magnetic resonance imaging of the lumbar discs

BACKGROUND

Recent advances in texture analysis and machine learning offer new opportunities to improve the application of imaging to intervertebral disc biomechanics. This study employed texture analysis and machine learning on MRIs to investigate the lumbar disc's response to loading.

METHODS

Thirty-five volunteers (30 (SD 11) yrs.) with and without chronic back pain spent 20 min lying in a relaxed unloaded supine position, followed by 20 min loaded in compression, and then 20 min with traction applied. T₂-weighted MR images were acquired during the last 5 min of each loading condition. Custom image analysis software was used to segment discs from adjacent tissues semi-automatically and segment each disc into the nucleus, anterior and posterior annulus automatically. A grey-level, co-occurrence matrix with one to four pixels offset in four directions (0°, 45°, 90° and 135°) was then constructed (320 feature/tissue). The Random Forest Algorithm was used to select the most promising classifiers. Linear mixed-effect models and Cohen's d compared loading conditions.

FINDINGS

All statistically significant differences (p < 0.001) were observed in the nucleus and posterior annulus in the 135° offset direction at the L4-5 level between lumbar compression and traction. Correlation (P_2-Offset, P_4-Offset) and information measure of correlation 1 (P_3-Offset, P_4-Offset) detected significant changes in the nucleus. Statistically significant changes were also observed for homogeneity (P_2-Offset, P_3-Offset), contrast (P_2-Offset), and difference variance (P_4-Offset) of the posterior annulus.

INTERPRETATION

MRI textural features may have the potential of identifying the disc's response to loading, particularly in the nucleus and posterior annulus, which appear most sensitive to loading.

LEVEL OF EVIDENCE

Diagnostic: individual cross-sectional studies with consistently applied reference standard and blinding.

Vertebrae, IVD and spinal canal boundary extraction on MRI, utilizing CT-trained active shape models

PURPOSE

Vertebrae, intervertebral disc (IVD) and spinal canal (SC) displacements are in the root of several spinal cord pathologies. The localization and boundary extraction of these structures, along with the quantification of their displacements, provide valuable clues for assessing each pathological condition. In this work, we propose a computational method for boundary extraction of vertebrae, IVD and SC in magnetic resonance images (MRI).

METHOD

Vertebrae shape priors derived from computed tomography (CT) images are used to guide vertebrae, IVD and SC boundary extraction in MRI. This strategy is dictated by three considerations: (1) CT is the modality of choice for highlighting solid structures such as vertebrae, (2) vertebrae boundaries indirectly impose constraints on the boundaries of neighbouring structures (IVD and SC), and (3) it can be observed that edges are similarly located in CT and MR images; therefore, gradient profiles and shape priors learned by active shape models (ASMs) from CT are also valid in MRI.

RESULTS

Experimental comparisons on two MR image datasets demonstrate that the proposed approach obtains segmentation results, which are comparable to the state of the art. Moreover, the adopted bimodal strategy is validated by demonstrating that CT-derived shape priors lead to more accurate boundary extraction than MRI-derived shape priors, even in the case of MR image applications.

CONCLUSION

Unlike existing bimodal methods, the proposed one is not dependent on the availability of CT/MR image pairs, which are not usually acquired from the same patient. In addition, unlike state-of-the-art deep learning-based methods, it is not dependent on large amounts of training data. The proposed method requires a limited amount of user intervention.

Deep neural network for automated simultaneous intervertebral disc (IVDs) identification and segmentation of multi-modal MR images

BACKGROUND AND OBJECTIVE

Lower back pain in humans has become a major risk. Classical approaches follow a non-invasive imaging technique for the assessment of spinal intervertebral disc (IVDs) abnormalities, where identification and segmentation of discs are done separately, making it a time-consuming phenomenon. This necessitates designing a robust automated and simultaneous IVDs identification and segmentation of multi-modality MRI images.

