X-rays have long been used to diagnose and monitor scoliosis until patients achieve skeletal maturity. Epidemiologic studies have demonstrated that the effects of radiation from repeated x-rays during childhood leads to increased malignancy rates later in life. This has created an increasing need for alternative modalities to monitor children with scoliosis during their adolescent growth.
Surface topography is a modality that has been tried with limited success in the past, but may be a more reliable solution as new systems provide more powerful analysis of topography data. The aim of this article is to:
- Review various techniques being used to measure surface topography
- Discuss the advantages and disadvantages of these techniques
- Describe the advances that have been made in the use of surface topography
Scoliosis is a relatively common disorder, with curves greater than 20 degrees occurring in <1/1000 (approximately 2-4%) of children age 6 to 14.1 To date, the gold standard for identifying and monitoring scoliosis has been standing AP and lateral scoliosis films. But as research has progressed, we have become increasingly aware of the disadvantages of radiographs.
- They are only two-dimensional and thus do not take into account the three-dimensional nature of the curvature.
- There is significant intra- and inter-observer variation in Cobb angle measurements, with a recent article from He et al showing an intra-observer variance of 5° and an inter-observer variance of 6.5°.2
- There is an increasing awareness of the potential oncogenic effects of radiation exposure.
- Ronckers et al found that cancer mortality was 8% higher than expected in patients who had repeated radiographs for scoliosis, and there was a four times greater relative risk of breast cancer in female patients with spinal disorders.3
- Ronckers et al studied radiation exposure from 137,000 spine procedures and determined there to be an estimated cumulative radiation dose of 10.9 to the breast, 4.1 to the lung, 7.4 to the thyroid and 1.0 cGY to the bone marrow.
- Curran et al. found there to be a 10 times higher risk of breast cancer.4
- Both Ardran et al and Doody et al found a mortality ratio of 4.1 from breast cancer for a woman receiving 50 or more radiographs with a lag time of 30 years.5,6
Because of these disadvantages and the potential risk associated with radiographs, there have been numerous attempts to find other ways to identify and monitor scoliosis. Wong et al evaluated the accuracy of scoliometers as a screening tool, which were found to have a false positive rate of 58.3% when a 5° cut off was used.7
The Ortelius 800 is a device that combines a manual scoliometer with surface topography. The scoliometer has a position sensor that takes hundreds of data points as it is moved across the patient’s back, and then each spinous process is identified by manual palpation and marked with the position sensor. Unfortunately, in a study by Knott et al, this technique only predicted the correct radiographic Cobb angle 6.4% of the time and was within an acceptable range 55.3% of the time. They concluded that the variability was too great and that Orthoscan did not accurately predict an individual’s curve progression.8 The limiting factors seemed to be the length of time that it took to acquire a patient scan (~ 5 minutes) and the movement of the patient during that time. The ability to accurately find each spinous process by palpation was also a source of variability.
Despite early setbacks, surface topography has gained favor as new algorithms that are more robust and accurate have been developed. Moiré technology was first developed in the 1960-70s and reported on separately by Takasaki and Meadows et al in 1970. However, it was not until the 1980s and 1990s that researchers began to see its utility in assessing back surface topography, which could then be used to extrapolate the shape of the spinal curvature.
Moiré technology was the earliest form of surface topography. It is based on the distortion that occurs when a grid is projected onto a 3D object, and its distortion creates shadows and fringes from which 3D information can be extracted.9 Early analysis was incredibly burdensome, required hours to process, and was highly influenced by excess noise. However, Moiré technology became far more widely used as computers and computer algorithms for processing the data improved. Major advances were made with the development of 1) ISEF (infinite size symmetric exponential filter), whose filter segmented and smoothed the fringes so that it was easier to locate and analyze scoliosis, and 2) Mota’s Discrete Relaxation Operator, which generated 3D surfaces from the 2D contour lines.10 As Drerup and Hierholzer developed new technologies based on Raster Stereography in the 1980s, and technology based on structured light increased, Moiré technology was slowly phased out.
InSpeck uses four optical digitizers, a structured light projector that projects a pattern of black and white narrow stripes onto the patient’s trunk, and skin markers. Data acquisition by all four digitizers takes between 4 and 6 seconds and consists of five data sets from the four digitizers, four with fringes and the fifth without fringes, from each camera.
