Prediction of myelopathic level in cervical spondylotic myelopathy using diffusion tensor imaging.

PURPOSE: To investigate the use of a newly designed machine learning-based classifier in the automatic identification of myelopathic levels in cervical spondylotic myelopathy (CSM). MATERIALS AND METHODS: In all, 58 normal volunteers and 16 subjects with CSM were recruited for diffusion tensor imaging (DTI) acquisition. The eigenvalues were extracted as the selected features from DTI images. Three classifiers, naive Bayesian, support vector machine, and support tensor machine, and fractional anisotropy (FA) were employed to identify myelopathic levels. The results were compared with clinical level diagnosis results and accuracy, sensitivity, and specificity were calculated to evaluate the performance of the developed classifiers. RESULTS: The accuracy by support tensor machine was the highest (93.62%) among the three classifiers. The support tensor machine also showed excellent capacity to identify true positives (sensitivity: 84.62%) and true negatives (specificity: 97.06%). The accuracy by FA value was the lowest (76%) in all the methods. CONCLUSION: The classifiers-based method using eigenvalues had a better performance in identifying the levels of CSM than the diagnosis using FA values. The support tensor machine was the best among three classifiers.

Quantitative assessment of column-specific degeneration in cervical spondylotic myelopathy based on diffusion tensor tractography.

PURPOSE: Cervical spondylotic myelopathy (CSM) is a common spinal cord disorder in the elderly. Diffusion tensor imaging (DTI) has been shown to be of great value for evaluating the microstructure of nerve tracts in the spinal cord. Currently, the quantitative assessment of the degeneration on the specific tracts in CSM is still rare. The aim of the present study was to use tractography-based quantification to investigate the column-specific degeneration in CSM. METHODS: A total of 43 volunteers were recruited with written informed consent, including 20 healthy subjects and 23 CSM patients. Diffusion MRI was taken by 3T MRI scanner. Fiber tractography was performed using TrackVis to reconstruct the white matter tracts of the anterior, lateral and posterior column on the bilateral sides. The DTI metrics acquired from tractography, including fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD) and radial diffusivity (RD), were compared between healthy subjects and CSM patients. RESULTS: Compared to healthy subjects, FA was found significantly lower in the lateral (Healthy 0.64 ± 0.07 vs. CSM 0.53 ± 0.08) and posterior column (Healthy 0.67 ± 0.08 vs. CSM 0.47 ± 0.08) (p < 0.001), while MD, AD and RD were significantly higher in the anterior, lateral and posterior column in CSM (p < 0.05). CONCLUSION: Loss of microstructural integrity was detected in the lateral and posterior column in CSM. Tractography-based quantification was capable of evaluating the subtle pathological insult within white matter on a column-specific basis, which exhibited potential clinical value for in vivo evaluation of the severity of CSM.

Potential use of diffusion tensor imaging in level diagnosis of multilevel cervical spondylotic myelopathy.

STUDY DESIGN: A prospective study on a series of consecutive patients. OBJECTIVE: To investigate the use of diffusion tensor imaging (DTI) and orientation entropy in level localization in patients diagnosed with multilevel cervical spondylotic myelopathy (CSM). SUMMARY OF BACKGROUND DATA: Multilevel CSM presents complex neurological signs that make level localization difficult. DTI is recently found to be able to assess the microstructural changes of the white matter caused by cord compression. METHODS: Sixteen patients with CSM with multilevel compression were recruited. The level(s) responsible for the clinical symptoms were determined by detailed neurological examination, T2-weighted (T2W) magnetic resonance imaging (MRI), and DTI. On T2W MRI, anterior-posterior compression ratio and increased signal intensities were used to determine the affected level(s). The level diagnosis results from T2W MRI, increased signal intensities, DTI, and combination method were correlated to that of neurological examination on a level-to-level basis, respectively. The accuracy, sensitivity, and specificity were calculated. RESULTS: When correlated with the clinical level determination, the weighted orientation entropy-based DTI analysis was found to have higher accuracy (82.76% vs. 75.86%) and sensitivity (84.62% vs. 76.92%) than those of the anterior-posterior compression ratio. The increased signal intensities have the highest specificity (100.00%) but the lowest accuracy (58.62%) and sensitivity (53.85%). When combined with the level diagnosis result of wOE with that of anterior-posterior compression ratio, it demonstrated the highest accuracy and sensitivity that were 93.10% and 96.15%, respectively, and equal specificity (66.67%) with using them individually. CONCLUSION: DTI can be a useful tool to determine the pathological spinal cord levels in multilevel CSM. This information from orientation entropy-based DTI analysis, in addition to conventional MRI and clinical neurological assessment, should help spine surgeons in deciding the optimal surgical strategy.

