Virtual preoperative measurement and surgical manipulation of sagittal spinal alignment using a novel research and educational software program

Transcription

1 Neurosurg Focus 28 (3):E2, 2010 Virtual preoperative measurement and surgical manipulation of sagittal spinal alignment using a novel research and educational software program Dav i d B. Pe t t i g r e w, Ph.D., 1 Ch a d J. Mo r g a n, M.D., 2 R. Br i a n And e r s o n, B.A., 3 Ph i l i p A. Wi l s e y, Ph.D., 4 a n d Ch a r l e s Ku n t z IV, M.D. 5 1 Department of Medical Education, University of Cincinnati College of Medicine; 3 Clifton Labs, Inc.; 4 Department of Electrical and Computer Engineering, University of Cincinnati; and 5 University of Cincinnati Neuroscience Institute: Department of Neurosurgery, University of Cincinnati College of Medicine, Mayfield Clinic and Spine Institute, Cincinnati, Ohio; and 2 Springfield Neurological and Spine Institute, Missouri State University, Springfield, Missouri Understanding regional as well as global spinal alignment is increasingly recognized as important for the spine surgeon. A novel software program for virtual preoperative measurement and surgical manipulation of sagittal spinal alignment was developed to provide a research and educational tool for spine surgeons. This first-generation software program provides tools to measure sagittal spinal alignment from the occiput to the pelvis, and to allow for virtual surgical manipulation of sagittal spinal alignment. The software was developed in conjunction with Clifton Labs, Inc. Photographs and radiographs were imported into the software program, and a 2D virtual spine was constructed from the images. The software then measured regional and global sagittal spinal alignment from the virtual spine construct, showing the user how to perform the measurements. After measuring alignment, the program allowed for virtual surgical manipulation, simulating surgical procedures such as interbody fusion, facet osteotomy, pedicle subtraction osteotomy, and reduction of spondylolisthesis, as well as allowing for rotation of the pelvis on the hip axis. Following virtual manipulation, the program remeasured regional and global sagittal spinal alignment. Computer software can be used to measure and manipulate sagittal spinal alignment virtually, providing a new research and educational tool. In the future, more comprehensive programs may allow for measurement and interaction in the coronal, axial, and sagittal planes. (DOI: / FOCUS09283) Ke y Wo r d s virtual measurement computer software spinal alignment spinal deformity In c r e a s i n g emphasis is being placed on understanding regional as well as global sagittal spinal alignment. Sagittal spinal alignment is becoming recognized as an important predictor of patient outcome after spine surgery. 1 3,6,9 A new educational and research software program was developed for virtual sagittal modeling and manipulation of the spine. The objective was to provide a tool that would aid in understanding measurement techniques and the effects of surgical manipulation on alignment. The system features an easy-to-use interface, including a canvas for viewing images and virtual spine models, a measurement window showing automatically calculated angles and displacements with their relationship to those calculated from the asymptomatic population, and a toolbox consisting of 4 possible virtual surgical adjustments that may be applied to the spine. As each adjustment is made, the effect on regional and global Abbreviations used in this paper: CBVA = chin-brow to vertical angle; HA = hip axis; PT = pelvic tilt; SVA = sagittal vertical axis. spinal alignment is immediately depicted by a graphic representation of the adjusted spine. This software system may be useful as a research and educational tool for spine surgeons. Methods The simulation software was developed as a graphic Windows desktop application suitable for use directly by surgeons. Building on high-level development frameworks such as.net 7 and Spring.NET 8 allowed for a greater focus on solving domain-related issues, and less time was needed for solving low-level implementation details. To allow for maximum flexibility over the course of the project while maintaining high confidence in the validity of the domain model, all software development followed an iterative and test-driven methodology. Because the software is user oriented, development began with a nonfunctional visual prototype used to identify the major modes of user interaction. Development 1