METHODS

We introduced a novel deep neural network architecture coined as 'RIMNet', a Region-to-Image Matching Network model, capable of performing an automated and simultaneous IVDs identification and segmentation of MRI images. The multi-modal input data is being fed to the network with a dropout strategy, by randomly disabling modalities in mini-batches. The performance accuracy as a function of the testing dataset was determined. The execution of the deep neural network model was evaluated by computing the IVDs Identification Accuracy, Dice coefficient, MDOC, Average Symmetric Surface Distance, Jaccard Coefficient, Hausdorff Distance and F1 Score.

RESULTS

Proposed model has attained 94% identification accuracy, dice coefficient value of 91.7±1% in segmentation and MDOC 90.2±1%. Our model also achieved 0.87±0.02 for Jaccard Coefficient, 0.54±0.04 for ASD and 0.62±0.02 mm Hausdorff Distance. The results have been validated and compared with other methodologies on dataset of MICCAI IVD 2018 challenge.

CONCLUSIONS

Our proposed deep-learning methodology is capable of performing simultaneous identification and segmentation on IVDs MRI images of the human spine with high accuracy.

Bimodal CT/MRI-Based Segmentation Method for Intervertebral Disc Boundary Extraction

Intervertebral disc (IVD) localization and segmentation have triggered intensive research efforts in the medical image analysis community, since IVD abnormalities are strong indicators of various spinal cord-related pathologies. Despite the intensive research efforts to address IVD boundary extraction based on MR images, the potential of bimodal approaches, which benefit from complementary information derived from both magnetic resonance imaging (MRI) and computed tomography (CT), has not yet been fully realized. Furthermore, most existing approaches rely on manual intervention or on learning, although sufficiently large and labelled 3D datasets are not always available. In this light, this work introduces a bimodal segmentation method for vertebrae and IVD boundary extraction, which requires a limited amount of intervention and is not based on learning. The proposed method comprises various image processing and analysis stages, including CT/MRI registration, Otsu-based thresholding and Chan–Vese-based segmentation. The method was applied on 98 expert-annotated pairs of CT and MR spinal cord images with varying slice thicknesses and pixel sizes, which were obtained from 7 patients using different scanners. The experimental results had a Dice similarity coefficient equal to 94.77(%) for CT and 86.26(%) for MRI and a Hausdorff distance equal to 4.4 pixels for CT and 4.5 pixels for MRI. Experimental comparisons with state-of-the-art CT and MRI segmentation methods lead to the conclusion that the proposed method provides a reliable alternative for vertebrae and IVD boundary extraction. Moreover, the segmentation results are utilized to perform a bimodal visualization of the spine, which could potentially aid differential diagnosis with respect to several spine-related pathologies.

Automatic Spine Tissue Segmentation from MRI Data Based on Cascade of Boosted Classifiers and Active Appearance Model

The study introduces a novel method for automatic segmentation of vertebral column tissue from MRI images. The paper describes a method that combines multiple stages of Machine Learning techniques to recognize and separate different tissues of the human spine. For the needs of this paper, 50 MRI examinations presenting lumbosacral spine of patients with low back pain were selected. After the initial filtration, automatic vertebrae recognition using Cascade Classifier takes place. Afterwards the main segmentation process using the patch based Active Appearance Model is performed. Obtained results are interpolated using centripetal Catmull–Rom splines. The method was tested on previously unseen vertebrae images segmented manually by 5 physicians. A test validating algorithm convergence per iteration was performed and the Intraclass Correlation Coefficient was calculated. Additionally, the 10-fold cross-validation analysis has been done. Presented method proved to be comparable to the physicians (FF = 90.19 ± 1.01%). Moreover results confirmed a proper algorithm convergence. Automatically segmented area correlated well with manual segmentation for single measurements (r¯=0.8336) and for average measurements (r¯=0.9068) with p = 0.05. The 10-fold cross-validation analysis (FF = 91.37 ± 1.13%) confirmed a good model generalization resulting in practical performance.

