The reliability and validity of body vision system for assessing posture

Introduction: Body vision is a novel method which examines postural indices through photogrammetric essentials. Nevertheless, its reliability and validity has not been appraised till now. We aimed to evaluate the reliability and validity of body vision system for posture assessment. Methods: This was a cross-sectional study in which two examiners evaluated photographs of 71 subjects using body vision system twice with a two-week interval. The body vision system involves a grid wall and a camera fixed in front of the grid wall at about 390 cm distances. Three standing photographs (anterior, right lateral, and posterior view) were captured for participants. Results: The results for inter-rater reliability analysis showed that most of the parameters (74%) had excellent 95% confidence interval (CI), 10 % had good to excellent 95% CI, 13% had moderate to good 95% CI, and 1% had poor to moderate 95% CI. The results for intra-rater reliability analysis showed that 70%-72% of the parameters had excellent 95% CI, 6%-9% had good to excellent 95% CI, 12%-13% had moderate to good 95% CI, and 9% had poor to moderate 95% CI. The comparison between known distances and angles on grid wall and those obtained from photogrammetric measurements showed that there was no statistically significant difference (P > 0.05). Also, the regression analysis showed that there was a significant and positive relationship between them (R2 = 1, P < 0.05). Conclusion: The results of this study showed that body vision system is a valid and reliable tool for measuring postural parameters. Article History: Received: 8 June. 2020 Accepted: 6 Sep. 2020 e-Published: 16 Jan. 2021


Introduction
Posture is the alignment between different parts of body. A good posture is where the musculoskeletal structures bear least stress due to balance between different components. 1,2 Deviations from proper postural alignment indicate that the balance has been disturbed and some parts are at risk of injury because of excess load. 3 Also, postural analysis has shown that disorders like spondylolisthesis, 4 adolescent idiopathic scoliosis, 5 osteoporosis, pain, 6 and nasal patency 7 might affect the subject's posture which causes additional problems. 8 Posture assessment can be done with X-ray exam which might expose radiation or noninvasive methods like visual inspection, goniometry, photogrammetry and 3D laser scan. 9,10 Photogrammetry is a simple and safe method for analyzing posture in a quantitative manner. There are different types of software programs like postural assessment software (SAPO) and Surgimap which helps to measure postural parameters faster and easier. The reliability of measurements using most of these software packages has been investigated previously. Body vision is a new system which analyzes postural parameters using photogrammetric basics. However, the reliability and validity of the measurements using body vision system has not been evaluated till now.
The purpose of this study was to determine the interrater and intra-rater reliability and validity of postural analysis using body vision system.

Subjects
Photographs were chosen from databases of posture assessment clinic of Imam Reza hospital. The protocol was reviewed and approved by Physical Medicine and Rehabilitation Research Center. The subjects wearing clothes more than underwear were excluded and 71 subjects remained for analyses. The objectives and the process of the study were explained and written informed consent was completed for each subject. The subjects were able to retract the projects during the study, without providing any reason. All subjects had three standing photographs (anterior, right lateral, and posterior view), looking forward and arms at sides. All photographs had been captured by posture clinic assistant who had marked some anatomical points with 1.5 * 1.5 cm sticking spheres. The percent of identified (and visible) markers in each view was as follows: Anterior view: Greater trochanter 22.5%, ASIS 95.7%, suprasternal notch 15%, xiphoid 6%, acromion 6%.

Body vision
The body vision system consists of an 80*195 cm grid wall (with 10 cm sectors) and a camera (CANON Zoom Lens EF-S 18-55 mm 1:3:5-5.6 IS II 58 mm) which has been fixed in front of the grid wall at about 390 cm distance (on about 100 cm height). The camera is connected to the computer and examiner can take photos, restore them and analysis different angle and distance parameters of posture using the body vision software (Table 1). In this version of body vision software, the examiner has to identify different landmarks. The device has been developed in Tabriz, Iran  by Tocea Tadbir Tavan Teb Company (Rehabsoon Co).

Procedures
Two physical therapists who were familiar with the use of the software were randomly chosen from students of Physical Medicine and Rehabilitation Department of Tabriz University of Medical Sciences. Evaluation of photographs was done using body vision software. The examiner defined the plumb line and then other landmarks orderly as shown in Figure 1. Each examiner evaluated the photographs twice with a two-week interval. The third investigator exported reports of each test in an excel form, stored them in other file and cleaned from software database in order to not allowing the examiners to see previous analysis.
To evaluate the validity of measurement, we used distances and angles on grid wall as the reference value and asked the examiners to mark and measure them using the body vision software. In order to perform blinding, the examiners were not allowed to see the results of measurements and the third investigator gathered the data.

Statistical analysis
The intra-rater reliability of measurements was evaluated by intraclass correlation test (ICC two ways mixed, average, absolute agreement). The inter-rater reliability of measurements was evaluated by ICC test (two ways random, average, absolute agreement). It has been suggested to define poor, moderate, good and excellent results if the ICC values equals to less than 0.5, 0.5 to 0.75, 0.75 to 0.9 and greater than 0.9, respectively. 11 Paired t test and regression analysis were used to evaluate the validity of measurements.
All analysis was done using SPSS software 16 and P value less than 0.05 was considered significant.

