Multimodal image registration of the scoliotic torso for surgical planning.

BACKGROUND: This paper presents a method that registers MRIs acquired in prone position, with surface topography (TP) and X-ray reconstructions acquired in standing position, in order to obtain a 3D representation of a human torso incorporating the external surface, bone structures, and soft tissues. METHODS: TP and X-ray data are registered using landmarks. Bone structures are used to register each MRI slice using an articulated model, and the soft tissue is confined to the volume delimited by the trunk and bone surfaces using a constrained thin-plate spline. RESULTS: The method is tested on 3 pre-surgical patients with scoliosis and shows a significant improvement, qualitatively and using the Dice similarity coefficient, in fitting the MRI into the standing patient model when compared to rigid and articulated model registration. The determinant of the Jacobian of the registration deformation shows higher variations in the deformation in areas closer to the surface of the torso. CONCLUSIONS: The novel, resulting 3D full torso model can provide a more complete representation of patient geometry to be incorporated in surgical simulators under development that aim at predicting the effect of scoliosis surgery on the external appearance of the patient's torso.

Impact of prone surgical positioning on the scoliotic spine.

OBJECTIVE: To study the impact of patient-specific prone positioning on the sagittal and coronal curves of scoliotic spines, including the impact of various patient and surgical frame factors. SUMMARY OF BACKGROUND DATA: Prone operative positioning has been shown to impact the geometry of various individual spinal segments. Its impact on global spinal geometry and influential factors remains unknown. METHODS: Lateral and coronal radiographs were acquired of 6 scoliotic patients while standing, prone on a dynamically adjustable surgical frame and intraoperatively on the Relton-Hall frame. Standing lateral bending radiographs were also acquired. Lordosis, kyphosis, and Cobb angles were measured in each position. Personalized finite element models (FEMs), including the spine, ribcage, pelvis, and lower limbs were generated for each patient based on their standing radiographs. The FEM's ability to reproduce prone spinal geometry was evaluated by using different values of intervertebral disc elastic moduli: published, optimized based on lateral bending radiographs and optimized based on prone radiographs. The 6 FEMs were then exploited to study the impact of surgical frame cushion configuration, standing curve magnitudes, and patient weight on spinal geometry changes due to prone positioning. RESULTS: All coronal and sagittal curves decreased in the prone position; averaging 12% in lordosis, 19% in kyphosis, 7%, 14%, and 26%, respectively, for proximal thoracic, main thoracic, and thoracolumbar/lumbar Cobb angles. FEM prone simulations yielded best results when optimized by using the prone position radiographs (Δ<5 degrees for all segmental curves). Lateral bending optimization yielded similar results by using published properties. Surgical frame cushion configuration, standing curve magnitudes, and patient weight all had an important impact on spinal geometries with the exception of thoracic cushion longitudinal position. A strong correlation (R=0.86) was found between standing kyphosis and its reduction in the prone position. CONCLUSIONS: Prone positioning results in a reduction of all spinal segmental curves which is dependent on a number of patient and surgical frame factors.

Biomechanical study of patient positioning: influence of lower limb positioning on spinal geometry.

OBJECTIVE: The objective of this study was to use finite element model (FEM) simulations and experimental testing to study the relationship between lower limb positioning for surgeries of the spine and changes in sagittal curves. METHODS: Four volunteers underwent lower limb flexibility and range of motion testing before being placed prone on a new surgical frame where lateral radiographs of their spines were taken in positions of hip flexion (average 48 degrees) and extension (average 13 degrees). Personalized FEMs were created representing each volunteer's spine, rib cage, pelvis, and lower limbs. Optimization of model behavior was performed by adjustment of lower limb muscle initial strains. The FEMs were exploited to examine the impact of more extreme and intermediate lower limb positions; 30 degrees of hip extension to 90 degrees of flexion at intervals of 20 degrees. RESULTS: With increased hip flexion, lordosis and kyphosis decreased to an average of 52% (35 degrees) and 16% (6 degrees), respectively. Personalization of the 4 FEMs allowed reproduction of the experimental results within 5 degrees and their subsequent exploitation showed the linear changes in lordosis and kyphosis between extreme positions decreasing an average of 84% (59 degrees) and 34% (13 degrees) with increased hip flexion. A strong correlation was found between experimental change in lordosis and individual hamstring flexibilities (R=-0.93) which allowed for the development of a predictive equation for lordosis in terms of hip flexion which factors straight leg raise test results. CONCLUSIONS: Knowledge gained through this study can be used to improve intraoperative control of sagittal curves through lower limb positioning.