Multi-Parameter Ensemble Learning for Automated Vertebral Body Segmentation in Heterogeneously Acquired Clinical MR Images

The development of quantitative imaging biomarkers in medicine requires automatic delineation of relevant anatomical structures using available imaging data. However, this task is complicated in clinical medicine due to the variation in scanning parameters and protocols, even within a single medical center. Existing literature on automatic image segmentation using MR data is based on the analysis of highly homogenous images obtained using a fixed set of pulse sequence parameters (TR/TE). Unfortunately, algorithms that operate on fixed scanning parameters do not avail themselves to real-world daily clinical use due to the existing variation in scanning parameters and protocols. Thus, it is necessary to develop algorithmic techniques that can address the challenge of MR image segmentation using real clinical data. Toward this goal, we developed a multi-parametric ensemble learning technique to automatically detect and segment lumbar vertebral bodies using MR images of the spine. We use spine imaging data to illustrate our techniques since low back pain is an extremely common condition and a typical spine clinic evaluates patients that have been referred with a wide range of scanning parameters. This method was designed with special emphasis on robustness so that it can perform well despite the inherent variation in scanning protocols. Specifically, we show how a single multi-parameter ensemble model trained with manually labeled T2 scans can autonomously segment vertebral bodies on scans with echo times varying between 24 and 147 ms and relaxation times varying between 1500 and 7810 ms. Furthermore, even though the model was trained using T2-MR imaging data, it can accurately segment vertebral bodies on T1-MR and CT, further demonstrating the robustness and versatility of our methodology. We believe that robust segmentation techniques, such as the one presented here, are necessary for translating computer assisted diagnosis into everyday clinical practice.

Evaluation and comparison of 3D intervertebral disc localization and segmentation methods for 3D T2 MR data: A grand challenge

The evaluation of changes in Intervertebral Discs (IVDs) with 3D Magnetic Resonance (MR) Imaging (MRI) can be of interest for many clinical applications. This paper presents the evaluation of both IVD localization and IVD segmentation methods submitted to the Automatic 3D MRI IVD Localization and Segmentation challenge, held at the 2015 International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI2015) with an on-site competition. With the construction of a manually annotated reference data set composed of 25 3D T2-weighted MR images acquired from two different studies and the establishment of a standard validation framework, quantitative evaluation was performed to compare the results of methods submitted to the challenge. Experimental results show that overall the best localization method achieves a mean localization distance of 0.8 mm and the best segmentation method achieves a mean Dice of 91.8%, a mean average absolute distance of 1.1 mm and a mean Hausdorff distance of 4.3 mm, respectively. The strengths and drawbacks of each method are discussed, which provides insights into the performance of different IVD localization and segmentation methods.

Regression Segmentation for M³ Spinal Images

Clinical routine often requires to analyze spinal images of multiple anatomic structures in multiple anatomic planes from multiple imaging modalities (M(3)). Unfortunately, existing methods for segmenting spinal images are still limited to one specific structure, in one specific plane or from one specific modality (S(3)). In this paper, we propose a novel approach, Regression Segmentation, that is for the first time able to segment M(3) spinal images in one single unified framework. This approach formulates the segmentation task innovatively as a boundary regression problem: modeling a highly nonlinear mapping function from substantially diverse M(3) images directly to desired object boundaries. Leveraging the advancement of sparse kernel machines, regression segmentation is fulfilled by a multi-dimensional support vector regressor (MSVR) which operates in an implicit, high dimensional feature space where M(3) diversity and specificity can be systematically categorized, extracted, and handled. The proposed regression segmentation approach was thoroughly tested on images from 113 clinical subjects including both disc and vertebral structures, in both sagittal and axial planes, and from both MRI and CT modalities. The overall result reaches a high dice similarity index (DSI) 0.912 and a low boundary distance (BD) 0.928 mm. With our unified and expendable framework, an efficient clinical tool for M(3) spinal image segmentation can be easily achieved, and will substantially benefit the diagnosis and treatment of spinal diseases.

