Radiation-free Assessment of Scoliosis: A Reliability and Validity Study for Ultrasound Angles

Abstract

Study design

A prospective study.

Objectives

This study aimed to evaluate the reliability and validity of ultrasound angles, and to explore the impact of body mass index (BMI) and scoliosis status on the accuracy of ultrasound measurements.

Methods

A total of 94 subjects were enrolled in this study and subsequently underwent both X-ray and ultrasound examinations. Two orthopedic surgeons independently measured the radiographic Cobb angle and the ultrasound angle to evaluate the intra- and interobserver reliability of ultrasound angles. Based on apical vertebra location, BMI, and scoliosis status, subjects were stratified into clinically relevant subgroups. Linear regression analyses and Bland-Altman plots were employed to assess the accuracy of the ultrasound measurements.

Results

Ultrasound angles demonstrated excellent intra- and interobserver reliability (intraclass correlation coefficients, ICCs ranging from 0.939 to 0.99), and manual ultrasound angle showed better correlation with the radiographic Cobb angle than the automatic ultrasound angle. The lowest R² of 0.425 was observed for the automatic ultrasound angle in the overweight group, while manual ultrasound angle maintained comparable correlation with the Cobb angle across different BMI subgroups. Furthermore, a higher correlation was found between ultrasound angles and the Cobb angle in the scoliosis group compared to the non-scoliosis group.

Conclusions

This study showed that ultrasound angles were reliable and valid. Manual measurements showed superior performance and should be prioritized. When using ultrasound angles to assess spinal curvature severity, the influence of BMI and scoliosis status must be considered.

Keywords

scoliosis automatic manual ultrasound angle reliability validity body mass index

Introduction

Scoliosis is defined as a lateral curvature of the spine, diagnosed by the gold standard of a Cobb angle ≥10° on anteroposterior radiographs.^1,2 Regardless of whether observation, bracing, or surgical treatment is selected, regular X-ray evaluation is necessary.³ Luan et al demonstrated that due to cumulative radiation exposure from full-spine radiographs, adolescents with scoliosis exhibited a cancer incidence and mortality rate 1.46 times and 1.5 times higher than those in the control group, respectively.⁴

Due to its radiation-free, non-invasive, real-time, and convenient nature, ultrasound technology is now increasingly applied in spinal imaging.^5-9 Volume projection imaging has been developed to reconstruct 3-D ultrasound data into coronal views of the spine.¹⁰ Automatic algorithms, such as the gradient vector flow snake model, have been proposed to detect spinous process landmarks on transverse images, enabling measurement of the spinous process angle. 1¹¹ Numerous studies have demonstrated that the ultrasound angles exhibit good reliability (ICCs ranging from 0.63 to 0.99) and a strong linear correlation with the radiographic Cobb angle (R² ranging from 0.61 to 0.99).^5,7-9,11-15 Kwan et al incorporated ultrasonography into the scoliosis screening and confirmed its potential as a radiation-free alternative to X-ray.¹⁶ However, the accuracy of ultrasound-derived spinal curvature measurements can be compromised by patient-specific factors, including body mass index (BMI), vertebral rotation and thoracic kyphosis.^8,15

Thus, this study aimed to investigate the reliability and validity of both automatic and manual ultrasound angles in assessing spinal curvature, and to explore the influence of BMI and the scoliosis status on the accuracy of ultrasound angles in predicting the Cobb angle.

Methods

Study Population

Children and adolescents who underwent scoliosis screening at Peking University People’s Hospital from April 2024 to October 2025 were recruited consecutively in this study. Inclusion criteria were: (1) age younger than 18 years; (2) agreement to undergo both standing whole-spine radiographs and ultrasound scan on the same day. Exclusion criteria were: (1) Presence of spinal pathology other than scoliosis; (2) Previous spinal surgery; (3) Allergy to ultrasound gel.

Image Acquisition

The study design is illustrated schematically in Figure 1. Radiographic and spinal ultrasound images were both acquired with the patient in a weight-bearing standing position. However, during ultrasound scanning, the probe must be placed in close contact with the back skin with a certain amount of pressure. To prevent the patient from leaning forward and thereby altering the sagittal or coronal spinal alignment, the patient was instructed to gently grasp the fixed handrails located on both sides of the machine to maintain body balance and a natural weight-bearing spinal posture. All subjects underwent two consecutive spinal ultrasound scans (Scan 1 by Observer 1 and Scan 2 by Observer 2), using a three-dimensional ultrasound imaging system (SCN802, Telefield Medical Imaging Ltd., Hong Kong). During the ultrasound acquisition, the operator first applied coupling gel to the subject’s back, adjusted the scanning range according to the subject’s height, and then scanned along the spine from L5 to T1.

