Multi-muscle deep learning segmentation to automate the quantification of muscle fat infiltration in cervical spine conditions.

Muscle fat infiltration (MFI) has been widely reported across cervical spine disorders. The quantification of MFI requires time-consuming and rater-dependent manual segmentation techniques. A convolutional neural network (CNN) model was trained to segment seven cervical spine muscle groups (left and right muscles segmented separately, 14 muscles total) from Dixon MRI scans (n = 17, 17 scans < 2 weeks post motor vehicle collision (MVC), and 17 scans 12 months post MVC). The CNN MFI measures demonstrated high test reliability and accuracy in an independent testing dataset (n = 18, 9 scans < 2 weeks post MVC, and 9 scans 12 months post MVC). Using the CNN in 84 participants with scans < 2 weeks post MVC (61 females, 23 males, age = 34.2 ± 10.7 years) differences in MFI between the muscle groups and relationships between MFI and sex, age, and body mass index (BMI) were explored. Averaging across all muscles, females had significantly higher MFI than males (p = 0.026). The deep cervical muscles demonstrated significantly greater MFI than the more superficial muscles (p < 0.001), and only MFI within the deep cervical muscles was moderately correlated to age (r > 0.300, p ≤ 0.001). CNN's allow for the accurate and rapid, quantitative assessment of the composition of the architecturally complex muscles traversing the cervical spine. Acknowledging the wider reports of MFI in cervical spine disorders and the time required to manually segment the individual muscles, this CNN may have diagnostic, prognostic, and predictive value in disorders of the cervical spine.

Macromolecular changes in spinal cord white matter characterize whiplash outcome at 1-year post motor vehicle collision.

Each year, whiplash injuries from motor vehicle collisions (MVC) affect millions worldwide, with no strong evidence of pathology. While the majority recover soon after the injury, the same is not true for roughly 20% reporting higher levels of pain and distress, without diagnostic options. This study used magnetization transfer (MT) imaging to quantify white matter integrity in 78 subjects with varying levels of pain, 1 year after MVC. MT images of the cervical spinal cord were collected parallel to the intervertebral disks. MT ratios (MTR) were calculated in select white matter tracts along with MTR homogeneity (MTRh) at each level. Significant differences were observed between clinical outcome groups in the left and right spinothalamic tracts (p = 0.003 and 0.020) and MTRh (p = 0.009). MTRh was elevated in females with poor recovery versus females reporting recovery (p < 0.001) or milder symptoms (p < 0.001), and in males reporting recovery (p = 0.007) or no recovery (p < 0.001). There was a significant interaction between recovery status and sex (p = 0.015). MT imaging identified tract specific and regional changes in white matter integrity suggesting potential insults to the cord. Additionally, significant MTRh differences between sexes were observed, characterizing the heterogeneity of whiplash recovery and worse outcomes in females.

Confirming the geography of fatty infiltration in the deep cervical extensor muscles in whiplash recovery.

Previous preliminary work mapped the distribution of neck muscle fat infiltration (MFI) in the deep cervical extensor muscles (multifidus and semispinalis cervicis) in a small cohort of participants with chronic whiplash associated disorders (WAD), recovered, and healthy controls. While MFI was reported to be concentrated in the medial portion of the muscles in all participants, the magnitude was significantly greater in those with chronic WAD. This study aims to confirm these results in a prospective fashion with a larger cohort and compare the findings across a population of patients with varying levels of WAD-related disability one-year following the motor vehicle collision. Sixty-one participants enrolled in a longitudinal study: Recovered (n = 25), Mild (n = 26) and Severe WAD (n = 10) were studied using Fat/Water magnetic resonance imaging, 12-months post injury. Bilateral measures of MFI in four quartiles (Q1-Q4; medial to lateral) at cervical levels C4 through C7 were included. A linear mixed model was performed, controlling for covariates (age, sex, body mass index), examining interaction effects, and comparing MFI distribution between groups. The recovered group had significantly less MFI in Q1 compared to the two symptomatic groups. Group differences were not found in the more lateral quartiles. Results at 12 months are consistent with the preliminary study, indicating that MFI is spatially concentrated in the medial portions of the deep cervical extensors regardless of WAD recovery, but the magnitude of MFI in the medial portions of the muscles is significantly larger in those with severe chronic WAD.

Muscle fat infiltration following whiplash: A computed tomography and magnetic resonance imaging comparison.

