Intra-Examiner Reliability and Validity of Sagittal Cervical Spine Mensuration Methods Using Deep Convolutional Neural Networks.

Background: The biomechanical analysis of spine and postural misalignments is important for surgical and non-surgical treatment of spinal pain. We investigated the examiner reliability of sagittal cervical alignment variables compared to the reliability and concurrent validity of computer vision algorithms used in the PostureRay® software 2024. Methods: A retrospective database of 254 lateral cervical radiographs of patients between the ages of 11 and 86 is studied. The radiographs include clearly visualized C1-C7 vertebrae that were evaluated by a human using the software. To evaluate examiner reliability and the concurrent validity of the trained CNN performance, two blinded trials of radiographic digitization were performed by an extensively trained expert user (US) clinician with a two-week interval between trials. Then, the same clinician used the trained CNN twice to reproduce the same measures within a 2-week interval on the same 254 radiographs. Measured variables included segmental angles as relative rotation angles (RRA) C1-C7, Cobb angles C2-C7, relative segmental translations (RT) C1-C7, anterior translation C2-C7, and absolute rotation angle (ARA) C2-C7. Data were remotely extracted from the examiner's PostureRay® system for data collection and sorted based on gender and stratification of degenerative changes. Reliability was assessed via intra-class correlations (ICC), root mean squared error (RMSE), and R2 values. Results: In comparing repeated measures of the CNN network to itself, perfect reliability was found for the ICC (1.0), RMSE (0), and R2 (1). The reliability of the trained expert US was in the excellent range for all variables, where 12/18 variables had ICCs ≥ 0.9 and 6/18 variables were 0.84 ≤ ICCs ≤ 0.89. Similarly, for the expert US, all R2 values were in the excellent range (R2 ≥ 0.7), and all RMSEs were small, being 0.42 ≤ RMSEs ≤ 3.27. Construct validity between the expert US and the CNN network was found to be in the excellent range with 18/18 ICCs in the excellent range (ICCs ≥ 0.8), 16/18 R2 values in the strong to excellent range (R2 ≥ 0.7), and 2/18 in the good to moderate range (R2 RT C6/C7 = 0.57 and R2 Cobb C6/C7 = 0.64. The RMSEs for expert US vs. the CNN network were small, being 0.37 ≤ RMSEs ≤ 2.89. Conclusions: A comparison of repeated measures within the computer vision CNN network and expert human found exceptional reliability and excellent construct validity when comparing the computer vision to the human observer.

Abnormal Static Sagittal Cervical Curvatures following Motor Vehicle Collisions: A Retrospective Case Series of 41 Patients before and after a Crash Exposure.

Previous investigations have found a correlation between abnormal curvatures and a variety of patient complaints such as cervical pain and disability. However, no study has shown that loss of the cervical curve is a direct result of exposure to a motor vehicle collision (MVC). This investigation presents a retrospective consecutive case series of patients with both a pre-injury cervical lateral radiograph (CLR) and a post-injury CLR after exposure to an MVC. Computer analysis of digitized vertebral body corners on CLRs was performed to investigate the possible alterations in the geometric alignment of the sagittal cervical curve. METHODS: Three spine clinic records were reviewed over a 2-year period, looking for patients where both an initial lateral cervical X-ray and an examination were performed prior to the patient being exposed to a MVC; afterwards, an additional exam and radiographic analysis were obtained. A total of 41 patients met the inclusion criteria. Examination records of pain intensity on numerical pain rating scores (NPRS) and neck disability index (NDI), if available, were analyzed. The CLRs were digitized and modeled in the sagittal plane using curve fitting and the least squares error approach. Radiographic variables included total cervical curve (ARA C2-C7), Chamberlain's line to horizontal (skull flexion), horizontal translation of C2 relative to C7, segmental translations (retrolisthesis and anterolisthesis), and circular modelling radii. RESULTS: There were 15 males and 26 females with an age range of 8-65 years. Most participants were drivers (28) involved in rear-end impacts (30). The pre-injury NPRS was 2.7 while the post injury was 5.0; p < 0.001. The NDI was available on 24/41 (58.5%) patients and increased after the MVC from 15.7% to 32.8%, p < 0.001. An altered cervical curvature was identified following exposure to MVC, characterized by an increase in the mean radius of curvature (265.5 vs. 555.5, p < 0.001) and an approximate 8° reduction of lordosis from C2-C7; p < 0.001. The mid-cervical spine (C3-C5) showed the greatest curve reduction with an averaged localized mild kyphosis at these levels. Four participants (10%) developed segmental translations that were just below the threshold of instability, segmental translations < 3.5 mm. CONCLUSIONS: The post-exposure MVC cervical curvature was characterized by an increase in radius of curvature, an approximate 8° reduction in C2-C7 lordosis, a mild kyphosis of the mid-cervical spine, and a slight increase in anterior translation of C2-C7 sagittal balance. The modelling result indicates that the post-MVC cervical sagittal alignment approximates a second-order buckling alignment, indicating a significant alteration in curve geometry. Future biomechanics experiments and clinical investigations are needed to confirm these findings.

