Multivariate Analysis of MRI Biomarkers for Predicting Neurologic Impairment in Cervical Spinal Cord Injury

Haefeli, Jenny; Mabray, Marc C.; Whetstone, William D.; Dhall, Sanjay S.; Pan, Jonathan Z.; Upadhyayula, Pavan S.; Manley, Geoffrey T.; Bresnahan, Jacqueline C.; Beattie, Michael S.; Ferguson, Adam R.; Talbott, Jason F.

doi:10.3174/ajnr.a5021

Cited by 57 publications

(37 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…SC segmentation for acute SCI is particularly challenging given the high frequency of SC distortion related to compression and associated geometric distortion as well as heterogeneous intramedullary signal abnormality. 2,29,30 Our targeted, disease-specific approach to network training likely, in part, explains performance differences between BASICseg and Deepseg algorithms for our SCI cohort. All CNN-based algorithms (BASICseg1-3 and Deepseg) outperformed Propseg; this difference highlights the value of CNN applications for SC segmentation.…”

Section: Discussionmentioning

confidence: 99%

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

McCoy

Gros

Cohen‐Adad

et al. 2019

AJNR Am J Neuroradiol

Self Cite

View full text Add to dashboard Cite

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 Ͻ .001) and DeepSeg (P Ͻ .05). Brain and Spinal Cord Injury Center segmentation showed significantly better adaptability to damaged areas compared with PropSeg (P Ͻ .001) and DeepSeg (P Ͻ .02). The contusion injury volumes based on automated segmentation were significantly associated with motor scores at admission (P ϭ .002) and discharge (P ϭ .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. ABBREVIATIONS: BASICseg ϭ Brain and Spinal Cord Injury Center segmentation; CNN ϭ convolutional neural network; SC ϭ spinal cord; SCI ϭ spinal cord injury; SCT ϭ Spinal Cord Toolbox T he natural history of recovery after spinal cord injury (SCI) is highly variable and often difficult to predict. 1,2 Early objective neurologic assessment of injury severity in these patients is challenging due to several confounding factors, such as traumatic brain injury, severe pain, intubation, spinal shock, and sedative medications, among others. 2,3 Early and accurate classification of patients with SCI as close to the time of injury as possible is imperative for the guidance of triage, acute management, prognostication, and for patient selection in the design of clinical trials in which novel therapies are tested. 2,3 MR imaging is the criterion standard imaging technique for evaluation of the ...

show abstract

Section: Discussionmentioning

confidence: 99%

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

McCoy

Gros

Cohen‐Adad

et al. 2019

AJNR Am J Neuroradiol

Self Cite

View full text Add to dashboard Cite

show abstract

“…37 The 2 PCs are PC1, which is represented by all imaging variables, with measures of intrinsic cord compression (BA-SIC score, linear length of injury, sagittal grade) showing highest loadings; and PC2, which is represented by markers of extrinsic cord compression. The PC1 predicted AIS at discharge, and among its components, the BASIC score consistently demonstrated the strongest distinction between severe, moderate, and mild AIS grading.…”

Section: 49mentioning

confidence: 99%

Update on critical care for acute spinal cord injury in the setting of polytrauma

Yue

Winkler

Rick

et al. 2017

Neurosurgical Focus

View full text Add to dashboard Cite

Traumatic spinal cord injury (SCI) often occurs in patients with concurrent traumatic injuries in other body systems. These patients with polytrauma pose unique challenges to clinicians. The current review evaluates existing guidelines and updates the evidence for prehospital transport, immobilization, initial resuscitation, critical care, hemodynamic stability, diagnostic imaging, surgical techniques, and timing appropriate for the patient with SCI who has multisystem trauma. Initial management should be systematic, with focus on spinal immobilization, timely transport, and optimizing perfusion to the spinal cord. There is general evidence for the maintenance of mean arterial pressure of > 85 mm Hg during immediate and acute care to optimize neurological outcome; however, the selection of vasopressor type and duration should be judicious, with considerations for level of injury and risks of increased cardiogenic complications in the elderly. Level II recommendations exist for early decompression, and additional time points of neurological assessment within the first 24 hours and during acute care are warranted to determine the temporality of benefits attributable to early surgery. Venous thromboembolism prophylaxis using low-molecular-weight heparin is recommended by current guidelines for SCI. For these patients, titration of tidal volumes is important to balance the association of earlier weaning off the ventilator, with its risk of atelectasis, against the risk for lung damage from mechanical overinflation that can occur with prolonged ventilation. Careful evaluation of infection risk is a priority following multisystem trauma for patients with relative immunosuppression or compromise. Although patients with polytrauma may experience longer rehabilitation courses, long-term neurological recovery is generally comparable to that in patients with isolated SCI after controlling for demographics. Bowel and bladder disorders are common following SCI, significantly reduce quality of life, and constitute a focus of targeted therapies. Emerging biomarkers including glial fibrillary acidic protein, S100β, and microRNAs for traumatic SCIs are presented. Systematic management approaches to minimize sources of secondary injury are discussed, and areas requiring further research, implementation, and validation are identified.