Assessment of two novel surgical positions for the reduction of scoliotic deformities: lateral leg displacement and hip torsion.

Cobb angles and apical vertebral rotations (AVR) are two of the main scoliosis deformity parameters which spinal instrumentation and fusion techniques aim to reduce. Despite this importance, current surgical positioning techniques do not allow the reduction of these parameters. Two new surgical frame accessory prototypes have been developed: (1) a lateral leg displacer (LLD) allows lateral bending of a patient's legs up to 75° in either direction and (2) a pelvic torsion device (PTD) which allows transverse plane twisting of a patient's pelvis at 30° in either direction while raising the thoracic cushion, opposite to the raised side of the pelvis, by 5 cm. The objective of this study was to evaluate the ability of the LLD and PTD to reduce Cobb angles and AVR. Experimental testing was performed pre-operatively on 12 surgical scoliosis patients prone on an experimental surgical frame. Postero-anterior radiographs of their spines were taken in the neutral prone position on a surgical frame, and then again for 6 with their legs bent towards the convexity of their lowest structural curve, 4 with their pelvis raised on the convex side of their lowest structural curve and one each in opposite LLD and PTD intended use. Use of the LLD allowed for an average supplementary reduction of 16° (39%) for Cobb angle and 9° (33%) for AVR in the lowest structural curve. Use of the PTD allowed for an average supplementary reduction of 9° (19%) for Cobb angle and 17° (48%) for AVR in the lowest structural curve. Both devices were most efficient on thoraco-lumbar/lumbar curves. Opposite of intended use resulted in an increase in both Cobb angle and AVR. The LLD and PTD provide interesting novel methods to reduce Cobb angles and AVR through surgical positioning which can be used to facilitate instrumentation procedures by offering an improved intra-operative geometry of the spine.

The impact of intra-operative sternum vertical displacement on the sagittal curves of the spine.

Patient positioning is an important step in spinal surgeries. Many surgical frames allow for lumbar lordosis modulation due to lower limb displacement, however, they do not include a feature which can modulate thoracic kyphosis. A sternum vertical displacer (SVD) prototype has been developed which can increase a subject's thoracic kyphosis relative to the neutral prone position on a surgical frame. The kyphosis increase is obtained by lifting the subject's torso off the thoracic cushions with a dedicated sternum cushion that can be displaced vertically. The objective of this study was to evaluate the impact of SVD utilization on the sagittal curves of the spine. Experimental testing was performed on six healthy volunteers. Lateral radiographs were taken in the neutral and sternum raised positions and then analyzed in order to compare the values of sagittal curves. The displacement of volunteers and surgical frame components between positions was recorded using an optoelectronic device. Finally, interface pressures between the volunteers and surgical frame cushions were recorded using a force sensing array. Average results show that passing from the neutral to sternum raised positions caused an increase of 53% in thoracic kyphosis and 24% in lumbar lordosis; both statistically significant. Sensors showed that the sternum was raised a total of 8 cm and that interface pressures were considerably higher in the raised position. The SVD provides a novel way of increasing a patient's thoracic kyphosis intra-operatively which can be used to improve access to posterior vertebral elements and improve sagittal balance. It is recommended that its use should be limited in time due to the increase in interface pressures observed.

Optimized use of multi-functional positioning frame features for scoliosis surgeries.

A multi-functional positioning frame (MFPF) has been developed which includes a number of positioning features allowing for hip flexion and extension, thorax vertical displacement, lateral leg displacement, pelvic torsion and thorax lateral displacement. The objective of this study was to develop a method allowing for optimized combined use of the MFPF features. Finite element models (FEMs) representing the osseo-ligamentous structures of the spine, ribcage, pelvis and lower limbs, including muscles, were created for three different curve types (main thoracic, double major, and triple major) using a radiographic bi-planar reconstruction technique. Each FEM was subjected to an experimental design in which MFPF features were independently and simultaneously varied between extreme positions and the resultant changes in spinal geometry measured. Optimization of individual spinal geometrical parameters showed variability between curve types and some patterns such as minimum Cobb with lower limbs displaced laterally towards the convexity, pelvis raised on the side of concavity, and thorax laterally displaced towards the thoracic concavity. A weighted and normalized global optimization equation was developed which accounts for the relative importance and desired values of each geometrical parameter. Combined use of MFPF features and adjustments offers a wider range of possible intra-operative spinal geometries than their individual use.

