Whole-spine Lateral Radiographs Research Articles

Artificial Intelligence Assistance for the Measurement of Full Alignment Parameters in Whole-Spine Lateral Radiographs

BackgroundMeasuring spinal alignment with radiological parameters is essential in patients with spinal conditions likely to be treated surgically. These evaluations are not usually included in the radiological report. As a result, spinal surgeons commonly perform the measurement, which is time-consuming and subject to errors. We aim to develop a fully automated artificial intelligence tool to assist in measuring alignment parameters in whole-spine lateral radiograph (WSL X-rays). Material and MethodsWe developed a tool called Vertebrai that automatically calculates the global spinal parameters (GSP): Pelvic incidence (PI), sacral slope (SS), pelvic tilt (PT), L1-L4 angle, L4-S1 lumbo-pelvic angle, T1 pelvic angle (TPA), sagittal vertical axis (SVA), cervical lordosis (CL), C1-C2 lordosis, lumbar lordosis (LL), mid-thoracic kyphosis (MTK), proximal thoracic kyphosis (PTK), global thoracic kyphosis (GTK), T1 slope (T1S), C2-C7 plummet (cSVA), spino-sacral angle (SSA), C7 tilt (C7T), global tilt (GT), spinopelvic tilt (T1SPi) and hip odontoid axis (OD-HA). We assessed human-AI interaction instead of AI performance alone. We compared the time to measure GSP and inter-rater agreement with and without AI assistance. Two institutional datasets were created with 2267 multilabel images for classification, and 784 WSL X-rays with reference standard landmark labeled by spinal surgeons. ResultsVertebrai significantly reduced the measurement time comparing spine surgeons with AI assistance and the AI algorithm alone, without human intervention (3 minutes vs 0.26 minutes; p < 0.05). Vertebrai achieved an average accuracy of 83% in detecting abnormal alignment values, with the SS parameter exhibiting the lowest accuracy at 61.5% and T1SPi demonstrating the highest accuracy at 100%. Intra-class correlation analysis revealed a high level of correlation and consistency in the global alignment parameters. ConclusionsVertebrai's measurements can accurately detect alignment parameters, making it a promising tool for measuring GSP automatically.

Machine Learning-Based Measurement of Regional and Global Spinal Parameters Using the Concept of Incidence Angle of Inflection Points.

This study delves into the application of convolutional neural networks (CNNs) in evaluating spinal sagittal alignment, introducing the innovative concept of incidence angles of inflection points (IAIPs) as intuitive parameters to capture the interplay between pelvic and spinal alignment. Pioneering the fusion of IAIPs with machine learning for sagittal alignment analysis, this research scrutinized whole-spine lateral radiographs from hundreds of patients who visited a single institution, utilizing high-quality images for parameter assessments. Noteworthy findings revealed robust success rates for certain parameters, including pelvic and C2 incidence angles, but comparatively lower rates for sacral slope and L1 incidence. The proposed CNN-based machine learning method demonstrated remarkable efficiency, achieving an impressive 80 percent detection rate for various spinal angles, such as lumbar lordosis and thoracic kyphosis, with a precise error threshold of 3.5°. Further bolstering the study's credibility, measurements derived from the novel formula closely aligned with those directly extracted from the CNN model. In conclusion, this research underscores the utility of the CNN-based deep learning algorithm in delivering precise measurements of spinal sagittal parameters, and highlights the potential for integrating machine learning with the IAIP concept for comprehensive data accumulation in the domain of sagittal spinal alignment analysis, thus advancing our understanding of spinal health.

AI-Driven Segmentation and Automated Analysis of the Whole Sagittal Spine from X-ray Images for Spinopelvic Parameter Evaluation.