show abstract

“…23,24 The longitudinal extent of T2 hyperintensity (in millimeters) was evaluated based on the National Institutes of Health/National Institutes of Neurological Disorders and Stroke SCI common data elements version 1.0. 22,25,26 Sagittal grading was evaluated based on previous studies 22,25 : grade 1, no spinal cord abnormal intensity; grade 2, one-level T2 hyperintensity; grade 3, more than a two-level T2 signal hyperintensity; and grade 4, T2 signal hyperintensity with lesions of hypointensity indicating hemorrhage.…”

Section: Discussionmentioning

confidence: 99%

“…27 Maximum canal compromise (MCC) and maximum spinal cord compression (MSCC) were evaluated on mid-sagittal images by differentiating the anteroposterior diameter of the canal (on sagittal T1WI for MCC) and that of the spinal cord (on sagittal T2WI for MSCC) by means of the canal or spinal cord above and below as reported previously. 25,28 The signal intensity ratio (SIR) at the narrowest level of the spinal cord on sagittal views of T1WI and T2WI was measured, Radiographs were also obtained using normal radiographic methods in which the tube was positioned on the C5 disc. The radiographic film cassette was 150 cm from the tube.…”

Section: Discussionmentioning

confidence: 99%

XGBoost, a Machine Learning Method, Predicts Neurological Recovery in Patients with Cervical Spinal Cord Injury

Inoue

Ichikawa

Ueno

et al. 2020

Neurotrauma Reports

View full text Add to dashboard Cite

The accurate prediction of neurological outcomes in patients with cervical spinal cord injury (SCI) is difficult because of heterogeneity in patient characteristics, treatment strategies, and radiographic findings. Although machine learning algorithms may increase the accuracy of outcome predictions in various fields, limited information is available on their efficacy in the management of SCI. We analyzed data from 165 patients with cervical SCI, and extracted important factors for predicting prognoses. Extreme gradient boosting (XGBoost) as a machine learning model was applied to assess the reliability of a machine learning algorithm to predict neurological outcomes compared with that of conventional methodology, such as a logistic regression or decision tree. We used regularly obtainable data as predictors, such as demographics, magnetic resonance variables, and treatment strategies. Predictive tools, including XGBoost, a logistic regression, and a decision tree, were applied to predict neurological improvements in the functional motor status (ASIA [American Spinal Injury Association] Impairment Scale [AIS] D and E) 6 months after injury. We evaluated predictive performance, including accuracy and the area under the receiver operating characteristic curve (AUC). Regarding predictions of neurological improvements in patients with cervical SCI, XGBoost had the highest accuracy (81.1%), followed by the logistic regression (80.6%) and the decision tree (78.8%). Regarding AUC, the logistic regression showed 0.877, followed by XGBoost (0.867) and the decision tree (0.753). XGBoost reliably predicted neurological alterations in patients with cervical SCI. The utilization of predictive machine learning algorithms may enhance personalized management choices through pre-treatment categorization of patients.

show abstract

Multivariate Analysis of MRI Biomarkers for Predicting Neurologic Impairment in Cervical Spinal Cord Injury

Cited by 57 publications

References 38 publications

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

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

Update on critical care for acute spinal cord injury in the setting of polytrauma

XGBoost, a Machine Learning Method, Predicts Neurological Recovery in Patients with Cervical Spinal Cord Injury

Contact Info

Product

Resources

About