Here we present a secondary analysis from a parent database of 97 acutely injured participants enrolled in a prospective inception cohort study of whiplash recovery after motor vehicle collision (MVC). The purpose was to investigate the deep and superficial neck extensor muscles with peri-traumatic computed tomography (CT) and longitudinal measures of magnetic resonance imaging (MRI) in participants with varying levels of whiplash-related disability. Thirty-six underwent standard care imaging of the cervical spine with CT at a level-1 trauma designated emergency department. All 36 participants were assessed with MRI of the cervical spine at <1-week, 2-weeks, 3-, and 12-months post-injury and classified into three groups using initial pain severity and percentage scores on the Neck Disability Index (recovered (NDI of 0-8%), mild (NDI of 10-28%), or severe (NDI ≥ 30%)) at 3-months post MVC. CT muscle attenuation values were significantly correlated to muscle fat infiltration (MFI) on MRI at one-week post MVC. There was no significant difference in muscle attenuation across groups at the time of enrollment. A trend of lower muscle attenuation in the deep compared to the superficial extensors was observed in the severe group. MFI values in the deep muscles on MRI were significantly higher in the severe group when compared to the mild group at 1-year post MVC. This study provides further evidence that the magnitude of 1) deep MFI appears unique to those at risk of and eventually transitioning to chronic WAD and that 2) pre- or peri-traumatic muscular health, determined by CT muscle attenuation, may be contribute to our understanding of long-term recovery.

Does Overall Cervical Spine Pathology Relate to the Clinical Heterogeneity of Chronic Whiplash?

BACKGROUND AND PURPOSE: There remains limited evidence for the clinical importance of most imaging findings in whiplash. However, it is possible the type and number of findings on Computed Tomography (CT) may contribute to prognostic recovery models. The purpose is to interpret cervical spine pathologies in the context of known factors influencing recovery. MATERIALS AND METHODS: This is a secondary analysis from a database of 97 acutely injured participants enrolled in a prospective inception cohort study. Thirty-eight participants underwent standard of care cervical spine CT in the emergency medicine department. All 38 participants were assessed at <1-week, 2-weeks, and 3-months post-injury and classified using percentage scores on the Neck Disability Index (recovered/mild (NDI of 0-28%) or moderate/severe (NDI ≥ 30%)). Between-group comparison of categorical variables (gender (male/female), presence of at least one CT finding (yes/no), and presence of ≥3 pathologies on CT (yes/no)) was conducted using 2-tailed Fisher's exact test. RESULTS: Participants from both groups demonstrated at least one observable pathology. The group with persistent moderate/severe symptoms presented with significantly more pathology at baseline than those who later reported recovery or milder symptoms at 3-months post injury (p = 0.02). CONCLUSIONS: This preliminary study, which needs replication in a larger cohort, provides foundation that the number of degenerative pathologies seen on initial post MVC CT may be associated with the subsequent clinical course of whiplash.

Motor vehicle crash reconstruction: Does it relate to the heterogeneity of whiplash recovery?

Whiplash injury is a common consequence of motor vehicle crashes (MVC), yet it is also one of the most poorly understood. While more than 50% of those injured should expect to rapidly recover, others are not as fortunate with approximately 25% of those exposed to and injured in an MVC transitioning from acute to chronic pain and disability. The purpose of this prospective study was to determine if the severity and direction of collisions involving participants enrolled in a longitudinal study of recovery from whiplash are able to differentiate between different recovery groups based on the neck disability index (NDI) percentage scores at 3-months, and if these crash specific parameters are associated with known risk factors for recovery. Here, we examined objective collision data, repair invoices, and characteristics of the crash for 37 acutely injured participants consented and enrolled at their emergency department visit and further assessed at three time points; < 1 week, 2-weeks, and 3-months post MVC. Collision data were used to reconstruct and estimate the severity of the crash and determine if they aligned with the heterogeneity of whiplash injury recovery. Wilcoxon rank sum tests were used to determine if % scores on the Neck Disability Index (NDI) at 3-months post MVC were associated with the following variables: sex, head turned at time of impact, seatbelt use, whether or not airbags deployed, if the vehicle was struck while stopped or while turning, or the principle direction of force (PDOF). Spearman's correlation coefficients were used to determine if NDI at 3-months post MVC was associated with age, Body Mass Index, pain-related disability at baseline, signs of post-traumatic distress, intrusion/hyperarousal, negative affect, pain intensity, estimated speed change from the impact, and damage estimates (in US$). There was a significant positive association between self-reported neck disability at 3-months post MVC, post-traumatic distress, negative affect and uncontrolled pain. There was no direct effect of participant characteristics, arousal, intrusion/hyperarousal sub-score, damage, PDOF, speed change, or other crash characteristics. Established crash parameters were not associated with the heterogeneity of whiplash injury recovery in a small sample of injured participants.

