1. A NON-INVASIVE THREE-DIMENSIONAL SPINAL MOTION ANALYSIS METHOD Tae Hong Lim, Ph.D. Jason C. Eck, B.S. Howard S. An, M.D. Linda M. Mc Grady, B.S.

2. INTRODUCTION l Spinal disorders frequently cause a significant and permanent decrease in quality of life. l The presence of abnormal motion is used as an indication of possible instability in the spine. l Efforts have been made to quantify spinal motion; however, the precise assessment of complex 3-D motion remains difficult.

3. PREVIOUS STUDIES l Spinal instability has been assessed using the following 2-D or invasive methods: – Dynamic (flexion/extension) radiography – Roentgen stereophotogrammetry – Electrogoniometric methods l The concept of using principal axes of the moment of inertia tensor to monitor rigid body orientation and position has been well described for the study of the wrist.

4. OBJECTIVES l Establish a computer-aided system for 3-D non-invasive spinal motion analysis. – Develop software to describe the position of the geometrical center (GC) and orientation of the principal axes (PA). – Describe the 3-D motion of the vertebrae in terms of translation of the GC and rotation of the PA. – Quantify the accuracy of the system using isolated cadaveric vertebrae.

5. METHODS l An isolated vertebra was held in various known rotated and translated position using a custom designed apparatus. l CT images were obtained of the vertebra in 0, 5, and 20 degrees of axial rotation, lateral bending, and extension. l 1.0 mm scans were taken using both axial and either sagittal or coronal planes. l 512 x 512 image matrix was used with a 22.0 cm field of view providing a system resolution of 0.043 cm.

6. METHODS Slider 0 5 10 15 20 25 Lever Arm Figure 1. Diagram of accuracy testing device.

7. METHODS l Images were analyzed using commercial software, C-Med l A region of interest function based on pixel intensity was used to select the vertebra. l A measure function provided data on area, centroid, and moments and product of inertia for each image. l 2-D data from C-Med were used in custom software to calculate 3-D moment of inertia tensor using the parallel axis theorem.

8. METHODS Figure 2. Portion of CT image selected using region of interest function.

9. METHODS l Eigenvalues and eigenvectors were calculated from the tensor using an IMSL subroutine. l Translations of the GC and rotations of the PA were calculated. l Geometrical and inertial properties of the vertebra were compared in various positions to assess reproducibility.

10. RESULTS l All major rotations were within an accuracy of 1.0 degree. l Off-axis rotational errors were found when using axial scans. l Translational accuracy was within 0.1 cm.

11. Table 1. Rotational values using axial CT scans.

12. Table 2. Rotational values using sagittal or coronal CT scans.

13. DISCUSSION l This motion analysis system has an accuracy of 1.0 degrees in rotation and 0.1 cm in translation. l Off-axis errors were found when using axial scans. No errors occurred when using either sagittal or coronal scans. The only variation is in the number of scans obtained. l Geometrical and inertial errors were higher in the axial scans where off-axis motion was predicted. l Thus, system reproducibility is affected by the number of images.

14. DISCUSSION l It was shown that greater accuracy was achieved when using additional scans, but it the optimal number of scans was not determined. l It is hoped that future hardware enhancements will allow thinner image thickness or the ability to scan in additional planes. l Current motion data is based on an isolated cervical vertebra. When this technique is applied to the lumbar spine fewer off-axis errors are expected due to the larger vertebral height.

15. CONCLUSIONS l This is the first system capable of noninvasively measuring 3-D motion segments. l This project has shown the promise of using a series of parallel scans from the CT to calculate the principal axes of the moment of inertia tensor. These can be tracked to calculate 3-D spinal motion. l System accuracy is greater than current 2-D methods and similar to invasive methods.

16. Project 1: In Vivo Analysis of Segmental Spine Motion of the Lumbar Spine Program Project Grant External Scientific Advisory Board Meeting April 25, 2003

17. Long-Term Goal l Develop new methods to diagnose and treat low back pain problems resulting from both segmental instability and degenerative changes in discs and/or facets. – Comprehensive studies of the biomechanical, biological and clinical aspects of degenerative spinal disorders

18. Essential Information In-vivo Relationship Degenerative Changes in Intervertebral Joints Segmental InstabilityLow Back Pain

19. Previous Studies l In vivo studies: – In-vivo relationship among degenerative changes in the intervertebral joint (disc, facets and surrounding structures), segmental hypermobility and low-back pain remains unclear and controversial. – Limitations Inaccurate and motion measurement (2-D measurement of 3-D motion) Subjective grading of degenerative changes Lack of control in experimental protocol (voluntary motion, subject population) l In vitro study : – Established non-invasive 3-D motion analysis method using CT images – In-vitro relationship between rotational flexibility and degenerative changes in IVDs and facets Increasing hypermobility with IVD degeneration up to grade 4 Torsional flexibility is most significantly affected Segmental flexibility is affected by the facet degeneration, too.