Dynamic positioning of scoliotic patients during spine instrumentation surgery.

STUDY DESIGN: Evaluation of a novel Dynamic Positioning Frame (DPF) for scoliosis surgery to improve presurgical correction. OBJECTIVE: To assess a DPF for scoliosis surgery and demonstrate how its design offers additional presurgical correction of the 3-dimensional deformity. SUMMARY OF BACKGROUND DATA: Patient positioning is an important step in scoliosis surgery. Although current positioning frames focus on supporting the patient and keeping the abdomen pendulous to reduce blood loss, not much has been carried out to explore the aspect of dynamic positioning during surgery. When lying prone, there is some spontaneous correction of the scoliotic deformity owing to gravity and anesthesia. METHODS: Trunk 3-dimensional geometry and pressure at the patient-cushion interface were measured for 12 unanesthetized patients in various positions-standing (PI), lying prone on the DPF (PII), lying prone on the DPF with applied corrective forces (PIII), and lying prone on the Relton-Hall (R-H) frame for comparison (PIV). RESULTS: Spine height increased significantly in prone as compared with that in standing position. When lying on the DPF with corrective forces, there was an improvement in the patients' transverse plane deformity and rib hump with greater retention of kyphosis. There was also an improvement in the rib hump as compared with that in the R-H frame. Higher pressures were recorded on the DPF as compared with the R-H frame, but were reduced to similar values when larger cushions were used. CONCLUSIONS: The DPF provides a novel way of modifying the patient's position preoperatively and intraoperatively. Dynamic patient positioning, coupled with applied corrective forces, allows for increased reduction of the scoliotic deformity as compared with the R-H frame. Further investigation is required to optimize cushion placement and thus, insure safe patient-cushion interface pressures.

Computer-assisted design and finite element simulation of braces for the treatment of adolescent idiopathic scoliosis using a coronal plane radiograph and surface topography.

BACKGROUND: Orthopedic braces made by Computer-Aided Design and Manufacturing and numerical simulation were shown to improve spinal deformities correction in adolescent idiopathic scoliosis while using less material. Simulations with BraceSim (Rodin4D, Groupe Lagarrigue, Bordeaux, France) require a sagittal radiograph, not always available. The objective was to develop an innovative modeling method based on a single coronal radiograph and surface topography, and assess the effectiveness of braces designed with this approach. METHODS: With a patient coronal radiograph and a surface topography, the developed method allowed the 3D reconstruction of the spine, rib cage and pelvis using geometric models from a database and a free form deformation technique. The resulting 3D reconstruction converted into a finite element model was used to design and simulate the correction of a brace. The developed method was tested with data from ten scoliosis cases. The simulated correction was compared to analogous simulations performed with a 3D reconstruction built using two radiographs and surface topography (validated gold standard reference). FINDINGS: There was an average difference of 1.4°/1.7° for the thoracic/lumbar Cobb angle, and 2.6°/5.5° for the kyphosis/lordosis between the developed reconstruction method and the reference. The average difference of the simulated correction was 2.8°/2.4° for the thoracic/lumbar Cobb angles and 3.5°/5.4° the kyphosis/lordosis. INTERPRETATION: This study showed the feasibility to design and simulate brace corrections based on a new modeling method with a single coronal radiograph and surface topography. This innovative method could be used to improve brace designs, at a lesser radiation dose for the patient.

Modelling skin pelvic landmark coordinates into corresponding internal bone for wheelchair users.

The purpose of this study was to investigate the relationships, by linear regression, between internal and external pelvic landmarks identified by two techniques: manual digitization or skin markers. It was hypothesized that the body mass index or the skinfold thickness are significant variables in these relationships. The internal pelvic landmarks were obtained with a stereoradiographic method. Results showed that the external coordinates are generally statistically different from the internal ones; manual digitization of the landmark reduces the soft tissue artifacts compared to the use of skin markers. Different regression models were obtained according to the external acquisition method. Body mass index or skinfold thickness was generally included as a significant variable in models along the direction of the soft tissue thickness: postero-anterior direction for the anterior-superior iliac spine, medio-lateral direction for the apex of the iliac crests. With the use of skin markers, models obtained for a specific internal landmark coordinate include generally many variables, such as the other two coordinates of the landmark, body mass index, or skinfold measurements. This study presented preliminary results on the relationships between internal and external pelvic landmark coordinates. More research is needed before the full relationships are understood and adequate models are developed.