2 D. B. Pettigrew et al. Fig. 1. Computer screen capture. The software interface contains a central canvas where imported images and the virtual spine model are displayed. The model definition window is to the left of the canvas. The measurements, images, and adjustments windows are shown to the right. STA = sagittal tilt angle. proceeded in sprints of roughly 1 month, each followed by hands-on user acceptance testing. This iterative development model was very powerful, enabling the surgeons and developers to make significant changes and additions to the software requirements midproject. Much of the development effort was dedicated to capturing a computational model of sagittal spinal manipulation that was both conceptually simple and powerful enough to express a wide range of 2D spinal adjustments. The final domain model consists of 4 major components: 1) a vector representation of each vertebra of the spine prior to adjustment; 2) a series of adjustments representing standard real-world surgical procedures and modeled as mathematical transformations of the vector spine model; 3) a vector representation of each vertebra of the spine following adjustment, automatically calculated by applying adjustments to the original spine model; and 4) a series of standard measurements against both spine models. As the user makes changes to the inputs of any of the components of the model, the entire model updates in real time, with visual feedback. Consequently, minimal user 2 effort is required to experiment with different adjustment options. The model does not require complete information, enabling adjustments and measurements based on the available data, permitting the user to focus on an arbitrary subset of vertebrae or adjustments. Last, the system has undo and reset (undo all adjustments) options to facilitate virtual surgery explorations. This first-generation software program is available to the public free of charge. The virtual modeling and virtual adjustment procedure consists of the following workflow: 1) clinical and radiographic images are imported into the software display; 2) a virtual spine model is constructed using the images as a guide; 3) inspection and analysis of this original spine model is performed; 4) virtual adjustments are performed; and 5) the adjusted spine model is reanalyzed. The interface consists of a central canvas for displaying images (Fig. 1). To construct the original (preoperative) spine model, the user imports a clinical photograph and lateral radiograph into the canvas. The software supports a variety of image types, including DICOM, gif, jpeg, bitmap, png, and tiff. The radiograph is overlaid

3 Virtual sagittal spinal measurement and manipulation Fig. 2. A case illustrating use of the software. Preoperative lateral photographic (A) and lateral radiographic (B) images are shown. Clinical evaluation reveals a horizontal gaze sagittal imbalance. Radiographic evaluation reveals a major structural lumbar kyphotic curve and a minor structural lumbosacral kyphotic curve. Global spinal alignment (horizontal gaze and spinal balance) reveals positive sagittal imbalance. Pelvic alignment reveals posterior sagittal rotation. C: Virtual spine model constructed from preoperative images using the software program. Postoperative lateral photographic (D) and lateral radiographic (E) images illustrate the outcome following L-2 and L-4 pedicle subtraction osteotomies with a spinopelvic fixation and fusion. F: Virtual spine model following virtual surgical L-2 and L-4 pedicle subtraction osteotomies with anterior sagittal rotation of the pelvis on the HA. on the clinical image, and soft tissue in the radiograph is used for fiduciary landmarks (for example, see Fig. 1). To get the images into register, either image can be rotated, flipped, rescaled, or cropped as necessary. The opacity, brightness, or contrast of each image can also be adjusted. Once the images have been imported and put into register, the original spine model can be constructed. To the left of the canvas is the model definition window used to build the original model of the spine from imported images (Fig. 1). The vertebral body at each level is traced manually, each modeled as a quadrilateral. The C-2 odontoid process (designated Peg in the model definition window) is similarly traced and represented as a quadrilateral. The sacrum is modeled by tracing a 7-sided polygon along the borders. As adjacent vertebral bodies are defined, the intervertebral discs are automatically drawn by the