Three dimensional evaluation of posture in standing with the PosturePrint: an intra- and inter-examiner reliability study.

BACKGROUND: Few digitizers can measure the complexity of upright human postural displacements in six degrees of freedom of the head, rib cage, and pelvis. METHODS: In a University laboratory, three examiners performed delayed repeated postural measurements on forty subjects over two days. Three digital photographs (left lateral, AP, right lateral) of each of 40 volunteer participants were obtained, twice, by three examiners. Examiners placed 13 markers on the subjects before photography and chose 16 points on the photographic images. Using the PosturePrint internet computer system, head, rib cage, and pelvic postures were calculated as rotations (Rx, Ry, Rz) in degrees and translations (Tx, Tz) in millimeters. For reliability, two different types (liberal = ICC(3,1) & conservative = ICC(2,1)) of inter- and intra-examiner correlation coefficients (ICC) were calculated. Standard error of measurements (SEM) and mean absolute differences within and between observers' measurements were also determined. RESULTS: All of the "liberal" ICCs were in the excellent range (> 0.84). For the more "conservative" type ICCs, four Inter-examiner ICCs were in the interval (0.5-0.6), 10 ICCs were in the interval (0.61-0.74), and the remainder were greater than 0.75. SEMs were 2.7 degrees or less for all rotations and 5.9 mm or less for all translations. Mean absolute differences within examiners and between examiners were 3.5 degrees or less for all rotations and 8.4 mm or less for all translations. CONCLUSION: For the PosturePrint system, the combined inter-examiner and intra-examiner correlation coefficients were in the good (14/44) and excellent (30/44) ranges. SEMs and mean absolute differences within and between examiners' measurements were small. Thus, this posture digitizer is reliable for clinical use.

Conservative treatment of a patient with previously unresponsive whiplash-associated disorders using clinical biomechanics of posture rehabilitation methods.

OBJECTIVE: To describe the treatment of a patient with chronic whiplash-associated disorders (WADs) previously unresponsive to multiple physical therapy and chiropractic treatments, which resolved following Clinical Biomechanics of Posture (CBP) rehabilitation methods. CLINICAL FEATURES: A 40-year-old man involved in a high-speed rear-impact collision developed chronic WADs including cervicothoracic, shoulder, and arm pain and headache. The patient was diagnosed with a confirmed chip fracture of the C5 vertebra and cervical and thoracic disk herniations. He was treated with traditional chiropractic and physical therapy modalities but experienced only temporary symptomatic reduction and was later given a whole body permanent impairment rating of 33% by an orthopedic surgeon. INTERVENTION AND OUTCOME: The patient was treated with CBP mirror-image cervical spine adjustments, exercise, and traction to reduce forward head posture and cervical kyphosis. A presentation of abnormal head protrusion resolved and cervical kyphosis returned to lordosis posttreatment. His initial neck disability index was 46% and 0% at the end of care. Verbal pain rating scales also improved for neck pain (from 5/10 to 0/10). CONCLUSION: A patient with chronic WADs and abnormal head protrusion, cervical kyphosis, and disk herniation experienced an improvement in symptoms and function after the use of CBP rehabilitation protocols when other traditional chiropractic and physical therapy procedures showed little or no lasting improvement.

Validation of a computer analysis to determine 3-D rotations and translations of the rib cage in upright posture from three 2-D digital images.