Spinal-pelvic parameters are utilized in orthopedics for assessing patients' curvature and body alignment in diagnosing, treating, and planning surgeries for spinal and pelvic disorders. Segmenting and autodetecting the whole spine from lateral radiographs is challenging. Recent efforts have employed deep learning techniques to automate the segmentation and analysis of whole-spine lateral radiographs. This study aims to develop an artificial intelligence (AI)-based deep learning approach for the automated segmentation, alignment, and measurement of spinal-pelvic parameters through whole-spine lateral radiographs. We conducted the study on 932 annotated images from various spinal pathologies. Using a deep learning (DL) model, anatomical landmarks of the cervical, thoracic, lumbar vertebrae, sacrum, and femoral head were automatically distinguished. The algorithm was designed to measure 13 radiographic alignment and spinal-pelvic parameters from the whole-spine lateral radiographs. Training data comprised 748 digital radiographic (DR) X-ray images, while 90 X-ray images were used for validation. Another set of 90 X-ray images served as the test set. Inter-rater reliability between orthopedic spine specialists, orthopedic residents, and the DL model was evaluated using the intraclass correlation coefficient (ICC). The segmentation accuracy for anatomical landmarks was within an acceptable range (median error: 1.7-4.1 mm). The inter-rater reliability between the proposed DL model and individual experts was fair to good for measurements of spinal curvature characteristics (all ICC values > 0.62). The developed DL model in this study demonstrated good levels of inter-rater reliability for predicting anatomical landmark positions and measuring radiographic alignment and spinal-pelvic parameters. Automated segmentation and analysis of whole-spine lateral radiographs using deep learning offers a promising tool to enhance accuracy and efficiency in orthopedic diagnostics and treatments.

Automatic Spine Segmentation and Parameter Measurement for Radiological Analysis of Whole-Spine Lateral Radiographs Using Deep Learning and Computer Vision.

Radiographic examination is essential for diagnosing spinal disorders, and the measurement of spino-pelvic parameters provides important information for the diagnosis and treatment planning of spinal sagittal deformities. While manual measurement methods are the golden standard for measuring parameters, they can be time consuming, inefficient, and rater dependent. Previous studies that have used automatic measurement methods to alleviate the downsides of manual measurements showed low accuracy or could not be applied to general films. We propose a pipeline for automated measurement of spinal parameters by combining a Mask R-CNN model for spine segmentation with computer vision algorithms. This pipeline can be incorporated into clinical workflows to provide clinical utility in diagnosis and treatment planning. A total of 1807 lateral radiographs were used for the training (n = 1607) and validation (n = 200) of the spine segmentation model. An additional 200 radiographs, which were also used for validation, were examined by three surgeons to evaluate the performance of the pipeline. Parameters automatically measured by the algorithm in the test set were statistically compared to parameters measured manually by the three surgeons. The Mask R-CNN model achieved an average precision at 50% intersection over union (AP50) of 96.2% and a Dice score of 92.6% for the spine segmentation task in the test set. The mean absolute error values of the spino-pelvic parameters measurement results were within the range of 0.4° (pelvic tilt) to 3.0° (lumbar lordosis, pelvic incidence), and the standard error of estimate was within the range of 0.5° (pelvic tilt) to 4.0° (pelvic incidence). The intraclass correlation coefficient values ranged from 0.86 (sacral slope) to 0.99 (pelvic tilt, sagittal vertical axis).

Odontoid Incidence: A Novel Cervical Parameter Influencing Cervical Alignment From Top to Bottom.

ObjectiveBy using angulation of the axis itself, this study aims to define and analyze odontoid incidence (OI) and odontoid tilt (OT) as novel cervical alignment parameters and investigate their correlations with cervical alignment. MethodsNovel and existing parameters were measured with whole-spine lateral plain radiographs and EOS images of 42 adults without cervical symptoms. The correlations of OI, OT, C2 slope (C2S), and T1 slope (T1S) were calculated. ResultsThe OI, OT, and C2S showed significant correlations with C2–7 angle (r = 0.43, r = -0.42, r = 0.62, respectively) and C0–2 angle (r = -0.33, r = 0.48, r = -0.61, respectively). OI, OT, T1S were independent predictors of the C2–7 angle in univariate regression analysis (adjusted-R2 = 0.17, R2 = 0.15, R2 = 0.28, respectively). OI, OT, and T1S were independent predictors in the multivariable regression analysis with estimated standardized coefficients of 0.36, -0.67, -0.69, respectively (adjusted- R2 = 0.80, p < 0.001). Regarding the C0–2 angle, OI and OT were independent predictors in the univariate regression analysis (adjusted-R2 = 0.08, R2 = 0.21, respectively). ConclusionOI, OT, and C2S had significant correlations with cervical alignment. As the pelvic incidence, the OI is the only anatomical and constant parameter that could be used as a reference point related to the cervical spine from the rostral end. The study results may serve as baseline data for further studies on the alignment and balance of the cervical spine.