Deep Learning Convolutional Neural Networks for the Automatic Quantification of Muscle Fat Infiltration Following Whiplash Injury.

Muscle fat infiltration (MFI) of the deep cervical spine extensors has been observed in cervical spine conditions using time-consuming and rater-dependent manual techniques. Deep learning convolutional neural network (CNN) models have demonstrated state-of-the-art performance in segmentation tasks. Here, we train and test a CNN for muscle segmentation and automatic MFI calculation using high-resolution fat-water images from 39 participants (26 female, average = 31.7 ± 9.3 years) 3 months post whiplash injury. First, we demonstrate high test reliability and accuracy of the CNN compared to manual segmentation. Then we explore the relationships between CNN muscle volume, CNN MFI, and clinical measures of pain and neck-related disability. Across all participants, we demonstrate that CNN muscle volume was negatively correlated to pain (R = -0.415, p = 0.006) and disability (R = -0.286, p = 0.045), while CNN MFI tended to be positively correlated to disability (R = 0.214, p = 0.105). Additionally, CNN MFI was higher in participants with persisting pain and disability (p = 0.049). Overall, CNN's may improve the efficiency and objectivity of muscle measures allowing for the quantitative monitoring of muscle properties in disorders of and beyond the cervical spine.

Short- and long-term reproducibility of diffusion-weighted magnetic resonance imaging of lower extremity musculature in asymptomatic individuals and a comparison to individuals with spinal cord injury.

BACKGROUND: Diffusion-weighted magnetic resonance imaging (DW-MRI) of skeletal muscle has the potential to be a sensitive diagnostic and/or prognostic tool in complex, enigmatic neuromusculoskeletal conditions such as spinal cord injury and whiplash associated disorder. However, the reliability and reproducibility of clinically accessible DW-MRI parameters in skeletal muscle remains incompletely characterized - even in individuals without neuromusculoskeletal injury - and these parameters have yet to be characterized for many clinical populations. Here, we provide normative measures of the apparent diffusion coefficient (ADC) in healthy muscles of the lower limb; assess the rater-based reliability and short- and long-term reproducibility of the ADC in the same muscles; and quantify ADC of these muscles in individuals with motor incomplete spinal cord injury. METHODS: Twenty individuals without neuromusculoskeletal injury and 14 individuals with motor incomplete spinal cord injury (SCI) participated in this investigation. We acquired bilateral diffusion-weighted MRI of the lower limb musculature in all participants at 3 T using a multi-shot echo-planar imaging sequence with b-values of 0, 100, 300 and 500 s/mm2 and diffusion-probing gradients applied in 3 orthogonal directions. Outcome measures included: (1) average ADC in the lateral and medial gastrocnemius, tibialis anterior, and soleus of individuals without neurological or musculoskeletal injury; (2) intra- and inter-rater reliability, as well as short and long-term reproducibility of the ADC; and (3) estimation of average muscle ADC in individuals with SCI. RESULTS: Intra- and inter-rater reliability of the ADC averaged 0.89 and 0.79, respectively, across muscles. Least significant change, a measure of temporal reproducibility, was 4.50 and 11.98% for short (same day) and long (9-month) inter-scan intervals, respectively. Average ADC was significantly elevated across muscles in individuals with SCI compared to individuals without neurological or musculoskeletal injury (1.655 vs. 1.615 mm2/s, respectively). CONCLUSIONS: These findings provide a foundation for future studies that track longitudinal changes in skeletal muscle ADC of the lower extremity and/or investigate the mechanisms underlying ADC changes in cases of known or suspected pathology.

Lateral Corticospinal Tract Damage Correlates With Motor Output in Incomplete Spinal Cord Injury.