[Tridimensional evaluation and optimization of the orthotic treatment of adolescent idiopathic scoliosis]. / Evaluation tridimensionnelle et optimisation du traitement orthopédique de la scoliose idiopathique adolescente.

Adolescent idiopathic scoliosis involves complex tridimensional deformities of the spine, rib cage and pelvis. Moderate curves generally are treated using an orthosis. This paper presents different studies performed over the last fifteen years related to the biomechanical evaluation and optimization of the orthopedic treatment of scoliotic deformities. Patient specific 3D models of the spine, pelvis and rib cage are computed from calibrated radiographs, and are used to calculate 2D and 3D clinical indices. The torso shape is acquired using surface topography. With such internal and external 3D models, the efficacy of the most frequently used orthoses can be analyzed and new treatments can be developed. Pressures generated by a brace on the patient's trunk were measured using a flexible matrix of pressure sensors and displayed over the patient's internal geometry in order to analyze the brace efficacy. Patient specific finite element models have been developed, including the osseo-ligamentous structures as well as the muscles, the neuro-control, trunk growth and its adaptation to the stress. These models were used to analyze the effects of the Boston brace. The electro-myographic activity also was measured to analyze the << active >> correction mechanisms. Adjustment techniques and software are used to help the orthotists with real time feedback when the brace is being fabricated and adjusted to the patient. Residual growth potential is also being added to the computer model to simulate the long term effect of a brace. The improvement of the orthotic treatments of scoliotic deformities is very encouraging. The exploitation of such tools is expected to allow reaching optimal treatment personalized to each patient. double dagger.

Accessing the influence of repositioning on the pelvis' 3-D orientation in wheelchair users.

This study aimed at evaluating the effects of mechanical repositioning, obtained by the increase in seat-to-back (STB) and system tilt angles, on the position of the pelvis with spinal-cord injured subjects seated in a wheelchair. The noninvasive method used combined magnetic resonance imaging (MRI) images of the whole pelvis obtained in a supine posture and ultrasound images of the pelvic iliac crests obtained in four seating positions. The matching of the two image data sets enabled the location of fourteen pelvic landmarks in the seated positions. From these landmarks, the pelvic tilt, obliquity, and transverse rotation, and the three-dimensional (3-D) motion of the pelvis were calculated. Results showed that the increase in STB angle is not equal to the calculated increase in pelvic tilt and that the pelvis rotated posteriorly, moved forward and downwards. An increase in the system tilt moved the pelvis rearwards and downwards, which counter-balanced the movement seen with the increase in STB. At the return to the first position, no significant changes were observed in the pelvis' position and orientation compared to the initial posture. Results also demonstrated the importance in calculating the total 3-D rotations and translations to characterize adequately the pelvic movement.

Biomechanical simulations of scoliotic spine correction due to prone position and anaesthesia prior to surgical instrumentation.

BACKGROUND: The positioning of patients during scoliosis surgery has been shown to affect the scoliosis curve, yet positioning has not been exploited to help improve surgical outcome from a biomechanics point of view. Biomechanical models have been used to study other aspects of scoliosis. The goal of this study is to simulate the specific influence of the prone operative position and anaesthesia using a finite element model with patient personalized material properties. METHODS: A finite element model of the spine, ribcage and pelvis was created from the 3D standing geometry of two patients. To this model various positions were simulated. Initially the left and right supine pre-operative bending were simulated. Using a Box-Benkin experimental design the material properties of the intervertebral disks were personalized so that the bending simulations best matched the bending X-rays. The prone position was then simulated by applying the appropriate boundary conditions and gravity loads and the 3D geometry was compared to the X-rays taken intra-operatively. Finally an anaesthesia factor was added to the model to relax all the soft tissues. FINDINGS: The behaviour of the model improved for all three positions once the material properties were personalized. By incorporating an anaesthesia factor the results of the prone intra-operative simulation better matched the prone intra-operative X-ray. However, the anaesthesia factor was different for both patients. For the prone position simulation with anaesthesia patient 1 corrected from 62 degrees to 47 degrees and 43 degrees to 31 degrees. Patient 2 corrected from 70 degrees to 55 degrees and 40 degrees to 32 degrees for the thoracic and lumbar curves respectively. INTERPRETATION: Positioning of the patient, as well as anaesthesia, provide significant correction of the spinal deformity even before surgical instrumentation is fixed to the vertebra. The biomechanical effect of positioning should be taken into consideration by surgeons and possibly modify the support cushions accordingly to maximise 3D curve correction. The positioning is an important step that should not be overlooked by when simulating surgical correction and biomechanical models could be used to help determine optimal cushion placement.