Automatic recognition of whole-spine sagittal alignment and curvature analysis through a deep learning technique

PurposeArtificial intelligence based on deep learning (DL) approaches enables the automatic recognition of anatomic landmarks and subsequent estimation of various spinopelvic parameters. The locations of inflection points (IPs) and apices (APs) in whole-spine lateral radiographs could be mathematically determined by a fully automatic spinal sagittal curvature analysis system.MethodsWe developed a DL model for automatic spinal curvature analysis of whole-spine lateral plain radiographs by using 1800 annotated images of various spinal disease etiologies. The DL model comprised a landmark localizer to detect 25 vertebral landmarks and a numerical algorithm for the generation of an individualized spinal sagittal curvature. The characteristics of the spinal curvature, including the IPs, APs, and curvature angle, could thus be analyzed using mathematical definitions. The localization error of each landmark was calculated from the predictions of 300 test images to evaluate the performance of the landmark localizer. The interrater reliability among a senior orthopedic surgeon, a radiologist, and the DL model was assessed using the intraclass correlation coefficient (ICC).ResultsThe accuracy of the landmark localizer was within an acceptable range (median error: 1.7–4.1 mm), and the interrater reliabilities between the proposed DL model and each expert were good to excellent (all ICCs > 0.85) for the measurement of spinal curvature characteristics.ConclusionThe interrater reliability between the proposed DL model and human experts was good to excellent in predicting the locations of IPs, APs, and curvature angles. Future applications should be explored to validate this system and improve its clinical efficiency.

Analysis of measurement changes in pelvic incidence according to pelvic rotation using a three-dimensional model

BackgroundPelvic incidence (PI) is used as a key parameter in surgical correction of adult spinal deformity (ASD). However, reflecting the exact center or inclination of the three-dimensional anatomical structures on the two-dimensional (2D) sagittal radiographs is limited, resulting in measurement errors. Therefore, we evaluated whether there is a change in PI measurement according to the actual rotation of the pelvis, and conducted a study on a more accurate method for PI measurement using 2D sagittal radiographs.MethodsFrom 2014 to 2015, the data of 30 patients who visited our outpatient clinic were analyzed retrospectively. CT scans including those of the lower lumbar spine, pelvis, and both femurs in the DICOM format were imported to Mimics Research 17.0 (Materialise NV, Belgium), SolidWorks (Dassault systems, France), and AutoCAD 2014 (AUTODESK, US). The changes in PI according to vertical and horizontal pelvic rotations were evaluated.ResultsThe average PIs according to the horizontal pelvic rotations measured on AutoCAD with 0°, 5°, 10°, 15°, 20°, 25°, 30°, 35°, and 40° were 48.8°, 48.7°, 48.3°, 47.8°, 46.9°, 45.6°, 44.0°, 42.2°, and 39.9°, respectively. The PI with an acceptable error of 6° on radiographs was 35° in the horizontal pelvic rotation. The average PIs according to the vertical pelvic rotations measured on AutoCAD with 0°, 5°, 10°, 15°, 20°, 25°, 30°, 35°, and 40° were 48.8°, 49.0°, 49.5°, 50.2°, 51.3°, 52.7°, 54.4°, 56.6°, and 59.4°, respectively. The PI with an acceptable error of 6° on radiographs was 30° in the vertical pelvic rotation.ConclusionsThis study revealed that the PI value could differ from the actual anatomical value due to the horizontal and vertical rotation of the pelvis while acquiring the radiograph. Regarding whole-spine lateral radiographs, errors in PI measurement may occur due to pelvic rotation or nonvertical projection of X-rays. In the standing pelvic lateral radiographs, ensuring superposition of the femoral heads at the center and obtaining the straight sacral endplate by referring to CT or magnetic resonance imaging would be a more accurate measurement method to define PI.