Is diffusion anisotropy a biomarker for disease severity and surgical prognosis of cervical spondylotic myelopathy?

PURPOSE: To explore the value of diffusion-tensor (DT) imaging in addressing the severity of cervical spondylotic myelopathy (CSM) and predicting the outcome of surgical treatment. MATERIALS AND METHODS: From July 2009 to May 2012, 65 volunteers were recruited for this institutional review board-approved study, and all gave informed consent; 20 volunteers were healthy subjects (age range, 41-62 years), and 45 were patients with CSM (age range, 43-86 years). Anatomic and DT 3.0-T magnetic resonance images were obtained. Surgical decompression was performed in 22 patients with CSM, and patients were followed up for 6 months to 2 years. The clinical severity of myelopathy and postoperative recovery were assessed by using the modified Japanese Orthopaedic Association (mJOA) score. A recovery ratio (comparison of postoperative with preoperative mJOA score) of more than 50% indicated a good clinical outcome of surgery. DT findings, patient age, T2 high signal intensity (HSI), and somatosensory evoked potential (SEP) were analyzed by using a logistic regression model to predict the surgical outcome of patients with CSM. RESULTS: A significant difference in cervical cord mean fractional anisotropy (FA) was found between healthy subjects and patients with CSM (0.65 ± 0.05 [standard deviation] vs 0.52 ± 0.13, P < .001). FA values were significantly correlated with the severity of neurologic dysfunction indicated by mJOA score (r(2) = 0.327, P = .016). Logistic regression analysis showed that mean FA (P = .030) and FA at the C2 vertebra (P = .035) enabled prediction of good surgical outcome; however, preoperative mJOA (P = .927), T2 HSI (P = .176), SEP amplitude (P = .154), and latency (P = .260) did not. CONCLUSION: FA is a biomarker for the severity of myelopathy and for subsequent surgical outcome.

Diffusion tensor imaging of somatosensory tract in cervical spondylotic myelopathy and its link with electrophysiological evaluation.

BACKGROUND AND CONTEXT: Abnormal somatosensory evoked potential (SEP) (ie, prolonged latency) has been associated with poor surgical prognosis of cervical spondylotic myelopathy (CSM). PURPOSE: To further characterize the extent of microstructural damage to the somatosensory tract in CSM patients using diffusion tensor imaging (DTI). STUDY DESIGN/SETTING: Retrospective study. PATIENT SAMPLE: A total of 40 volunteers (25 healthy subjects and 15 CSM patients). OUTCOME MEASURES: Clinical, electrophysiological, and radiological evaluations were performed using the modified Japanese Orthopedic Association (mJOA) scoring system, SEP, and cord compression ratio in anatomic magnetic resonance (MR) images, respectively. Axial diffusion MR images were taken using a pulsed gradient, spin-echo-echo-planar imaging sequence with a 3-T MR system. The diffusion indices in different regions of the spinal cord were measured. METHODS: Comparison of diffusion indices among healthy and myelopathic spinal cord with intact and impaired SEP responses were performed using one-way analysis of variance. RESULTS: In healthy subjects, fractional anisotropy (FA) values were higher in the dorsal (0.73±0.11) and lateral columns (0.72±0.13) than in the ventral column of white matter (0.58±0.10) (eg, at C4/5) (p<.05). FA was dramatically dropped in the dorsal (0.54±0.16) and lateral columns (0.51±0.13) with little change in the ventral column (0.48±0.15) at the compressive lesions in CSM patients. There were no significant differences in the mJOA scores or cord compression ratios between CSM patients with or without abnormal SEP. However, patients with abnormal SEP showed an FA decrease in the dorsal column cephalic to the lesion (0.56±0.06) (ie, at C1/2, compared with healthy subjects [0.66±0.02]), but the same decrease was not observed for those without a SEP abnormality (0.67±0.02). CONCLUSION: Spinal tracts were not uniformly affected in the myelopathic cervical cord. Changes in diffusion indices could delineate focal or extensive myelopathic lesions in CSM, which could account for abnormal SEP. DTI analysis of spinal tracts might provide additional information not available from conventional diagnostic tools for prognosis of CSM.

Quantitative analysis of fiber tractography in cervical spondylotic myelopathy.