20. Specific Aims of Project 1 l To establish the in-vivo relationship between segmental flexibility and degenerative changes in IVDs, facets and surrounding structures: – Normal asymptomatic subjects – Symptomatic back-pain patients (age and gender matched) l To investigate if there are significant differences between the asymptomatic group and the symptomatic patients’ group.

21. General Methods l Radiographs – Dynamic flexion/extension – Comparisons with previous work l Magnetic Resonance Imaging (MRI) – T2 sagittal images – Proton density axial images of facet joints – Disc and facet grading and morphologic measurements l Computer Tomography (CT) – Five positions (neutral and rotated) – Measure translation and rotation of lumbar vertebra – In vivo, non-invasive measurement of motion

22. General Methods l Subject Selection: – Normal subjects (80) – Age and gender matched symptomatic patients (80) l Degenerative changes in the Intervertebral Joints – MRI (T2 sagittal and Proton density axial images) – CT – Plain radiographs l Motion Measurement – CT images to measure passive AR motion – Dynamic radiographs to measure voluntary FLX/EXT motion

23. Subjects to date Age (yrs)NMNFSMSF 20-2910 30-3910 40-4910 50-5910 Total = 160 Remaining = 130

24. Currently tested subjects Data shown in this presentation

25. Radiographs FlexionExtension NF40-01 Dynamic Flexion/Extension

26. Radiographic Data Flexion Extension NF40-01

27. MRI- Disc Grading Fujiwara, et al 2000 Thompson’s Grading Sagittal T2- NF40-01 1- Normal 2- Less distinct np/af 3- Crack in np or af 4- Decreased height 5- Collapse Thompson’s scale

28. MRI- Disc Grading Data (sagittal) 1- Normal 2- Radial tear (HIZ, SI, BA) 3- Collapse Alternative scale

29. MRI- Measurements (sagittal) NF40-01 sagittal T2

30. MRI- Measurement Data

31. MRI- Facet Grading Fujiwara et al. 2000 1- Normal 2- Erosion 3- Cartilage half gone 4- Absence of cartilage Cartilage Degeneration

32. MRI- Subchondral Grading 1- Normal 2- Focal thickening 3- Up to 50% with thickening 4- Greater than 50% with thickening Subchondral Sclerosis Fujiwara et al. 2000

33. MRI- Osteophyte Grading 1- None 2- Possible osteophyte 3- Small osteophyte 4- Large osteophyte Osteophytes Fujiwara et al. 2000

34. MRI- Grading Data (axial)

35. MRI- Measurements (axial) NF40-01 Axial T2

36. CT l Determine torsional stability l Scan subjects – Using Torso Rotation Control Apparatus (TRCA) – 5 Positions Neutral (supine) Rotate Left and Right 30  Rotate Left and Right Full (up to 50  )

37. Torso Rotation Control Apparatus TRCA Rotation Ring TRCA (upper section) CT Scanner RingCT Scanner Bed Neutral Right Full (50  ) Straps to support upper body and head

38. CT Image Processing l Individual CT slices – Use Mimics software to: Threshold for cortical shell definition Hand colored to define vertebrae Create solid model

39. CT Solid Model from Mimics NF40-01 Neutral

40. CT Solid Models (con’t) Neutral Left 30  Left FullRight Full Right 30  Data from NF40-01

41. CT Image Processing l 3-D Images from Mimics l Pro-Engineer 2000i 2 – Centroids – Moment of Inertia Tensor l Matlab® software – Custom in-house programs – Calculate eigen vectors – Calculate rotation and translation between rigid bodies

42. Rigid Body Motion Inertial properties of a rigid body Moment of Inertia -Calculate the principle axes (eigen vectors) -The orientation of the principle axes does not change wrt the rigid body Lim 1994

43. CT Image Processing (con’t) Geometric PropertiesInertial Properties +X= Extension +Y= Left Lateral Bending +Z= Left Axial Rotation

44. Further CT Examples NF40-01

45. Future Plans l Continue data acquisition and processing – Start recruting low-back pain subjects Summer 2003 l By the summer 2004, we expect to produce sufficient data for meaningful statistical analyses and publication.