Figure 1.

Schematic of the study design

Scoliosis Measurements

Spinal curvature was measured using the Cobb method on X-ray and the transverse process method on ultrasound images, with only major curvature analyzed in both modalities. The measurements were performed by two orthopedic surgeons (observer1, O1; observer2, O2) with 4 and 6years of clinical experience, respectively. Both observers were required to independently measure ultrasound angles from 60 scoliosis patients until they felt confident in identifying bony landmarks on ultrasound images. Additionally, they were blinded to all patient clinical information and to each other. Cobb angle was measured independently by two observers on the same radiograph and their results were averaged. For each image, the system automatically identified skeletal landmarks to derive an automatic ultrasound angle. Furthermore, each of the two observers performed two repeated measurements on both Scan 1 and Scan 2. The average of these eight measurements was calculated as the manual ultrasound angle. Figures 2 and 3 showed representative cases of successful and failed automatic measurements, respectively, as compared with radiographic Cobb angles.

Figure 2.

The coronal Cobb angle, measured on radiographs (Left), was compared with automatic ultrasound angle (right). Both images demonstrated a right-convex thoracic curve with the upper and lower end vertebrae at T5 and L1, respectively

Figure 3.

A 14-year-old female with adolescent idiopathic scoliosis, BMI 17.5, upper end vertebra T6 and lower end vertebra T11. The radiographic Cobb angle was 23°, and the ultrasound angle was 15°. This discrepancy can be attributed to two main factors: 1) scapular-spinous process gap, which degraded probe coupling and transverse process visibility; 2) the inherent difference between anterior-based radiographic Cobb angles and posterior-based ultrasound angles, as the posterior column tends to exhibit less displacement in scoliosis

Subgroup Stratification

Subjects were stratified into the following clinically relevant subgroups: (1) apical vertebral location — thoracic (apical vertebra located between T2 and the T11–T12 intervertebral disc) and (thoraco)lumbar (apical vertebra located between T12 and L1, or between the L1–L2 intervertebral disc and L4); (2) body mass index (BMI) — underweight (BMI < 18.5kg/m²), normal-weight (18.5kg/m² ≤ BMI <24kg/m²) and overweight (BMI ≥ 24 kg/m²); (3) Scoliosis status — non-scoliosis and scoliosis (Cobb angle ≥10°).

Statistical Analysis

All statistical analyses were conducted with SPSS (Version 24.0; IBM, Chicago, IL, USA). Descriptive data were presented as mean ± standard deviation. Intra- and interobserver reliability were evaluated using intraclass correlation coefficients (ICCs), which were calculated using the 2-way random effects model with absolute agreement and interpreted as follows: 0.80–1.00 indicating very reliable, 0.60–0.79 indicating moderately reliable, and ≤0.60 indicating questionably reliable. Validity was assessed through simple linear regression; correlation coefficients were classified as 0.25–0.50 (poor), 0.50–0.75 (moderate to good), and 0.75–1.00 (very good to excellent). Agreement between ultrasound angles and Cobb angle was assessed using Bland-Altman analysis with 95% confidence intervals for bias and limits of agreement. To measure the differences in agreement for the two different ultrasound measurements, the mean absolute differences (MADs) between the Cobb angle and the ultrasound angles were calculated. The degree of measurement error for the ultrasound angles was expressed using the standard error of measurement (SEM) and the minimal detectable change (MDC). The MDC, which represents the threshold for true change at a 95% confidence level, was computed as MDC=1.96×√2×SEM. Values smaller than the MDC are considered to lie within the measurement error. The significance level was set at 0.05 for all analyses.

Results

Demographic and clinical characteristics were presented in Table 1. The mean (± standard deviation) age, height, weight, and BMI of the 94 subjects were 12.8 ± 3.6 years, 158.9 ± 15.1 cm, 47.9 ± 15.7 kg, and 18.5 ± 3.9 kg/m², respectively. The radiographic Cobb angle, automatic and manual ultrasound angle were, on average, 14.1±9.1°, 13±7.2°, and 13.2±8°.