Pazos et al compared the clavicle position (routinely used for scoliosis films) and anatomic position using the InSpeck system and found that both provided reproducible and accurate results. They also found that the anatomic position was better for visualizing anterior thoracic areas.11 Seoud et al used the InSpeck system in conjunction with radiographs to create a 3D geometric model of the rib cage, placing markers on the trunk surface of 39 patients. The accuracy of their measurements was 1.1±0.9 mm over the entire trunk surface and a 1.4° surface rotation error.12
Fortin et al described the use of the InSpeck system in conjunction with a pressure mat to help in the development of spinal braces. This combination was used to develop the 3D model upon which the brace was based, rather than the traditional plaster cast. Then the pressure mat, with 192 sensors covering the patient’s entire trunk, was placed between the brace and the patient’s trunk to further allow the orthotist to adjust the brace.13 Labelle et al performed a randomized control study of 48 AIS patients undergoing bracing with Boston braces and found that the InSpeck group had a statistically significant improvement in curve correction compared with the control group, with an average in-brace correction of 12° in the InSpeck group versus 7° in the control group for thoracic curves, and 10° in the InSpeck group versus 6° in the control group for lumbar curves.
A commercial version of the Integrated Shape Imaging System (ISIS1) was developed between 1984 and 1988, with a scan time of 2 seconds and 10 minutes needed to analyze the photograph. ISIS2, based on the ISIS1 system, uses a combination of digital photography and the structured light of parallel fringes with a fringe frequency of 0.16 fringes/mm. It has an accuracy of +/- 1mm when providing a color 3D contour plot of the patient’s back in a coronal, transverse, and sagittal plane. Scan time is < 0.1 seconds.
Zubovic et al performed 520 scans on 242 patients and found good repeatability. They then compared the scans of 111 patients with their radiographs and found no statistically significant differences. However, they did not indicate which measures they were comparing between the radiographs and the scans.14
Berryman et al examined 168 patients, 28 on two occasions and 2 on three occasions. Patients were placed with their feet against positioning blocks directly below their ASIS, the abdomen resting against a crossbar and arms supported on rests. Stickers were placed on multiple landmarks on the patient’s spine, with the number of stickers increased in more severe scoliotic deformities. Interpolation was not required to obtain height data for every pixel, unlike ISIS1, Quantec, and Formetric. The correlate of the Cobb angle in Berryman et al's study was lateral asymmetry, which they found to have good correlation (R= 0.84), and to be within 10° of the Cobb angle in 80% of patients. The use of lateral asymmetry, they found, was limited in patients who were extremely obese or very muscular.15
Berryman et al also used ISIS2 to measure 60 patients’ rib humps, with two measurements at each session. ISIS2 was used to reconstruct a wire frame model and a contour plot. The researchers found a difference in rib hump height between pairs of measurements of -0.08 mm, with a 95% CI of -9.82 to 9.66 mm.16
They used ISIS2 to measure thoracic kyphosis as well, measuring 61 patients twice, with markers placed on the patients' backs, vertebral prominences, lumbar dimples, and spinous processes. In this study, a modified Cobb technique was used, measuring from T1/2 to the point of inflection in the curve. The average kyphosis angle of each photo was 33.8°, and the mean difference between the pairs of measurements was -0.02° with a 95% CI of -7.41° to7.38°.17
Carmen et al found a +/-10.6° 95% CI inter-observer variation and a +/-10.4° 95% CI intra-observer variation in their study of inter- and intra-observer variance in kyphosis measurements in radiographs.18
The Quantec system also uses raster stereography for 3D assessment through a single photographic image of a fringe pattern projected onto the patient’s back, which is then reconstructed into a surface representation. It has been used by McArdle since 1994 and is a highly portable system. Markers are placed over T1, T12, dimples of Venus, and occasionally additional spinous processes. The data acquisition process is extremely rapid, taking 250,000 data points with an accuracy of 0.25 mm in only 1/50th of a second, thus making the Quantec system less susceptible to patient motion.19,20
Klos et al followed 105 single curve subjects and 62 double curve subjects for a total of 543 clinical scans over a 3-year period. In addition to the markers, their patients stood on a foot post with their arms abducted, performing three trials per visit. Klos et al analyzed their data using the Functional Capacity System,21 comparing the within-visit accuracy as well as the between-visit accuracy of the Q angle versus the Cobb angle (the Q angle being the surface topography equivalent of the Cobb angle). Their intraday Q angle variation was <5°. When the between-visit increase was less than 5°, the Q angle varied more; when between-visit increase was more than 5°, the Cobb angle varied more. This indicates that the Q angle alone may not be sufficient to monitor the change in Cobb angle.22
McDonald et al used the Quantec system to analyze the effects of maximum sway on the upper curve, lower curve, thoracic curve, lumbar curve, vertical alignment, and pelvic tilt by having the patient stand entirely on one foot, then the other, and comparing that to their baseline of standing equally on both feet. The researchers found that the thoracic curve and pelvic tilt measurements were most profoundly affected and that the changes in all the measures were small except for pelvic tilt.23
McArdle et al looked at 57 patients with kyphoscoliosis over a 16-year period, with a minimum of three measurements of thoracic sagittal curve on five or more occasions prior to surgery. They placed six to eight markers on the spinous processes, which identified the "spine line" in the sagittal projection, and used a positioning frame. The average standard deviation was 3.8°.24
Formetric system has been utilized in several European countries. It uses Raster technology, which is then detected using one camera. It does not use external markers or a positioning frame. One set of measurements takes 6 seconds at a rate of 2 pictures per second, during which time it acquires 12 complete sets of measurements that can be averaged.