OBJECTIVE: To investigate the relationship between spinal cord damage and specific motor function in participants with incomplete spinal cord injury (iSCI). DESIGN: Single-blinded, cross-sectional study design. SETTING: University setting research laboratory. PARTICIPANTS: Individuals with chronic cervical iSCI (N=14; 1 woman, 13 men; average age ± SD, 43±12y). INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Axial T2-weighted magnetic resonance imaging (MRI) of spinal cord damage was performed in 14 participants with iSCI. Each participant's damage was processed for total damage quantification, lateral corticospinal tract (LCST) and gracile fasciculus (GF) analysis. Plantarflexion and knee extension were quantified using an isokinetic dynamometer. Walking ability was assessed using a 6-minute walk test. RESULTS: Total damage was correlated with plantarflexion, knee extension, and distance walked in 6 minutes. Right LCST damage was correlated with right plantarflexion and right knee extension, while left LCST damage was correlated with left-sided measures. Right and left GF damage was not correlated with the motor output measures. CONCLUSIONS: MRI measures of spinal cord damage were correlated to motor function, and this measure appears to have spatial specificity to descending tracts, which may offer prognostic value after SCI.

Lower extremity muscle structure in incomplete spinal cord injury: a comparison between ultrasonography and magnetic resonance imaging.

INTRODUCTION: Ultrasound (US) measures of muscle thickness are used in the management and monitoring of muscle changes during rehabilitation, but it is unknown how this measure compares to magnetic resonance imaging (MRI) measures of muscle cross-sectional area (CSA) in patients with spinal cord injury (SCI). CASE PRESENTATION: Six participants with incomplete SCI underwent US imaging and MRI of their gastrocnemius and tibialis anterior (TA) muscles. DISCUSSION: Significant correlations were found between US muscle thickness and MRI CSA in the gastrocnemius (R=0.91, P<0.001), and TA muscles (R=0.58, P<0.05). US muscle thickness in the gastrocnemius and TA of patients with incomplete SCI may be used as a cheaper alternative measure of CSA as measured using MRI, and this measure may be clinically useful to track progress in muscle gains during rehabilitation.

MRI measures of fat infiltration in the lower extremities following motor incomplete spinal cord injury: reliability and potential implications for muscle activation.

Muscle fat infiltration (MFI) is an expected consequence of incomplete spinal cord injury (iSCI). The MFI magnitude may have clinical value in determining functional recovery. However, there is a lack of understanding of how MFI relates to the volitional muscle activity in people with motor incomplete spinal cord injury (iSCI). Five iSCI and 5 uninjured age-matched control subjects participated in the study. In this preliminary study, we established the reliability of MFI quantification of select lower extremity muscles across different raters. Secondly, we assessed the magnitude and distribution of MFI in the lower legs of iSCI and uninjured control participants. Thirdly, we explored the relationship between MFI in the plantar flexor muscles and the ability to volitionally activate these muscles. High levels of inter-rater reliability were observed. The iSCI group had significantly elevated and a vastly different MFI distribution in the lower leg muscles compared to healthy controls. MFI was negatively correlated with volitional activation in iSCI. Our preliminary results sanction the importance of lower extremity MFI quantification as a potential measure in determining the functional outcomes in iSCI, and the subsequent pathological sequelae.

Potential associations between chronic whiplash and incomplete spinal cord injury.

STUDY DESIGN: This research utilized a cross-sectional design with control group inclusion. OBJECTIVES: Preliminary evidence suggests that a portion of the patient population with chronic whiplash may have sustained spinal cord damage. Our hypothesis is that in some cases of chronic whiplash-associated disorders (WAD), observed muscle weakness in the legs will be associated with local signs of a partial spinal cord injury of the cervical spine. SETTING: University based laboratory in Chicago, IL, USA. METHODS: Five participants with chronic WAD were compared with five gender/age/height/weight/body mass index (BMI) control participants. For a secondary investigation, the chronic WAD group was compared with five unmatched participants with motor incomplete spinal cord injury (iSCI). Spinal cord motor tract integrity was assessed using magnetization transfer imaging. Muscle fat infiltration (MFI) was quantified using fat/water separation magnetic resonance imaging. Central volitional muscle activation of the plantarflexors was assessed using a burst superimposition technique. RESULTS: We found reduced spinal cord motor tract integrity, increased MFI of the neck and lower extremity muscles and significantly impaired voluntary plantarflexor muscle activation in five participants with chronic WAD. The lower extremity structural changes and volitional weakness in chronic WAD were comparable to participants with iSCI. CONCLUSION: The results support the position that a subset of the chronic whiplash population may have sustained partial damage to the spinal cord. SPONSORSHIP: NIH R01HD079076-01A1, NIH T32 HD057845 and the Foundation for Physical Therapy Promotion of Doctoral Studies program.