Table 1.

Demographic and Clinical Characteristics of Study Population

	Total	Thoracic	(Thoraco)Lumbar
Number	94	67	27
Gender (male: female)		32:35	8:19
Age (year)	12.8±3.6	13±3.7	12.3±3.3
Height (cm)	158.9±15.1	160±15.7	156.1±13.6
Weight (kg)	47.9±15.7	49.2±16.7	44.8±12.6
BMI (kg/㎡)	18.5±3.9	18.7±4.1	17.9±3.2
Underweight		37	18
Normal-weight		8	6
Overweight		22	3
Radiographic Cobb angle (°)	14.1±9.1	13.3±9.1	16.2±9.2
Non-scoliosis		31	8
Scoliosis		36	19
Automatic ultrasound angle (°)	13±7.2	12.5±7.4	14.3±6.9
Manual ultrasound angle (°)	13.2±8	12.5±8	15±7.9

Reliability of Ultrasound Assessment

Based on the same scans, the intra- and interobserver reliabilities (ICCs) for the manual ultrasound angle were 0.99 (95% confidence interval: 0.984–0.994) and 0.97 (0.955–0.98) (Table 2). The automatic ultrasound angle yielded an ICC of 1.00 on the same scans, since the automatic method produced identical angle values. For different scans, the ICCs for the automatic ultrasound angle were 0.939 (0.905–0.956), while the manual ultrasound angle achieved intra- and interobserver ICCs of 0.969 (0.954–0.98) and 0.96 (0.941–0.973), respectively (Table 2).

Table 2.

Intra- and Interoperator Reliability of Ultrasound Assessments

	Same scan		Different scan
	Intra(ICC)	Inter(ICC)	Intra(ICC)	Inter(ICC)
Automatic ultrasound angle	1.000	/	/	0.939
Automatic ultrasound angle	1.000	/	/	(0.905-0.956)
Manual ultrasound angle	0.99^a	0.97^b	0.969^c	0.96^d
Manual ultrasound angle	(0.984-0.994)	(0.955-0.98)	(0.954-0.98)	(0.941-0.973)

^aCalculated with O1 first and second measurements for scan1.

^bCalculated with O1 first measurements and O2 first measurements for scan1.

^cCalculated with O1 first measurements for scan1 and O1 first measurements for scan2.

^dCalculated with O1 first measurements for scan1and O2 first measurements for scan2.

The MDC was calculated for both automatic and manual ultrasound angle measurements. For the automatic ultrasound angle, the ICC was 0.939, resulting in an SEM of 1.78° and an MDC of 4.93°. For the manual ultrasound angle, ICC values ranged from 0.96 to 0.99, corresponding to an SEM range of 0.8°–1.6° and an MDC range of 2.22°–4.43° (Table 3).

Table 3.

Minimal Detectable Change for Automatic and Manual Ultrasound Angle Measurements

	ICC	SD	SEM	MDC
Automatic ultrasound angle	0.939	7.2	1.78	4.93
Manual ultrasound angle	0.96-0.99	8	0.8-1.6	2.22-4.43

ICC, intraclass correlation coefficient; SD, standard deviation; SEM, standard error of measurement; MDC, minimal detectable change.

Validity of Ultrasound Assessment

Simple linear regression indicated significant correlations between the Cobb angle and the ultrasound angles regardless of apical vertebral location (Figure 4). The manual ultrasound angle demonstrated better correlation with the Cobb angle than the automatic ultrasound angle, both in thoracic (R² = 0.985 vs. R² = 0.809) and (thoraco)lumbar groups (R² = 0.977 vs. R² = 0.645).

Figure 4.

Equations between the Cobb angle (y) and the automatic (A) and manual (B) ultrasound angles (x) were created. The scatter plot, linear regression and equation: Cobb angle = regression coefficient × ultrasound angle + constant, were shown

Correspondingly, the Bland-Altman plots demonstrated a good agreement between the ultrasound angles and the Cobb angle (Figure 5). The bias between the automatic ultrasound angle and the Cobb angle was 1.15° (95% CI: 0.22° to 2.07°), with 95% LoA from -7.69° (-9.29° to -6.09°) to 9.99° (8.39° to 11.59°). For the manual ultrasound angle, the bias against the Cobb angle was 0.96° (0.63° to 1.29°), with 95% LoA from -2.17° (-2.74° to -1.60°) to 4.09° (3.52° to 4.66°). To compare the measurements in terms of validity, the MAD was measured between ultrasound angles and the Cobb angle. A significant difference between the two ultrasound angles was found (MAD: automatic ultrasound angle 3°±3.5°, manual ultrasound angle 1.2°±1.4°, p<0.001).