Formetric uses a very robust algorithm based on the analysis of hundreds of scoliosis x-rays and topography scans. Knott et al performed two reproducibility studies using Formetric 4D and found the scoliosis angle of the major curve (equivalent of the Cobb angle) to have an average standard deviation of +/- 3.2°. The first study of 12 patients with AIS had 30 data sets of 10 measurements, and the second study with a set of 14 patients (9 with small or no scoliosis and 5 with curves ranging from 15° to 40°) consisted of 30 data sets of 10 measurements each.25,26
Hackenberg et al used the Formetric system to examine the back surface in standing posture and forward bending posture. Their goal was to determine if axial rotation could be better assessed in these positions. They found a 3.2° increase in back surface rotation between the two postures, with a standard deviation of 6.1°, and a poor correlation between the axial rotation in standing and forward bending positions measured with both surface topography (r2 = 0.41) and scoliometer (r2 =0.35). They also found large differences between the rotation measured by surface topography and the scoliometer.27
Mohokum et al performed a reliability study with the Formetric system using 51 healthy volunteers whose exams consisted of three pictures each, which were analyzed by three investigators. They used a positioning board but did not use markers. For their measurement of the kyphotic angle, they found a Cronbach alpha intra-tester reliability value between 0.921 and 0.992, and an inter-tester reliability for kyphosis of 0.979. There was less intra-tester reliability when measuring the lordosis angle, ranging from 0.884 to 0.972, and an inter-tester reliability of 0.961. They also analyzed the data to determine the influence of BMI on their outcomes, separating the patients into two groups based on BMI above and below 24.99. BMI had no influence on reproducibility.28
Knott et al, in their reproducibility study analyzing 30 measurements of 14 patients, found increased variability in scoliosis angles at greater BMIs. Their scoliosis curve correlated with BMI (r= 0.65). Despite this, they still found that the patient with the highest BMI had scoliosis curve measurements that were only +/- 4.6°.29
Several techniques to measure surface topography are currently available. Although studies demonstrating each individual system’s accuracy and reproducibility are relatively limited, enough data are available to establish that the Cobb angle equivalents of these systems do approximate the Cobb angle fairly well, even if they are not exact representations of the Cobb angle. Thus, changes in the Cobb angle equivalents parallel changes in the Cobb angle to the point that it is becoming feasible to use surface topography to both screen and follow patients with scoliosis, with radiographs obtained only when there has been a change in their measures. At this point, surface topography has not eliminated the need for radiographs, as an x-ray is still necessary to evaluate the morphology of the spine. Evidence suggests, however, that surface topography has the potential to dramatically decrease the number of radiographs taken over a patient's lifespan.
Surface topography still has a number of limitations, as it has primarily been used in slimmer patients in whom the contours of the back are more pronounced, as well as in patients who are capable of standing. To date there is limited information on patients with a BMI over 29. Additionally, several studies are currently being conducted that examine the use of surface topography in populations of patients with neuromuscular disorders who would not be capable of standing or holding still for the exam.
A number of posturing devices may be of particular use to patients with neuromuscular disorders. Many believe that these devices result in more reproducible images (both radiographic and surface topography) by achieving approximately the same posture each time and by minimizing the effects of sway. The downside is that these devices may potentially force patients into unnatural positions, which may affect the ability to measure the severity of a very flexible curve. Furthermore, positioning devices might impair acquisition of the images in systems that use multiple devices, such as the InSpeck system.