Three-dimensional hemodynamics in intracranial aneurysms: influence of size and morphology.

PURPOSE: To use four-dimensional (4D)-flow MRI for the comprehensive in vivo analysis of hemodynamics and its relationship to size and morphology of different intracranial aneurysms (IA). We hypothesize that different IA groups, defined by size and morphology, exhibit different velocity fields, wall shear stress, and vorticity. MATERIALS AND METHODS: The 4D-flow MRI (spatial resolution = 0.99-1.8 × 0.78-1.46 × 1.2-1.4 mm(3) , temporal resolution = 44-48 ms) was performed in 19 IAs (18 patients, age = 55.4 ± 13.8 years) with saccular (n = 16) and fusiform (n = 3) morphology and different sizes ranging from small (n = 8; largest dimension = 6.2 ± 0.4 mm) to large and giant (n = 11; 25 ± 7 mm). Analysis included quantification of volumetric spatial-temporal velocity distribution, vorticity, and wall shear stress (WSS) along the aneurysm's 3D surface. RESULTS: The 4D-flow MRI revealed distinct hemodynamic patterns for large/giant saccular aneurysms (Group 1), small saccular aneurysms (Group 2), and large/giant fusiform aneurysms (Group 3). Saccular IA (Groups 1, 2) demonstrated significantly higher peak velocities (P < 0.002) and WSS (P < 0.001) compared with fusiform aneurysms. Although intra-aneurysmal 3D velocity distributions were similar for Group 1 and 2, vorticity and WSS was significantly (P < 0.001) different (increased in Group 1 by 54%) indicating a relationship between IA size and hemodynamics. Group 3 showed reduced velocities (P < 0.001) and WSS (P < 0.001). CONCLUSION: The 4D-flow MRI demonstrated the influence of lesion size and morphology on aneurysm hemodynamics suggesting the potential of 4D-flow MRI to assist in the classification of individual aneurysms.

Cardiac magnetic resonance T2 mapping in the monitoring and follow-up of acute cardiac transplant rejection: a pilot study.

BACKGROUND: Acute rejection is a major factor impacting survival in the first 12 months after cardiac transplantation. Transplant monitoring requires invasive techniques. Cardiac magnetic resonance (CMR), noninvasive testing, has been used in monitoring heart transplants. Prolonged T2 relaxation has been related to transplant edema and possibly rejection. We hypothesize that prolonged T2 reflects transplant rejection and that quantitative T2 mapping will concur with the pathological and clinical findings of acute rejection. METHODS AND RESULTS: Patients were recruited within the first year after transplantation. Biopsies were graded according to the International Society for Heart Lung Transplant system for cellular rejection with immunohistochemistry for humoral rejection. Rejection was also considered if patients presented with signs and symptoms of hemodynamic compromise without biopsy evidence of rejection who subsequently improved with treatment. Patients underwent a novel single-shot T2-prepared steady-state free precession 4-chamber and 3 short axis sequences and regions of interest were drawn overlying T2 maps by 2 independent blinded reviewers. A total of 74 (68 analyzable) CMRs T2 maps in 53 patients were performed. There were 4 cellular, 2 humoral, and 2 hemodynamic rejection cases. The average T2 relaxation time for grade 0R (n=46) and grade 1R (n=17) was 52.5±2.2 and 53.1±3.3 ms (mean±SD), respectively. The average T2 relaxation for grade 2R (n=3) was 59.6±3.1 ms and 3R (n=1) was 60.3 ms (all P value <0.05 compared with controls). The T2 average in humoral rejection cases (n=2) was 59.2±3.3 ms and the hemodynamic rejection (n=2) was 61.1±1.8 ms (P<0.05 versus controls). The average T2 relaxation time for all-cause rejection versus no rejection is 60.1±2.1 versus 52.8±2.7 ms (P<0.05). All rejection cases were rescanned 2.5 months after treatment and demonstrated T2 normalization with average of 51.4±1.6 ms. No difference was found in ventricular function between nonrejection and rejection patients, except in ventricular mass 107.8±10.3 versus 127.5±10.4 g (P < 0.05). CONCLUSIONS: Quantitative T2 mapping offers a novel noninvasive tool for transplant monitoring, and these initial findings suggest potential use in characterizing rejections. Given the limited numbers, a larger multi-institution study may help elucidate the benefits of T2 mapping as an adjunctive tool in routine monitoring of cardiac transplants.