Figure 5.

Bland-Altman plots showed the differences between the ultrasound angles and Cobb angle. Limits of agreement between the automatic ultrasound angle and the Cobb angle were –7.69° to 9.99°, and between the manual ultrasound angle and the Cobb angle were –2.17° to 4.09°

Factors Influencing Ultrasound Assessment

The agreement between ultrasound angles and Cobb angle across different BMI subgroups were shown in Figure 6. The lowest R² value of 0.425 and the highest R² value of 0.991 were observed for the automatic and manual ultrasound angles, respectively, within the overweight group.

Figure 6.

Correlation and regression analysis between ultrasound angles and Cobb angle stratified by BMI. (A, B) Underweight: BMI < 18.5kg/m²; (C, D) normal-weight: 18.5kg/m² ≤ BMI <24kg/m²; (E, F) overweight: BMI ≥ 24 kg/m²

Figure 7 presented the results of the subgroup analysis based on the scoliosis status. The scoliosis group demonstrated stronger correlations than non-scoliosis group for both the automatic (R² = 0.623 vs. R² = 0.38) and manual ultrasound angle (R² = 0.974 vs. R² = 0.871).

Figure 7.

Correlation and regression analysis between ultrasound angles and Cobb angle stratified by scoliosis status. The red and green colors correspond to the non-scoliosis and scoliosis groups, respectively

Discussion

The cumulative radiation exposure associated with repeated radiographs during scoliosis management raises concerns about malignancy risk.^4,17 In contrast, ultrasound imaging offers a promising, radiation-free alternative. This study investigated the reliability and validity of ultrasound angles, and further investigated how these measurements vary across different BMI subgroups and between populations with and without scoliosis.

In this study, both automatic and manual ultrasound measurements were based on the transverse processes and were compared with the radiographic Cobb angle, respectively. The ultrasound angles demonstrated strong linear correlations with the radiographic Cobb angle (R² > 0.6), although the ultrasound measurements consistently yielded lower values than the Cobb angle. Brink et al suggested that this discrepancy occurred because ultrasound measurements were derived from structures situated more posteriorly than the vertebral body, which served as the reference for the Cobb angle, resulting in a different projection of the complex three-dimensional deformity.¹² Furthermore, manual ultrasound angle showed better correlation with the Cobb angle compared to automatic ultrasound angle, as reported by Brink et al.¹² However, previous studies indicated that ultrasound angles were more likely to be underestimated in the thoracic region than the (thoraco)lumbar region,^5,12 which was not observed in the present study. It should be noted that the Cobb angle was a coronal simplification of the spinal deformity. Although yielding different values, ultrasound measurements gave a reliable and valid impression of the severity of deformity, proving to be a useful tool for scoliosis screening.

Automatic and manual ultrasound angles achieved high intra- and interobserver reliability (ICCs≥0.939), which was consistent with previous studies (ICCs ranging from 0.63 to 0.99).^5,8,12,15The Bland-Altman plots illustrated that the data points of both automatic and manual measurements were tightly clustered within the limits of agreement (LoA), suggesting excellent agreement with the Cobb angle. Moreover, both manual and automatic ultrasound angles produced MAD within the clinically acceptable error of 5°, with the manual method demonstrating a significantly lower MAD.¹¹

The MDC values for the automatic (4.93°) and manual (up to 4.43°) ultrasound angles were close to the widely accepted threshold of 5° for clinically relevant curve progression on radiographs.¹⁸ Langensiepen et al reported that even with standardized manual Cobb angle measurement, repeated measurements may vary by up to 5°, and a change of 5° or more is considered true progression.¹⁸ Our findings suggested that both automatic and manual ultrasound measurements had a measurement error comparable to that of conventional radiography, supporting their potential for longitudinal monitoring of scoliosis curvature. However, it should be noted that the 5° threshold was originally established for radiographic Cobb angles, and its direct application to ultrasound-derived angles warrants further validation.