BACKGROUND CONTEXT: Diffusion tensor fiber tractography is an emerging tool for the visualization of spinal cord microstructure. However, there are few quantitative analyses of the damage in the nerve fiber tracts of the myelopathic spinal cord. PURPOSE: The aim of this study was to develop a quantitative approach for fiber tractography analysis in cervical spondylotic myelopathy (CSM). STUDY DESIGN/SETTING: Prospective study on a series of patients. MATERIALS AND METHODS: A total of 22 volunteers were recruited with informed consent, including 15 healthy subjects and 7 CSM patients. The clinical severity of CSM was evaluated using modified Japanese Orthopedic Association (JOA) score. The microstructure of myelopathic cervical cord was analyzed using diffusion tensor imaging. Diffusion tensor imaging was performed with a 3.0-T magnetic resonance imaging scanner using pulsed gradient, spin-echo, echo-planar imaging sequence. Fiber tractography was generated via TrackVis with fractional anisotropy threshold set at 0.2 and angle threshold at 40°. Region of interest (ROI) was defined to cover C4 level only or the whole-length cervical spinal cord from C1 to C7 for analysis. The length and density of tracked nerve bundles were measured for comparison between healthy subjects and CSM patients. RESULTS: The length of tracked nerve bundles significantly shortened in CSM patients compared with healthy subjects (healthy: 6.85-77.90 mm, CSM: 0.68-62.53 mm). The density of the tracked nerve bundles was also lower in CSM patients (healthy: 086±0.03, CSM: 0.80±0.06, p<.05). Although the definition of ROI covering C4 only or whole cervical cord appeared not to affect the trend of the disparity between healthy and myelopathic cervical cords, the density of the tracked nerve bundle through whole myelopathic cords was in an association with the modified JOA score in CSM cases (r=0.949, p=.015), yet not found with ROI at C4 only (r=0.316, p=.684). CONCLUSIONS: The quantitative analysis of fiber tractography is a reliable approach to detect cervical spondylotic myelopathic lesions compared with healthy spinal cords. It could be employed to delineate the severity of CSM.

Orientation entropy analysis of diffusion tensor in healthy and myelopathic spinal cord.

The majority of nerve fibers in the spinal cord run longitudinally, playing an important role in connecting the brain to the peripheral nerves. There is a growing interest in applying diffusion tensor imaging (DTI) to the evaluation of spinal cord microarchitecture. The current study sought to compare the organization of longitudinal nerve fibers between healthy and myelopathic spinal cords using entropy-based analysis of principal eigenvector mapping. A total of 22 subjects were recruited, including 14 healthy subjects, seven cervical myelopathy (CM) patients with single-level compression, and one patient suffering from multi-level compression. Diffusion tensor magnetic resonance (MR) images of the cervical spinal cord were obtained using a pulsed gradient, spin-echo echo-planar imaging (SE-EPI) sequence with a 3T MR system. Regions of interest (ROIs) were drawn manually to cover the spinal cord, and Shannon entropy was calculated in principal eigenvector maps. The results revealed no significant differences in orientation entropy values along the whole length of cervical spinal cord in healthy subjects (C2-3: 0.73±0.05; C3-4: 0.71±0.07; C4-5: 0.72±0.048; C5-6: 0.71±0.07; C6-7: 0.72±0.07). In contrast, orientation entropy values in myelopathic cord were significantly higher at the compression site (0.91±0.03), and the adjacent levels (above: 0.85±0.03; below: 0.83±0.05). This study provides a novel approach to analyze the orientation information in diffusion MR images of healthy and diseased spinal cord. These results indicate that orientation entropy can be applied to determine the contribution of each compression level to the overall disorganization of principal nerve tracts of myelopathic spinal cord in cases with multi-level compression.

Entropy-based analysis for diffusion anisotropy mapping of healthy and myelopathic spinal cord.

The present study utilized diffusion MR imaging and fractional anisotropy (FA) mapping to delineate the microstructure of spinal cord. The concept of Shannon entropy was introduced to analyze the complex microstructure of healthy and injured spinal cords based on FA map. A total of 30 volunteers were recruited in this study with informed consent, including 13 healthy adult subjects (group A, 25±3 years), 12 healthy elderly subjects (group B, 53±7 years) and 5 cervical spondylotic myelopathy (CSM) patients (group C, 53±15 years). Diffusion MRI images of cervical spinal cord were taken using pulsed gradient spin-echo-echo-planar imaging (SE-EPI) sequence with a 3T MR system. The region of interest was defined to cover the spinal cord in FA maps. The Shannon entropy of FA values of voxels in the cord was calculated as well as the average FA values. The significant differences were determined among three groups using one-way ANOVA and post-hoc test. As compared with adult and elderly healthy subjects, the entropy of whole spinal cord was significantly lower in CSM patients (group A: 6.07±0.18; B: 6.01±0.23; C: 5.32±0.44; p<0.05). Whereas there were no significant difference in FA values among groups (group A: 0.62±0.08; B: 0.64±0.09; C: 0.64±0.12). In CSM patients, there was a loss of architectural structural complexity in the cervical spinal cord tissue as noted by the lower Shannon entropy value. It indicated the potential application of entropy-based analysis for the diagnosis of the severity of chronic compressive spinal cord injuries, i.e. CSM.