Localization and Segmentation of 3D Intervertebral Discs in MR Images by Data Driven Estimation

This paper addresses the problem of fully-automatic localization and segmentation of 3D intervertebral discs (IVDs) from MR images. Our method contains two steps, where we first localize the center of each IVD, and then segment IVDs by classifying image pixels around each disc center as foreground (disc) or background. The disc localization is done by estimating the image displacements from a set of randomly sampled 3D image patches to the disc center. The image displacements are estimated by jointly optimizing the training and test displacement values in a data-driven way, where we take into consideration both the training data and the geometric constraint on the test image. After the disc centers are localized, we segment the discs by classifying image pixels around disc centers as background or foreground. The classification is done in a similar data-driven approach as we used for localization, but in this segmentation case we are aiming to estimate the foreground/background probability of each pixel instead of the image displacements. In addition, an extra neighborhood smooth constraint is introduced to enforce the local smoothness of the label field. Our method is validated on 3D T2-weighted turbo spin echo MR images of 35 patients from two different studies. Experiments show that compared to state of the art, our method achieves better or comparable results. Specifically, we achieve for localization a mean error of 1.6-2.0 mm, and for segmentation a mean Dice metric of 85%-88% and a mean surface distance of 1.3-1.4 mm.

Statistical shape model reconstruction with sparse anomalous deformations: Application to intervertebral disc herniation

Many medical image processing techniques rely on accurate shape modeling of anatomical features. The presence of shape abnormalities challenges traditional processing algorithms based on strong morphological priors. In this work, a sparse shape reconstruction from a statistical shape model is presented. It combines the advantages of traditional statistical shape models (defining a 'normal' shape space) and previously presented sparse shape composition (providing localized descriptors of anomalies). The algorithm was incorporated into our image segmentation and classification software. Evaluation was performed on simulated and clinical MRI data from 22 sciatica patients with intervertebral disc herniation, containing 35 herniated and 97 normal discs. Moderate to high correlation (R=0.73) was achieved between simulated and detected herniations. The sparse reconstruction provided novel quantitative features describing the herniation morphology and MRI signal appearance in three dimensions (3D). The proposed descriptors of local disc morphology resulted to the 3D segmentation accuracy of 1.07±1.00mm (mean absolute vertex-to-vertex mesh distance over the posterior disc region), and improved the intervertebral disc classification from 0.888 to 0.931 (area under receiver operating curve). The results show that the sparse shape reconstruction may improve computer-aided diagnosis of pathological conditions presenting local morphological alterations, as seen in intervertebral disc herniation.

Validity and reliability of computerized measurement of lumbar intervertebral disc height and volume from magnetic resonance images

BACKGROUND CONTEXT

Magnetic resonance (MR) examinations of morphologic characteristics of intervertebral discs (IVDs) have been used extensively for biomechanical studies and clinical investigations of the lumbar spine. Traditionally, the morphologic measurements have been performed using time- and expertise-intensive manual segmentation techniques not well suited for analyses of large-scale studies..

PURPOSE

The purpose of this study is to introduce and validate a semiautomated method for measuring IVD height and mean sagittal area (and volume) from MR images to determine if it can replace the manual assessment and enable analyses of large MR cohorts.

STUDY DESIGN/SETTING

This study compares semiautomated and manual measurements and assesses their reliability and agreement using data from repeated MR examinations.

METHODS

Seven healthy asymptomatic males underwent 1.5-T MR examinations of the lumbar spine involving sagittal T2-weighted fast spin-echo images obtained at baseline, pre-exercise, and postexercise conditions. Measures of the mean height and the mean sagittal area of lumbar IVDs (L1-L2 to L4-L5) were compared for two segmentation approaches: a conventional manual method (10-15 minutes to process one IVD) and a specifically developed semiautomated method (requiring only a few mouse clicks to process each subject).

RESULTS

Both methods showed strong test-retest reproducibility evaluated on baseline and pre-exercise examinations with strong intraclass correlations for the semiautomated and manual methods for mean IVD height (intraclass correlation coefficient [ICC]=0.99, 0.98) and mean IVD area (ICC=0.98, 0.99), respectively. A bias (average deviation) of 0.38 mm (4.1%, 95% confidence interval 0.18-0.59 mm) was observed between the manual and semiautomated methods for the IVD height, whereas there was no statistically significant difference for the mean IVD area (0.1%±3.5%). The semiautomated and manual methods both detected significant exercise-induced changes in IVD height (0.20 and 0.28 mm) and mean IVD area (5.7 and 8.3 mm(2)), respectively.