Based on the Chinese BMI classification standards, subjects were stratified into underweight, normal-weight, and overweight subgroups. When assessing the agreement between automatic ultrasound angle and the Cobb angle, the normal-weight group demonstrated the highest correlation coefficient (R² = 0.888), whereas the overweight group showed the lowest (R² = 0.425). A plausible explanation was that the excessive thickness of back soft tissue in the overweight group impeded clear identification of bony landmarks, while the excessively thin soft tissue in the underweight group resulted in poor contact between the probe and the skin, both of which may reduce the accuracy of ultrasound imaging.¹⁵ However, the correlation between manual ultrasound angle and the Cobb angle remained comparable across different BMI subgroups. Despite reduced ultrasound image quality due to excessively thin or thick soft tissue, skilled observers could still accurately identify bony landmarks, thereby maintaining measurement consistency.

41.5% of the subjects were confirmed by X-ray as non-scoliosis individuals. Subgroup analysis based on the scoliosis status demonstrated that the ultrasound angles showed stronger correlations with the Cobb angle in the scoliosis group. Yang et al who classified scoliosis patients into mild, moderate, and severe curve groups (Cobb angle ≥ 35°), reported that ultrasound angles demonstrated the poorest agreement with Cobb angle in the severe curve group.⁸ They attributed this phenomenon to vertebral rotation, rib humps and thoracic cage deformation resulting from severe scoliosis. However, no trend was observed where the correlation between ultrasound angles and the Cobb angle decreased with increasing scoliosis severity, as non-severe curves (Cobb angle <35°) accounted for 96.8% of the enrolled cohort.

This research has several limitations. Firstly, the single-center nature of this study restricts generalizability, highlighting the need for future multicenter studies with a larger sample size to evaluate the reliability and accuracy of ultrasound angles. Secondly, the absence of follow-up data for scoliosis patients in this study precludes the assessment of ultrasound’s effectiveness in monitoring curve progression. Thirdly, the ultrasound scans and radiographs were obtained non-simultaneously, with an interval of within 2 hours. Consequently, potential changes in subject posture between the two examinations may lead to discrepancies in spinal curvature measurements derived from ultrasound images versus radiographs. Lastly, the enrolled cohort consisted predominantly of patients with mild curves, and the current findings should not be generalized to moderate or severe scoliosis (Cobb angle ≥ 35°) without further validation.

Conclusion

This study demonstrated that ultrasound provided reliable and valid measurements of spinal curvature, offering a radiation-free alternative to conventional radiography. When utilizing ultrasound for scoliosis screening, emphasis should be placed on manual ultrasound angle. Additionally, the influence of the subject’s BMI and scoliosis status on ultrasound measurements should be considered in the assessment of curvature severity.

Footnotes

ORCID iDs

Shuaiqi Zhu

Shuai Xu

Ethical Considerations

This prospective study received ethical approval from the Institutional Review Board of Peking University People’s Hospital (2024PHB135). Informed consent was obtained from all patients and/or their parents.

Consent to Participate

All authors have signed patient consent forms.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by (1) Capital’s Funds for Health Improvement and Research (CFH) [grant number ShouFa2026-2G-40812], (2) Research and Development Fund of Peking University People’s Hospital [grant number RS-2025-03], (3) the Beijing Natural Science Foundation [grant number 7232182], (4) Peking University Clinical Scientist Training Program (grant number BMU2024PYJH016), (5) China Association for Science and Technology: Innovative Service for Management of Scoliosis in Tibetan Adolescents (grant number: N/A).

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Altaf

Gibson

Dannawi

Noordeen

. Adolescent idiopathic scoliosis. Bmj. 2013;346(apr30 1):f2508. doi:10.1136/bmj.f2508.

Zhu

Zhao

, et al. Scoliosis Screening for Students From Primary and Junior Schools Using Electronic Devices. Spine. 2025;50(24):1788-1797. doi:10.1097/brs.0000000000005377.

Cool

Streekstra

van Schuppen

Stadhouder

van den Noort

van Royen

. Estimated cumulative radiation exposure in patients treated for adolescent idiopathic scoliosis. European Spine Journal. 2023;32(5):1777-1786. doi:10.1007/s00586-023-07651-2.