Results
The mean age of subjects was 21.13 years old (ranging from 4 to 67 year) and 62% of them were male.
The results for inter-rater reliability analysis showed that most of the parameters (74%) had excellent 95% confidence interval (CI), 10 % had good to excellent 95% CI, 13% had moderate to good 95% CI, and 1% had poor to moderate 95% CI ( Table 2).
The comparison between known distances and angles on grid wall and those obtained from photogrammetric measurements showed that there was no statistically significant difference (P > 0.05). Also the regression analysis showed that there was a significant and positive relationship between them (R 2 = 1, P < 0.05) ( Table 3).

Discussion
According to the results obtained in this study, most parameters had good to excellent inter-rater and intrarater reliability. The low levels of reliability in some parameters might be attributed to the low resolution of the photographs or hiding of the markers due to relaxed position of the subject, for example in the case of tibial tuberosity or greater trochanter landmarks, respectively. So, it is necessary to mark those landmarks which might not be clear in photo and all landmarks should be visible; otherwise, the analysis will not be reliable. The results of other studies using different software were the same. Souza et al. have shown that postural assessment software (SAPO) is a reliable tool for postural analysis when bony landmarks were identified before photogrammetry. 12 Santos et al and Hazar et al have found same results in children and adolescents, respectively. 13,14 Helmya et al have shown that the Surgimap spine software is a reliable tool for posture assessment when placing markers on certain anatomical landmarks. 15 Muniandy et al concluded that web plot digitizer software was a reliable tool to assess forward  ASIS to greater trochanter distance D4 Distance between the ASIS and greater trochanter on same side.

Q angle A7
The angle formed between the line connecting ASIS to patellar center and the line connecting tibial tuberosity to the patellar center.
Lateral thigh leg angle A8 The angle formed between the line connecting greater trochanter to lateral mid knee joint line and the line connecting lateral mid knee joint line to the lateral malleolus in anterior view.
Femoral-tibial angle (anatomic knee angle) A9 The angle between distal part of thigh and proximal part of leg Intercondylar distance D5 Horizontal distance between right and left medial femoral condyles.
Tibial tuberosity tilt angle A10 The angle formed between the line connecting right and left tibial tuberosity with the horizontal line.
Femoral segment length D6 Distance between greater Trochanter and lateral mid knee joint line on each side.
Tibial segment length D7 Distance between lateral mid knee joint line and lateral malleolus on each side.
Lateral malleolus tilt angle A11 The angle formed between the line connecting right and left lateral malleolus with the horizontal line.
Intermalleolar distance D8 Horizontal distance between right and left medial malleolus.
True leg length D9 Distance between the ASIS and medial malleolus on same side.
True discrepancy D10 Difference between right and left true leg length.
Apparent leg length D11 Distance between the umbilicus and medial malleolus on same side. The angle formed by the straight lines between the C7 spinous process and the spinous process of T12 and that intersects the horizontal line between T7 (between T6 and T8 ) and the true vertical line Lateral spinal angle A18 The angle formed between the line connecting C7 spinous process to T12 and the line connecting T12 to the greater trochanter.

Lumbar lordosis angle A19
The angle formed by the straight lines between the T12 spinous process and the spinous process of L5 (between L4 & S2) and that intersects the horizontal line between L3 (between L2 & L4 ) and the true vertical line Lateral lumbar angle A20 The angle formed between the line connecting T12 spinous process to ASIS and the line connecting ASIS to the greater trochanter.
Pelvic tilt angle A21 The angle formed between the line connecting ASIS and PSIS and a horizontal line through PSIS.
Trunk sway angle A22 The angle formed between the line connecting grater trochanter and acromion and a vertical line through grater trochanter.
J Res Clin Med, 2020, 8: 3 4 head posture when using marker placement. 16 Szucs and Brown had found same results about mobile application for posture analysis. 17 Salahzadeh et al showed that the level of ICC to assess the reliability of photogrammetry was ranged from 0.75 to 0.94 when examining forward head and the intra subject reliability was 0.84 to 0.89. 18 Ferreira et al found that the SAPO software would be a reliable tool for posture analysis when marking the anatomical points and the low levels of reliability might be due to primary calibration of the photos. 19 We did not meet this issue because the calibration had been done once for all and before photogrammetry, so the raters did not calibrate each photo again.
The low values of reliability analysis in our study are most probably due to unclear location of anatomical points in some cases. Hébert-Losier and Abd Rahman did not use surface marker with the rational that it would be more practical when assessing great number of subjects and they found the inter-rater reliability of posture pro 8 software would be fair (ICC values of 0.40 to 0.75) to measure most parameters of posture analysis. 20 So, it is necessary to find out the least landmarks needed to be marked in order to obtain reliable results. According to data presented in our study, the authors suggest that it is necessary to mark C7, T7, T12, L1, L3, and L5 spinous process, bilateral ASIS, acromion, PSIS, greater trochanter, inferior angle of scapula, tibial tuberosity, mid-popliteal, lateral malleolus, mid-distal leg, Achilles tendon and calcaneus to have reliable results for all postural parameters. These results are consistent with the review of do Rosário. 21 The other factor that is said to affect the results is the feet position. 4 However, Antoniolli et al found that the position of feet did not interfere with the results of the posture assessment.
Another aspect of using a measurement method is the validity which explain how results are accurate. 22 The validity of any measurement tool could be assessed when comparing the results with those obtained using standard methods.
Walicka-Cupryś et al have shown that the results of photogrammetry and inclinometer are comparable to measure thoracic kyphosis and lumbar lordosis. 23 Ludwig et al had compared the photometric method for posture index measurement with visual assessment to obtain the validity of photometric method. 24 Marques et al showed that goniometry and photogrammetry measurements of hip abduction and flexion have comparable results. 25 However, the different landmarks in each test 26,27 or different unit calculation 28,29 might cause differences between results of various methods. As Marchetti et al had shown there might be ± 5.9 and 6.9-degree difference when comparing spinous process marker for measuring Lateral trunk angle A23 The angle formed between the line connecting grater trochanter and acromion and a line connecting grater trochanter and lateral malleolus.