software. The hips are represented by tracing circles around the femoral heads. The chin-brow and MacGregor lines are drawn using the defining landmarks in the clinical photograph and radiograph, respectively. The dorsoventral axis is defined by specifying the patient as either right- or leftfacing. Finally, to assign physical units to displacement measurements, the model is calibrated using a known length standard contained in the image. After the original model is defined, analysis and adjustments may be performed. The software automatically measures a variety of angles and displacements from the occiput to the pelvis from the virtual spine model. These values are displayed to the right of the canvas in the measurements window; mean values ± SDs for the asymptomatic adult population are also displayed.4,5 Values lying within 2 SDs of the mean are displayed in green; values lying outside this range are displayed in red. Virtual adjustments to the spine may be made using 3

4 D. B. Pettigrew et al. the adjustments window (Fig. 1). To simulate shaving osteotomy wedges from the vertebral bodies, the angle of either the upper or lower endplate of each vertebra may be adjusted. Material can be added to either the anterior or posterior aspect of each intervertebral disc space. Listhesis can be simulated by specifying a translational shift in either direction between any two vertebrae. Finally, the hip rotation on the HA may be adjusted. This last adjustment is not intended to simulate a surgical procedure per se, but to educate the user regarding the role that hip rotation plays in global sagittal spinal alignment. The user may toggle between the original and adjusted spine models as desired. Angle and displacement measurements are displayed for both models in the measurement window. The user may specify angles to be measured in addition to those that are hard-coded by the software system. The software also constructs a log of all adjustments that are made. Case Illustration A case illustrating use of the software is shown in Fig. 2. This 67-year-old man with ankylosing spondylitis presented with low-back pain and difficulty ambulating. A preoperative lateral photographic image (Fig. 2A) revealed a horizontal gaze sagittal imbalance (CBVA, +67 ). Standing lateral cervical and long-cassette radiographic images (Fig. 2B) revealed a major structural lumbar kyphotic curve (L1 5, +3 ) and a minor structural lumbosacral kyphotic curve (L4 S1, 3 ). Global spinal alignment revealed positive sagittal imbalance (CBVA, +67 ; C7 S1 SVA, +346 mm). Pelvic alignment revealed posterior sagittal rotation (PT, +37 ). Postoperative lateral photographic (Fig. 2D) and long-cassette radiographic (Fig. 2E) images showed improvement in regional as well as global sagittal spinal alignment following L-2 and L-4 pedicle subtraction osteotomies with a spinopelvic fixation and fusion. With improvement in regional and global sagittal spinal alignment, the pelvis had rotated anteriorly, with an improvement in the PT. After importing a preoperative lateral clinical image into the software program, preoperative lateral radiographic images were imported into the software program. The radiographic images brightness, contrast, and opacity were adjusted. The radiographic images were then scaled and rotated to overlay the clinical image, using soft tissue in the radiographs as fiduciary landmarks. A virtual model of the spine was then constructed (Fig. 2C). The software program automatically measured the angles and displacements from the occiput to the pelvis based on the virtual spine model. These values were displayed to the right of the canvas in the measurements window; mean values ± SDs for the asymptomatic adult population were also displayed. 4,5 Values lying within 2 SDs of the mean were displayed in green; values lying outside this range were displayed in red. Using the adjustments window, 30 wedge osteotomies were virtually performed at L-2 and L-4 with a 11 anterior rotation of the pelvis on the HA, resulting in an adjusted model (Fig. 2F) that closely resembled the postoperative condition and confirming that the first-generation software could simulate the operative procedure and provide an educational and research tool. Discussion Spine surgeons are increasingly recognizing the importance of the maintenance or restoration of normal neutral upright sagittal spinal alignment. From occipitocervical fusion to lumbosacral fusion, preservation of neutral upright sagittal spinal alignment has been reported to prevent the postoperative development of deformity and adjacent-segment disease as well as to provide improved postoperative clinical outcomes. 