CONCLUSIONS

The presented semiautomated method provides an alternative to time- and expertise-intensive manual procedures for analysis of larger, cross-sectional, interventional, and longitudinal MR studies for morphometric analyses of lumbar IVDs.

Human intervertebral disc stiffness correlates better with the Otsu threshold computed from axial T2 map of its posterior annulus fibrosus than with clinical classifications

Degeneration of the intervertebral disc, sometimes associated with low back pain and abnormal spinal motions, represents a major health issue with high costs. A non-invasive degeneration assessment via qualitative or quantitative MRI (magnetic resonance imaging) is possible, yet, no relation between mechanical properties and T2 maps of the intervertebral disc (IVD) has been considered, albeit T2 relaxation time values quantify the degree of degeneration. Therefore, MRI scans and mechanical tests were performed on 14 human lumbar intervertebral segments freed from posterior elements and all soft tissues excluding the IVD. Degeneration was evaluated in each specimen using morphological criteria, qualitative T2 weighted images and quantitative axial T2 map data and stiffness was calculated from the load-deflection curves of in vitro compression, torsion, lateral bending and flexion/extension tests. In addition to mean T2, the OTSU threshold of T2 (TOTSU), a robust and automatic histogram-based method that computes the optimal threshold maximizing the distinction of two classes of values, was calculated for anterior, posterior, left and right regions of each annulus fibrosus (AF). While mean T2 and degeneration schemes were not related to the IVDs' mechanical properties, TOTSU computed in the posterior AF correlated significantly with those classifications as well as with all stiffness values. TOTSU should therefore be included in future degeneration grading schemes.

Personalized identification of abdominal wall hernia meshes on computed tomography

An abdominal wall hernia is a protrusion of the intestine through an opening or area of weakness in the abdominal wall. Correct pre-operative identification of abdominal wall hernia meshes could help surgeons adjust the surgical plan to meet the expected difficulty and morbidity of operating through or removing the previous mesh. First, we present herein for the first time the application of image analysis for automated identification of hernia meshes. Second, we discuss the novel development of a new entropy-based image texture feature using geostatistics and indicator kriging. Third, we seek to enhance the hernia mesh identification by combining the new texture feature with the gray-level co-occurrence matrix feature of the image. The two features can characterize complementary information of anatomic details of the abdominal hernia wall and its mesh on computed tomography. Experimental results have demonstrated the effectiveness of the proposed study. The new computational tool has potential for personalized mesh identification which can assist surgeons in the diagnosis and repair of complex abdominal wall hernias.

Customized first and second order statistics based operators to support advanced texture analysis of MRI images

Texture analysis is the process of highlighting key characteristics thus providing an exhaustive and unambiguous mathematical description of any object represented in a digital image. Each characteristic is connected to a specific property of the object. In some cases the mentioned properties represent aspects visually perceptible which can be detected by developing operators based on Computer Vision techniques. In other cases these properties are not visually perceptible and their computation is obtained by developing operators based on Image Understanding approaches. Pixels composing high quality medical images can be considered the result of a stochastic process since they represent morphological or physiological processes. Empirical observations have shown that these images have visually perceptible and hidden significant aspects. For these reasons, the operators can be developed by means of a statistical approach. In this paper we present a set of customized first and second order statistics based operators to perform advanced texture analysis of Magnetic Resonance Imaging (MRI) images. In particular, we specify the main rules defining the role of an operator and its relationship with other operators. Extensive experiments carried out on a wide dataset of MRI images of different body regions demonstrating usefulness and accuracy of the proposed approach are also reported.

Multimodal image registration of the scoliotic torso for surgical planning

BACKGROUND

This paper presents a method that registers MRIs acquired in prone position, with surface topography (TP) and X-ray reconstructions acquired in standing position, in order to obtain a 3D representation of a human torso incorporating the external surface, bone structures, and soft tissues.

METHODS

TP and X-ray data are registered using landmarks. Bone structures are used to register each MRI slice using an articulated model, and the soft tissue is confined to the volume delimited by the trunk and bone surfaces using a constrained thin-plate spline.