Lateral hip angle A24
The angle formed between the line connecting greater trochanter to lateral mid knee joint line and the line connecting greater trochanter to ASIS.
Hip-plumb line distance D15 The horizontal distance of grater trochanter from plumb line Lateral knee angle A25 The angle formed between the line connecting greater trochanter to lateral mid knee joint line and the line connecting lateral mid knee joint line to the lateral malleolus.
Knee -plumb line distance D16 Horizontal distance from lateral mid knee joint to the plumb line.
Lateral ankle angle A26 The angle formed between the line connecting lateral mid knee joint line to the lateral malleolus and the horizontal line through the lateral malleolus.
Body sway angle A27 The angle formed between the line connecting lateral malleolus and acromion and a vertical line through lateral malleolus.
Inferior angle -midline distance D17 The horizontal distance from inferior angle of scapula to the mid line of thoracic spine.
Inferior angle -T2 distance D18 The distance between inferior angle of scapula and T2 spinous process.

Scapular tilt angle A28
The angle formed between the line connecting inferior angles of scapula of both sides with the horizontal line.
Thoracic spine alignment angle A29 Elbow -trunk distance D19 Horizontal distance between medial elbow and trunk Elbow trunk difference D20 Difference between bilateral elbow-trunk distances  cobb angle with the standard method. 30 Wilczyński et al have compared photogrammetry and radiology for spinal curvature diagnosis and they have found that there is a significant but low correlation between them. 31 Döhnert and Tomasi found that photogrammetry would not be useful for screening and diagnoses of mild scoliosis. 32 Each parameter of posture assessment could be evaluated with its standard method of diagnosis.
Although there are different gold standard methods like X-ray, goniometry, inclinometer, etc to diagnose postural abnormalities, the aim of this study was not to compare all these with photogrammetry but it was to investigate how measurements are accurate. Therefore, we defined fixed points on the grid wall with known distances and angles and compared the value of each parameter with those obtained by photogrammetry. Schwertner et al had used similar way to verify the validity of SPGAP (posture evaluation rotating platform system). 33 They compared the values obtained through SPGAP system measurement of an object with its known dimensions. Ruivo et al used the metal pieces and goniometer as the reference values for assessing the validity of SAPO software. 34 Ferreira et al also had used an object with known distances and angles to investigate the validity of SAPO software. 19 They found that the mean difference for angle and distance measurement were 0.11 degree and 1.8 mm, respectively. We found that the mean difference for angle and distance measurement were 0.02 degree and 0.03 mm, respectively. There was no statistically significant difference (P > 0.05) and the correlation between measurements was almost complete (R 2 = 1). The magnification error of measurements was 0.02 which is inevitable in two dimensional analyses. So, it could be said that the body vision is a reliable and valid tool for analysis of postural parameters.
Standing posture is not static and may change during time. 35 Sacco et al found that photogrammetry is not suitable for follow up purposes because the intra-subject reliability of rear foot angle measurement is low when analyzing the subjects with one-week interval. 36 In this study, we did not examine intra-subject reliability which is necessary to be evaluated in future studies. The experience of examiner whom marks the anatomical points, as well as the computer experience of examiners might affect the results and have to be investigate. 19,37 Conclusion It could be claimed that the use of body vision system is a reliable and valid tool for measuring postural parameters. Although the measures are reliable and valid, the utility of these analyses should be investigated in follow up studies.

What is current knowledge?
• Body vision is a novel method which examines postural indices through photogrammetric essentials. Nevertheless, its reliability and validity has not been appraised till now.

What is new here?
• TThe body vision system is a valid and reliable tool for measuring postural parameters.