1 3,6,9 Many measurement techniques for evaluating regional and global sagittal spinal alignment have come about in the recent past. For spine surgeons it can be a daunting task to begin to understand measurements of regional and global spinal alignment and the effects of surgical manipulation. To better understand regional and global sagittal spinal alignment, we developed a first-generation, novel software program for virtual preoperative measurement and surgical manipulation of sagittal spinal alignment. By assuming that the spine is a rigid column, simple geometrical principles and fiducial representations of key vertebral segments can serve as markers within the program. Virtual surgical manipulation can then be portrayed and analyzed as it impacts regional and global alignment. Using this working model, the vertebral level selected for an osteotomy, the amount of angular bone removal required during an osteotomy, and their anticipated impact can be assessed for research and education. Conclusions This first-generation program is limited to the sagittal plane, and the surgical manipulation is oversimplified, treating the spine as a rigid column. Despite these limitations, the program does provide educational information to the user on measurement techniques and the effects of virtual surgical manipulation. Future software generations will be more comprehensive, allowing for measurement and interaction in the coronal, axial, and sagittal planes. Disclosure Dr. Kuntz is a stockholder in Mayfield Clinic, Mayfield Spine Center, Precision Radiotherapy, Priority Consult, Cincinnati Imaging, and CKIV Alignment; receives research or education funding from Synthes, Stryker, BioAxone, and AO Spine; and is a consultant for Synthes Spine. Dr. Wilsey is a consultant for Clifton Labs, Inc. Author contributions to the study and manuscript preparation include the following. Conception and design: C Kuntz IV, CJ Morgan, RB Anderson, PA Wilsey. Acquisition of data: CJ Morgan, RB Anderson, PA Wilsey. Analysis and interpretation of data: C Kuntz IV, CJ Morgan, RB Anderson, PA Wilsey. Drafting the article: DB Pettigrew, CJ Morgan, RB Anderson. Critically revising the article: C Kuntz IV, DB Pettigrew, CJ Morgan, PA Wilsey. Reviewed final version of the manuscript and approved it for submission: C Kuntz IV, DB Pettigrew, CJ Morgan, RB Anderson, PA Wilsey. Administrative/technical/material support: RB Anderson, PA Wilsey. Study supervision: C Kuntz IV, PA Wilsey. 4

5 Virtual sagittal spinal measurement and manipulation Acknowledgment The authors thank Martha E. Headworth, M.S., for her assistance with the figures. References 1. Glassman SD, Bridwell K, Dimar JR, Horton W, Berven S, Schwab F: The impact of positive sagittal balance in adult spinal deformity. Spine (Phila Pa 1976) 30: , Hioki A, Miyamoto K, Kodama H, Hosoe H, Nishimoto H, Sakaeda H, et al: Two-level posterior lumbar interbody fusion for degenerative disc disease: improved clinical outcome with restoration of lumbar lordosis. Spine J 5: , Kumar MN, Baklanov A, Chopin D: Correlation between sagittal plane changes and adjacent segment degeneration following lumbar spine fusion. Eur Spine J 10: , Kuntz C IV, Levin LS, Ondra SL, Shaffrey CI, Morgan CJ: Neutral upright sagittal spinal alignment from the occiput to the pelvis in asymptomatic adults: a review and resynthesis of the literature. J Neurosurg Spine 6: , Kuntz C IV, Shaffrey CI, Ondra SL, Durrani AA, Mummaneni PV, Levin LS, et al: Spinal deformity: a new classification derived from neutral upright spinal alignment measurements in asymptomatic juvenile, adolescent, adult, and geriatric individuals. Neurosurgery 63 (3 Suppl):25 39, Matsunaga S, Onishi T, Sakou T: Significance of occipitoaxial angle in subaxial lesion after occipitocervical fusion. Spine (Phila Pa 1976) 26: , Microsoft: The.NET Framework ( NET) [Accessed January 7, 2010] 8. Spring.NET Application Framework ( framework.net) [Accessed January 7, 2010] 9. Suk KS, Kim KT, Lee SH, Kim JM: Significance of chin-brow vertical angle in correction of kyphotic deformity of ankylosing spondylitis patients. Spine (Phila Pa 1976) 28: , 2003 Manuscript submitted November 16, Accepted December 16, Address correspondence to: Charles Kuntz IV, M.D., Division of Spine and Peripheral Nerve Surgery, Department of Neurosurgery, University of Cincinnati, ML 0515, Cincinnati, Ohio charleskuntz@yahoo.com. 5