Since thoracic cage posture affects lumbar spine coupling and loads on the spinal tissues and extremities, a scientific analysis of upright posture is needed. Common posture analyzers measure human posture as displacements from a plumb line, while the PosturePrint claims to measure head, rib cage, and pelvic postures as rotations and translations. In this study, it was decided to evaluate the validity of the PosturePrint Internet computer system's analysis of thoracic cage postures. In a university biomechanics laboratory, photographs of a mannequin thoracic cage were obtained in different postures on a stand in front of a digital camera. For each mannequin posture, three photographs were obtained (left lateral, right lateral, and AP). The mannequin thoracic cage was placed in 68 different single and combined postures (requiring 204 photographs) in five degrees of freedom: lateral translation (Tx), lateral flexion (Rz), axial rotation (Ry), flexion-extension (Rx), and anterior-posterior translation (Tz). The PosturePrint system requires 13 reflective markers to be placed on the subject (mannequin) during photography and 16 additional "click-on" markers via computer mouse before a set of three photographs is analyzed by the PosturePrint computer system over the Internet. Errors were the differences between the positioned mannequin and the calculated positions from the computer system. Average absolute value errors were obtained by comparing the exact inputted posture to the PosturePrint computed values. Mean and standard deviation of computational errors for sagittal displacements of the thoracic cage were Rx=0.3+/-0.1 degrees , Tz=1.6+/-0.7 mm, and for frontal view displacements were Ry=1.2+/-1.0 degrees , Rz=0.6+/-0.4 degrees , and Tx=1.5+/-0.6 mm. The PosturePrint system is sufficiently accurate in measuring thoracic cage postures in five degrees of freedom on a mannequin indicating the need for a further study on human subjects.

Modeling of the sagittal cervical spine as a method to discriminate hypolordosis: results of elliptical and circular modeling in 72 asymptomatic subjects, 52 acute neck pain subjects, and 70 chronic neck pain subjects.

STUDY DESIGN: Computer analysis of digitized vertebral body corners on lateral cervical radiographs. OBJECTIVES: Using elliptical and circular modeling, the geometric shape of the path of the posterior bodies of C2-C7 was sought in normal, acute pain, and chronic pain subjects. To determine the least squares error per point for paths of geometric shapes, minor axis to major axis elliptical ratios (b/a), Cobb angles, sagittal balance of C2 above C7, and posterior tangent segmental and global angles. SUMMARY OF BACKGROUND DATA: When restricted to cervical lordotic configurations, normal, acute pain, and chronic pain subjects have not been compared for similarities or differences of these parameters. Conventional Cobb angles provide only a comparison of the endplates of the distal vertebrae, while geometric modeling provides the shape of the entire sagittal curves, the orientation of the spine, and segmental angles. METHODS: Radiographs of 72 normal subjects, 52 acute neck pain subjects, and 70 chronic neck pain subjects were digitized. For normal subjects, the inclusion criteria were no kyphotic cervical segments, no cranial-cervical symptoms, and less than +/- 10 mm horizontal displacement of C2 above C7. In pain subjects, inclusion criteria were no kyphotic cervical segments and less than 25 mm of horizontal displacement of C2 above C7. Measurements included segmental angles, global angles of lordosis (C1-C7 and C2-C7), height-to-length ratios, anterior weight bearing, and from modeling, circular center, and radius of curvature. RESULTS: In the normal group, a family of ellipses was found to closely approximate the posterior body margins of C2-C7 with a least squares error of less than 1 mm per vertebral body point. The only ellipse/circle found to include T1, with a least squares error of less than 1 mm, was a circle. Compared with the normal group, the pain group's mean radiographic angles were reduced and the radius of curvature was larger. For normal, acute, and chronic pain groups, the mean angles between posterior tangents on C2-C7 were 34.5 degrees, 28.6 degrees, and 22.0 degrees, C2-C7 Cobb angles were 26.8 degrees, 16.5 degrees, and 12.7 degrees, and radius of curvature were r = 132.8 mm, r = 179 mm, and r = 245.4 mm, respectively. CONCLUSIONS: The mean cervical lordosis for all groups could be closely modeled with a circle. Pain groups had hypolordosis and larger radiuses of curvature compared with the normal group. Circular modeling may be a valuable tool in the discrimination between normal lordosis and hypolordosis in normal and pain subjects.