Luan

F-J

Wan

Mak

K-C

C-J

Wang

H-Q

. Cancer and mortality risks of patients with scoliosis from radiation exposure: a systematic review and meta-analysis. European Spine Journal. 2020;29(12):3123-3134. doi:10.1007/s00586-020-06573-7.

de Reuver

Brink

Lee

TTY

Zheng

Y-P

Beek

FJA

Castelein

. Cross-validation of ultrasound imaging in adolescent idiopathic scoliosis. European Spine Journal. 2020;30(3):628-633. doi:10.1007/s00586-020-06652-9.

Lee

TT-Y

Lai

KK-L

Cheng

JC-Y

Castelein

Lam

T-P

Zheng

Y-P

. 3D ultrasound imaging provides reliable angle measurement with validity comparable to X-ray in patients with adolescent idiopathic scoliosis. Journal of Orthopaedic Translation. 2021;29:51-59. doi:10.1016/j.jot.2021.04.007.

Liu

Zhao

, et al. Validity and reliability of 3D automatic ultrasound imaging for measuring coronal and sagittal angles in idiopathic scoliosis. Scientific Reports. 2025;15(1):33558. doi:10.1038/s41598-025-18538-1.

Yang

, et al. Three-dimensional ultrasonography: Advancing the clinical evaluation of adolescent idiopathic scoliosis. Ultrasonics Sonochemistry. 2025;119:107391. doi:10.1016/j.ultsonch.2025.107391.

Zimmermann

Hülstrunk

Schmitt

, et al. Comparison of two ultrasound methods from Scolioscan® for measuring spinal curvature: spinous processes (SP) versus transverse processes (TP). European Spine Journal. 2025;34(7):2677-2686. doi:10.1007/s00586-025-09007-4.

10.

Huang

Zhao

Leung

FHF

, et al. Joint Spine Segmentation and Noise Removal From Ultrasound Volume Projection Images With Selective Feature Sharing. IEEE Transactions on Medical Imaging. 2022;41(7):1610-1624. doi:10.1109/tmi.2022.3143953.

11.

Zeng

H-Y

Lou

S-H

Liu

Z-C

Zheng

. Automatic Detection and Measurement of Spinous Process Curve on Clinical Ultrasound Spine Images. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control. 2021;68(5):1696-1706. doi:10.1109/tuffc.2020.3047622.

12.

Brink

Wijdicks

SPJ

Tromp

, et al. A reliability and validity study for different coronal angles using ultrasound imaging in adolescent idiopathic scoliosis. The Spine Journal. 2018;18(6):979-985. doi:10.1016/j.spinee.2017.10.012.

13.

Lai

KK-L

Lee

TT-Y

Lee

MK-S

Hui

JC-H

Zheng

Y-P

. Validation of Scolioscan Air-Portable Radiation-Free Three-Dimensional Ultrasound Imaging Assessment System for Scoliosis. Sensors. 2021;21(8):2858. doi:10.3390/s21082858.

14.

Nguyen

TNN

Emery

Stampe

Hryniuk Southon

Lou

. Reliability and accuracy of scoliotic parameters on using a wireless handheld 3D ultrasound for children with adolescent idiopathic scoliosis: a pilot study. European Spine Journal. 2024;33(10):3784-3791. doi:10.1007/s00586-024-08445-w.

15.

Babaee

Zou

SLY

Beygi

Luo

Wong

. Influence of body mass index on the reliability and validity of ultrasound assessments in adolescent idiopathic scoliosis: An original research. PLOS One. 2025;20(11):e0335472. doi:10.1371/journal.pone.0335472.

16.

Kwan

Yang

Hung

, et al. Ultrasound curve angle is more accurate than spinous process angle and can reduce unnecessary radiography in school scoliosis screening: a prospective and validation study in 507 adolescent females. Bone&Joint. 2025;107-B:1220-1228. (Electronic).

17.

Doody

Lonstein

Stovall

Hacker

Luckyanov

Land

. Breast Cancer Mortality After Diagnostic Radiography Findings From the U.S. Scoliosis Cohort Study. Spine. 2000;25(16):2052-2063.

18.

Langensiepen

Semler

Sobottke

, et al. Measuring procedures to determine the Cobb angle in idiopathic scoliosis: a systematic review. European Spine Journal. 2013;22(11):2360-2371. doi:10.1007/s00586-013-2693-9.