Development and evaluation of a new body-seat interface shape measurement system.

A new system has been developed to capture the body-seat interface shape. It can repeatedly and accurately measure interface deformation. The shape sensing array system uses optical fiber technology and is noninvasive. The system can cover an interface as large as 400 x 480 mm and the shape is measured over a 10 x 12 array of sensors laminated on ribbon substrates. The accuracy and repeatability of this system were assessed. Measurement errors were evaluated by comparing the shape with a reference shape obtained by a mechanical digitizer. The root-mean-square error in the Z direction for the system was 3.79 mm. The repeatability of the system was within 0.38 mm under controlled conditions. Different interface materials noticeably affected measurements. With the development of this interface shape measurement device, the basic information gathered through its use may prove to be fundamental in the successful design of generic-shape contoured support surfaces. Furthermore, we expect that the new shape measurement device will provide a quick and effective tool for cushion evaluation and clinical guidelines for cushion prescription.

Differences between pelvic skin and bone landmark identification in different seated positions on spinal-cord injured subjects.

The purpose of this paper was to determine the differences between internal and external pelvic landmark locations in different seating positions. A computer tool developed for the registration of two series of images was used to obtain the internal geometry. First, images of the pelvis were acquired by magnetic resonance imaging (MRI) for each subject, in a supine position; internal landmarks were then identified on the images. Second, ultrasound images of the iliac crests were acquired in four seated positions. A registration algorithm was applied to obtain the transformation matrix between the two image reference systems. The MRI anatomical landmarks were, therefore, transferred into the ultrasound referential, to obtain their three-dimensional (3-D) location in the different seating positions. The external landmarks in those seated positions were identified with a 3-D digitizer. The results revealed that generally the internal and external coordinates of corresponding landmarks are statistically different. The differences are not only due to soft tissue thickness but also to different interpretations of the landmarks' locations between the supine and the seated postures. However, these differences generally did not affect significantly the accuracy with which orientation indexes can be estimated (pelvic tilt, obliquity, transverse rotation). Correlations were found between the internal and external coordinates, implying that linear regressions can be established.

Indices of torso asymmetry related to spinal deformity in scoliosis.

OBJECTIVE: To develop indices that quantify 360 degrees torso surface asymmetry sufficiently well to estimate the Cobb angle of scoliotic spinal deformity within the clinically important 5-10 degrees range. DESIGN: Prospective study in 48 consecutive adolescent scoliosis patients (Cobb angles 10-71 degrees ). BACKGROUND: Scoliotic surface asymmetry has been quantified on the back surface by indices such as back surface rotation (BSR) and curvature of the spinous process line and torso centroid line, though with limited success in spinal deformity estimation. Quantification of 360 degrees torso shape may enhance surface-spine correlation and permit reduced use of harmful X-rays in scoliosis. METHODS: For each patient a 3D torso surface model was generated concurrently with postero-anterior X-rays. We computed indices describing principal axis orientation, back surface rotation, and asymmetry of the torso centroid line, left and right half-areas and the spinous process line. We calculated correlations of each index to the Cobb angle and used stepwise regression to estimate the Cobb angle. RESULTS: Several torso asymmetry indices correlated well to the Cobb angle (r up to 0.8). The Cobb angle was best estimated by age, rib hump and left-right variation in torso width in unbraced patients and by centroid lateral deviation in braced patients. A regression model estimated the Cobb angle from torso indices within 5 degrees in 65% of patients and 10 degrees in 88% (r=0.91, standard error=6.1 degrees ). CONCLUSION: Consideration of 360 degrees torso surface data yielded indices that correlated well to the Cobb angle and estimated the Cobb angle within 10 degrees in 88% of cases. RELEVANCE: The torso asymmetry indices developed here show a strong surface-spine relation in scoliosis, encouraging development of a model to detect scoliosis magnitude and progression from the surface shape with minimal X-ray radiation.

Comparison of Cobb angles measured manually, calculated from 3-D spinal reconstruction, and estimated from torso asymmetry.