Positioning devices can take on a variety of forms. In 1994, the SRS recommended the use of supports to position the ASIS with respect to the cassette for improved radiographic imaging.30 While these supports primarily affect pelvic orientation, they may also minimize sway. Given their location, they would be unlikely to affect the shape or magnitude of the patients curve. However, they would not address the full affects of sway, nor would they address hand position (ie, clavicle position version position of comfort) or give stability to a patient unable to hold up himself or herself.
Other devices are more extensive, such as the one described by Berryman et al, where they had patients place their feet against positioning blocks, rest their abdomen against a crossbar, and place their arms on rests.25 The affects of merely resting the arms on a support may potentially change shoulder position and thus the ability to accurately assess back surface.
Alternative positioning devices utilize the activation of the patient’s own muscles to achieve a more consistent stance, such as the Balanc Aid used by Dewi et al. This device consisted of a square board with a cylindrical disc at the center and bars on either side of the patient, which they could grab for balance. As a result of using the device, patients were placed in a position of forced balance, which had better reproducibility than standing on ground alone.31
Unfortunately, although these devices may potentially serve a role in the AIS population, many would be of little use in patients with neuromuscular disorders, who lack the muscle control to use the Balanc Aid and whose curves might be altered by other devices because their curves are so flexible. Yet this lack of muscular control means these patients are unable stand for any type of imaging study, forcing them to require a device that may cause us to underestimate the degree of their curve. More research is needed to determine the ideal patient positioning and whether that involves a positioning device. The answer is likely to vary between and potentially within the populations we study.
Another topic of debate is the use of surface markers; three of the four systems discussed above use them. Only Formetric does not, although they are available. The thought is that surface markers improve the accuracy of the 3D reconstruction by facilitating data acquisition. To date, no study has demonstrated this. Pazos et al, using the InSpeck system, found additional sources of error associated with the use of markers secondary to poor acquisition due to lighting changes and the orientation of the digitizers affecting the ability to record the reflective surfaces.32
Additionally, the question has been raised that as these devices have become better at analyzing the surface of the back, they may be more accurate than the human eye at truly assessing the exact position of these landmarks. At this point, there are few data to support or reject the use of surface markers, and it remains up to the investigator as to whether they are used in all patients or potentially only in special populations, such as obese patients or patient with neuromuscular disorders, where the landmarks may only be palpable or visible in certain positions.
Ultimately, surface topography poses numerous potential benefits over traditional radiographs. But more research and standardization between and within the different systems are needed to improve the consistency and accuracy of the images that are taken. This is true not only for surface topography, but also for traditional radiographs.
Thus, we need to be more vigilant about the way in which we image our patients and the frequency with which we do so, no matter what which technique we use.
- Stirling AJ, Howel D, Millner PA, Sadiq S, Sharples D, Dickson RA. Late-onset idiopathic scoliosis in children six to fourteen years old. J Bone Joint Surg 1996;78-A:1330--6.
- He JW, Yan ZH, Liu J, Yu ZK, Wang XY, Bai GH, Ye XJ, Zhang X. Accuracy and repeatability of a new method for measuring scoliosis curvature. Spine 2009 Apr 20; 34(9): E323-9
- Ronckers C.M., Land C.E., Miller J.S., Stovall M., Lonstein J.E., Doody M.M.Cancer mortality among women frequently exposed to radiographic examinations for spinal disorders. Radiat Res. 2010 Jul;174(1):83-90.
- Curran P, Groves D. Assessing spinal deformities, Image Processing, September/October 14-16 (1990).
- Ardran GM, Coates R, Dickson RA, et al. Assessment of scoliosis in children: low dose radiographic technique. Br J Radiol 1980;53:146--7.
- Doody MM, Lonstein JE, Stoval MS, Luckyanov N, Land CF. Breast cancer following diagnostic x-rays among women with scoliosis. Proceedings of the 33rd Annual Meeting of the Scoliosis Research Society, New York, USA, 1998:104.