RESULTS

The method is tested on 3 pre-surgical patients with scoliosis and shows a significant improvement, qualitatively and using the Dice similarity coefficient, in fitting the MRI into the standing patient model when compared to rigid and articulated model registration. The determinant of the Jacobian of the registration deformation shows higher variations in the deformation in areas closer to the surface of the torso.

CONCLUSIONS

The novel, resulting 3D full torso model can provide a more complete representation of patient geometry to be incorporated in surgical simulators under development that aim at predicting the effect of scoliosis surgery on the external appearance of the patient's torso.

MRI signal distribution within the intervertebral disc as a biomarker of adolescent idiopathic scoliosis and spondylolisthesis

Background

Early stages of scoliosis and spondylolisthesis entail changes in the intervertebral disc (IVD) structure and biochemistry. The current clinical use of MR T2-weighted images is limited to visual inspection. Our hypothesis is that the distribution of the MRI signal intensity within the IVD in T2-weighted images depends on the spinal pathology and on its severity. Therefore, this study aims to develop the AMRSID (analysis of MR signal intensity distribution) method to analyze the 3D distribution of the MR signal intensity within the IVD and to evaluate their sensitivity to scoliosis and spondylolisthesis and their severities.

Methods

This study was realized on 79 adolescents who underwent a MRI acquisition (sagittal T2-weighted images) before their orthopedic or surgical treatment. Five groups were considered: low severity scoliosis (Cobb angle ≤50°), high severity scoliosis (Cobb angles >50°), low severity spondylolisthesis (Meyerding grades I and II), high severity spondylolisthesis (Meyerding grades III, IV and V) and control. The distribution of the MRI signal intensity within the IVD was analyzed using the descriptive statistics of histograms normalized by either cerebrospinal fluid or bone signal intensity, weighted centers and volume ratios. Differences between pathology and severity groups were assessed using one- and two-way ANOVAs.

Results

There were significant (p < 0.05) variations of indices between scoliosis, spondylolithesis and control groups and between low and high severity groups. The cerebrospinal fluid normalization was able to detect differences between healthy and pathologic IVDs whereas the bone normalization, which reflects both bone and IVD health, detected more differences between the severities of these pathologies.

Conclusions

This study proves for the first time that changes in the intervertebral disc, non visible to the naked eye on sagittal T2-weighted MR images of the spine, can be detected from specific indices describing the distribution of the MR signal intensity. Moreover, these indices are able to discriminate between scoliosis and spondylolisthesis and their severities, and provide essential information on the composition and structure of the discs whatever the pathology considered. The AMRSID method may have the potential to complement the current diagnostic tools available in clinics to improve the diagnostic with earlier biomarkers, the prognosis of evolution and the treatment options of scoliosis and spondylolisthesis.

Automated detection, 3D segmentation and analysis of high resolution spine MR images using statistical shape models

Recent advances in high resolution magnetic resonance (MR) imaging of the spine provide a basis for the automated assessment of intervertebral disc (IVD) and vertebral body (VB) anatomy. High resolution three-dimensional (3D) morphological information contained in these images may be useful for early detection and monitoring of common spine disorders, such as disc degeneration. This work proposes an automated approach to extract the 3D segmentations of lumbar and thoracic IVDs and VBs from MR images using statistical shape analysis and registration of grey level intensity profiles. The algorithm was validated on a dataset of volumetric scans of the thoracolumbar spine of asymptomatic volunteers obtained on a 3T scanner using the relatively new 3D T2-weighted SPACE pulse sequence. Manual segmentations and expert radiological findings of early signs of disc degeneration were used in the validation. There was good agreement between manual and automated segmentation of the IVD and VB volumes with the mean Dice scores of 0.89 ± 0.04 and 0.91 ± 0.02 and mean absolute surface distances of 0.55 ± 0.18 mm and 0.67 ± 0.17 mm respectively. The method compares favourably to existing 3D MR segmentation techniques for VBs. This is the first time IVDs have been automatically segmented from 3D volumetric scans and shape parameters obtained were used in preliminary analyses to accurately classify (100% sensitivity, 98.3% specificity) disc abnormalities associated with early degenerative changes.