While scoliotic spinal deformity is traditionally measured by the Cobb angle, we seek to estimate scoliosis severity from the torso surface without X-ray radiation. Here, we measured the Cobb angle in three ways: by protractor from postero-anterior X-ray, by computer from a 3-D digitized model of the vertebral body line, and by neural-network estimation from indices of torso surface asymmetry. The estimates of the Cobb angle by computer and by neural network were equally accurate in 153 records from 52 patients (standard deviation of 6 degrees from the Cobb angle, r=0.93), showing that torso asymmetry reliably predicted spinal deformity. Further improvements in predictive accuracy may require estimation of other 3-D indices of spinal deformity besides the Cobb angle with its wide measurement variability.

Powered tilt/recline systems: why and how are they used?

Prolonged static sitting can lead to discomfort, pain, pressure sores, spinal curvatures, and loss of functional independence. In order to counteract these harmful effects, adjustable tilt and/or recline systems are often prescribed. Considering the current context of assistive technology service delivery and budget cuts, it is essential to have a better knowledge of the use of these technical aids and user's satisfaction with them. The purpose of this study was to characterize the use of powered tilt and recline systems. A questionnaire was developed for this purpose, and 40 subjects were interviewed at home. They were asked to identify, from a list of 25 objectives, the reasons for which they used their repositioning system and to rank these reasons in order of importance. For each objective, they were also asked to identify the frequency and range of use as well as their satisfaction level with their system. Results revealed that 97.5% of the subjects were using their powered tilt and recline system everyday, and their satisfaction was high. The main objectives for using this type of assistive technology were to increase comfort and to promote rest. Although mainly descriptive, results are of clinical relevance and can be helpful when selecting wheelchairs.

3D reconstruction and analysis of the whole trunk surface for non-invasive follow-up of scoliotic deformities.

The purpose of this study was to evaluate the 3D reconstruction accuracy of a new technology that allows the acquisition of the whole trunk and to develop a software to analyse the trunk surface asymmetry. A non-invasive active vision system provides a 3D textured reconstruction of the whole trunk. The analysis system provides the clinician with quantitative indices that characterize the whole external trunk asymmetry.

Relationships between strap tension, interface pressures and spine correction in brace treatment of scoliosis.

The Boston brace has been shown to efficiently prevent scoliosis curve progression. However, it rarely achieves complete 3-D correction; its adjustment being often empirical and its biomechanical modes of action still remaining poorly understood. This study investigates the in-brace spinal shape (correction) in relationship with the patient's out-of-brace deformation, and the adjustment parameters of the brace, namely the strap tensions and the equivalent forces calculated at the patient-brace interface. Many of the observed relationships illustrate the fact that the in-brace spinal shape is strongly related to the characteristics of the patient's out-of-brace deformation. The pattern of pressure distribution as described by the equivalent forces computed in the thoracic, lumbar, pelvic and sternal regions has important effect on the in-brace Cobb angles, lumbar lordosis, frontal and sagittal imbalances and the apical axial rotations. The complex role of the strap tension on the correction has not been explained and needs further investigation. This project has the potential to give insight into the biomechanical effects of brace treatment by providing a statistical model leading to more rational and personalized brace adjustments.

Prediction of spinal deformity in scoliosis from geometric torsion.

The shape of a curved line that passes through thoracic and lumbar vertebrae is often used to study spinal deformity with measurements in "auxiliary" planes that are not truly three-dimensional (3D). Here we propose a new index, the geometric torsion, which could uniquely describe the spinal deformity. In this study we assessed whether geometric torsion could be effectively used. to predict spinal deformity with the aid of multiple linear regression. Anatomical landmarks were obtained from multi-view radiographic reconstruction and used to generate 3D model of the spine and rib cage of 28 patients. Fourier series best fitted to the vertebral centroids approximated the spinal shape. For each patient, spinal deformity indices were computed. Torsion was calculated and 20 derived parameters were recorded. Torsion inputs were used in a multiple linear regression model for prediction of key spinal indices. The primary clinical Cobb angle (mainly thoracic) was predicted well, with r=0.89 using all 20 inputs of torsion or r=0.83 using just two. Torsion was also well related to the orientation of plane of maximal deformity (r=0.87). Torsion was less accurate but still significant in predicting maximal vertebral axial rotation (r=0.77). This preliminary study showed promising results for the use of geometric torsion as an alternative 3D index of spinal deformity.