- Hee-Kit Wong, Ponnampalam Balasubramaniam, Uma Rajan, and Seo-Yi Chng. Direct Spinal Curvature Digitization in Scoliosis Screening - A comparative Study with Moire Contourgraphy. Journal of Spinal Disoders. 1997:10 (3): 185-192
- Knott et al. Electromagnetic Topographical Technique of Curve Evaluation for Adolescent Idiopathic Scoliosis. Spine. 2006: 31(24): E911--E915
- M. Batouche, M, Benlarmi, R and Kholladi, MK. A computer Vision System for Diagnosing Scoliosis Using Moire Images. Compu Biol Med Vol. 26, No. 4, pp. 339-353. 1996
- Batouche M, Benlamri R, Kholladi MK. A Computer Vision System for Diagnosing Scoliosis using Moire Images. Comput. Biol. Med. 1996:26 (4): 339-353
- Pazos et al. Reliability of trunk shape measurements based on 3-D surface reconstructions. Eur Spine J. 2007: 16:1882--1891
- Seoud et al. Novel Method for 3-D Reconstruction of Scoliotic Ribs from Radiographs. IEEE Transactions On Biomedical Engineering, 2011; 58(5): 1135-46
- Fortin D et al. A 3D Visualization Tool for the Design and Customization of Spinal Braces. Computerized Medical Imaging and Graphics 31. 2007: 612-624
- Zubovic et al. New Method of Scoliosis Deformity Assessment: ISIS2 System. Research into Spinal Deformities 6. IOS Press, 2008: 157-160.
- Berryman F, Pynsent P, Fairbank J, Disney S. A new System for Measuring Three-Dimensional Back Shape in Scoliosis. Eur Spine J (2008) 17:663--672
- Berryman F, Pysnsent P, Fairbank J. Measuring the Rib Hump in Scoliosis with ISIS2. Research into Spinal Deformities 6. IOS Press, 2008: 65-67
- Berryman F, Pynsent P, Fairbank J. Thoracic Kyphosis Angle Measurements with ISIS2. Research into Spinal Deformities 6. IOS Press, 2008: 68-71.
- Carman DL, Browne RH, Birch JG. Measurement of Scoliosis and Kyphosis Radiographs. JBJS. 72-A(3), pp 328-333. March 1990
- User Guide. Quantec Image Processing and Spinal Measurement System. Lancashire, UK: Quantec Image Processing; 1997:1--19.
- Oxborrow. Assessing the child with scoliosis: the role of surface topography. Arch Dis Child 2000; 83:453--455
- Liu XC, Thometz JG, Lyon RM, et al. Functional classification of patients with idiopathic scoliosis assessed by the Quantec system: a discriminant functional analysis to determine patient curve magnitude. Spine 2001;26: 1274--9.
- Klos et al. Reliability of a Functional Classification System in the Monitoring of Patients With Idiopathic Scoliosis. SPINE Volume 32, Number 15, pp 1662--1666
- Macdonald AM, Griffiths CJ, MacArdle FJ, Gibson MJ. The effect of posture on Quantec Measurements. Research into Spinal Deformities 4. IOS Press 2002. 190-193
- McArdle et al. Monitoring The Thoracic Sagittal Curvature in Kyphoscoliosis with Surface Topography: a Trend Analysis of 57 Patients. Research into Spinal Deformities 4. IOS Press, 2002
- Knott et al. Evaluating the Reporducibility of the Frometric 4D Measurements for Scoliosis. Eur Spine J 2010:19(supp 3): S241-242
- Knott P, Mardjetko S, Frerich JM, Hertzler K. Comparison of Radiographic and Surface Topography Measurements in Adolescent Scoliosis. Scoliosis Research Society IMAST Meeting, July 2011, Copenhagen, Denmark. (poster #523)
- Hackenberg L et al. Rasterstereographic Analysis of Axial Back Surface Rotation in Standing versus Forward Bending Posture in Idiopathic Scoliosis. Eur Spine Journal 2006;15: 1144--1149
- Mohokum et al. Reproducibility of Rasterstereography for Kyphotic and Lordotic Angles, Trunk Length, and Trunk Inclination: A Reliability Study. SPINE 2010: 35(14): 1353--1358
- Knott et al. The Influence of Body Mass Index (BMI) on the Reproducibility of Surface Topography Measurements. International Conference on Conservative Management of Spinal deformities SOSORT meeting 2011. 05/01/2011
- Stokes, I. A. F. Three-dimensional terminology of spinal deformity: a report presented to the Scoliosis Research Society by the Scoliosis Research Society Working Group on 3-D terminology of spinal deformity. Spine 19:236--248, 1994.
- Dewi, D E I. et al. Reproducibility of Standing Posture for X-Ray Radiography: A Feasibility Study of the BalancAid with Healthy Young Subjects. Annals of Biomedical Engineering, 2010. 38(10): 3237--3245
- Pazos, V. et al. Accuracy assessment of human trunk surface 3D reconstructions from an optical digitising system. Med. Biol. Eng. Comput., 2005. 43: 11-15