Quantitative evaluation of an automatic segmentation method for 3D reconstruction of intervertebral scoliotic disks from MR images

Background

For some scoliotic patients the spinal instrumentation is inevitable. Among these patients, those with stiff curvature will need thoracoscopic disk resection. The removal of the intervertebral disk with only thoracoscopic images is a tedious and challenging task for the surgeon. With computer aided surgery and 3D visualisation of the interverterbral disk during surgery, surgeons will have access to additional information such as the remaining disk tissue or the distance of surgical tools from critical anatomical structures like the aorta or spinal canal. We hypothesized that automatically extracting 3D information of the intervertebral disk from MR images would aid the surgeons to evaluate the remaining disk and would add a security factor to the patient during thoracoscopic disk resection.

Methods

This paper presents a quantitative evaluation of an automatic segmentation method for 3D reconstruction of intervertebral scoliotic disks from MR images. The automatic segmentation method is based on the watershed technique and morphological operators. The 3D Dice Similarity Coefficient (DSC) is the main statistical metric used to validate the automatically detected preoperative disk volumes. The automatic detections of intervertebral disks of real clinical MR images are compared to manual segmentation done by clinicians.

Results

Results show that depending on the type of MR acquisition sequence, the 3D DSC can be as high as 0.79 (±0.04). These 3D results are also supported by a 2D quantitative evaluation as well as by robustness and variability evaluations. The mean discrepancy (in 2D) between the manual and automatic segmentations for regions around the spinal canal is of 1.8 (±0.8) mm. The robustness study shows that among the five factors evaluated, only the type of MRI acquisition sequence can affect the segmentation results. Finally, the variability of the automatic segmentation method is lower than the variability associated with manual segmentation performed by different physicians.

Conclusions

This comprehensive evaluation of the automatic segmentation and 3D reconstruction of intervertebral disks shows that the proposed technique used with specific MRI acquisition protocol can detect intervertebral disk of scoliotic patient. The newly developed technique is promising for clinical context and can eventually help surgeons during thoracoscopic intervertebral disk resection.

Graph cuts with invariant object-interaction priors: application to intervertebral disc segmentation

This study investigates novel object-interaction priors for graph cut image segmentation with application to intervertebral disc delineation in magnetic resonance (MR) lumbar spine images. The algorithm optimizes an original cost function which constrains the solution with learned prior knowledge about the geometric interactions between different objects in the image. Based on a global measure of similarity between distributions, the proposed priors are intrinsically invariant with respect to translation and rotation. We further introduce a scale variable from which we derive an original fixed-point equation (FPE), thereby achieving scale-invariance with only few fast computations. The proposed priors relax the need of costly pose estimation (or registration) procedures and large training sets (we used a single subject for training), and can tolerate shape deformations, unlike template-based priors. Our formulation leads to an NP-hard problem which does not afford a form directly amenable to graph cut optimization. We proceeded to a relaxation of the problem via an auxiliary function, thereby obtaining a nearly real-time solution with few graph cuts. Quantitative evaluations over 60 intervertebral discs acquired from 10 subjects demonstrated that the proposed algorithm yields a high correlation with independent manual segmentations by an expert. We further demonstrate experimentally the invariance of the proposed geometric attributes. This supports the fact that a single subject is sufficient for training our algorithm, and confirms the relevance of the proposed priors to disc segmentation.

Heterogeneous computing for vertebra detection and segmentation in x-ray images

The context of this work is related to the vertebra segmentation. The method we propose is based on the active shape model (ASM). An original approach taking advantage of the edge polygonal approximation was developed to locate the vertebra positions in a X-ray image. Despite the fact that segmentation results show good efficiency, the time is a key variable that has always to be optimized in a medical context. Therefore, we present how vertebra extraction can efficiently be performed in exploiting the full computing power of parallel (GPU) and heterogeneous (multi-CPU/multi-GPU) architectures. We propose a parallel hybrid implementation of the most intensive steps enabling to boost performance. Experimentations have been conducted using a set of high-resolution X-ray medical images, showing a global speedup ranging from 3 to 22, by comparison with the CPU implementation. Data transfer times between CPU and GPU memories were included in the execution times of our proposed implementation.