Deep learning approach for automatic landmark detection and alignment analysis in whole-spine lateral radiographs

Human spinal balance assessment relies considerably on sagittal radiographic parameter measurement. Deep learning could be applied for automatic landmark detection and alignment analysis, with mild to moderate standard errors and favourable correlations with manual measurement. In this study, based on 2210 annotated images of various spinal disease aetiologies, we developed deep learning models capable of automatically locating 45 anatomic landmarks and subsequently generating 18 radiographic parameters on a whole-spine lateral radiograph. In the assessment of model performance, the localisation accuracy and learning speed were the highest for landmarks in the cervical area, followed by those in the lumbosacral, thoracic, and femoral areas. All the predicted radiographic parameters were significantly correlated with ground truth values (all p < 0.001). The human and artificial intelligence comparison revealed that the deep learning model was capable of matching the reliability of doctors for 15/18 of the parameters. The proposed automatic alignment analysis system was able to localise spinal anatomic landmarks with high accuracy and to generate various radiographic parameters with favourable correlations with manual measurements.

Relationships and Usefulness of Cervical Lateral Radiographs Compared with Whole-Spine Lateral Radiographs for Evaluating Cervical Sagittal Alignment

Relationships and Usefulness of Cervical Lateral Radiographs Compared with Whole-Spine Lateral Radiographs for Evaluating Cervical Sagittal Alignment

Relationships and Usefulness of Cervical Lateral Radiographs Compared with Whole-Spine Lateral Radiographs for Evaluating Cervical Sagittal Alignment

Relationships and Usefulness of Cervical Lateral Radiographs Compared with Whole-Spine Lateral Radiographs for Evaluating Cervical Sagittal Alignment

Evaluating Cervical Sagittal Alignment in Cervical Myelopathy: Are Sitting Cervical Radiographs and Standing Whole-Spine Radiographs Equally Useful?

Study Design:Retrospective review of medical charts and radiographic data.Objectives:We aimed to clarify the differences in cervical alignment findings between sitting cervical lateral radiographs and standing whole-spine lateral radiographs with clavicle positioning in cervical spondylotic myelopathy (CSM) patients.Methods:We retrospectively evaluated the radiographs of 50 consecutive patients who underwent cervical surgery for CSM in our hospital. Cervical sagittal alignment was evaluated based on the C0-2 angles and C2-7 Gore and Cobb angles. Head position was evaluated in terms of the center of gravity of the head to C7 (CGH-C7) angle and the McGregor angle (ie, the angle between the McGregor line and a horizontal line). The T1-slope was also evaluated.Results:The mean values of the CGH-C7 angle and T1-slope were significantly lower, while the mean value of the McGregor angle was significantly higher on whole-spine lateral radiographs with clavicle positioning than on sitting cervical lateral radiographs. The mean values of the C0-2 and C2-7 angles did not differ significantly between the 2 radiographic positioning approaches.Conclusions:Using whole-spine lateral radiographs with clavicle positioning may result in a significantly lower T1-slope and a posterior tilt of the head. In the absence of a compensatory change in cervical alignment, clavicle positioning may force patients to adopt an upward gazing position of the head. These compensatory mechanisms should be considered while evaluating cervical alignment on whole-spine lateral radiographs with clavicle positioning. Surgical planning should take into account the effect of posture on the radiographic appearance of cervical alignment.

Influence of the Type of Sagittal Profile on Clinical Results of Lumbar Total Disk Replacement After a Mean Follow-Up of 39 Months

Retrospective analysis of clinical and radiologic data of a prospective cohort study. To research the clinical differences after lumbar total disk replacement (TDR) with respect to the preoperative global and the adaptation at the local sagittal profile (SP) of the spine. It was suggested that facet loads and degeneration are dependent on epidemiologically defined types of SP. Moreover, the success of TDR was related to segmental facet joint loads. The influences of the preoperative SP or of the changes of the local SP after TDR on the clinical outcome after TDR remain unclear. Fifty-two patients included in a prospective cohort study regarding lumbar single-level TDR L4/5 (n=22) or L5/S1 (n=30) because of degenerative disk disease (Modic ≤2 degrees) were clinically (visual analog scale for back, leg, and overall pain; Oswestry Disability Index) and radiologically (extension-flexion radiographs, plain-spine, and whole-spine lateral radiographs in upright standing position) reevaluated after a minimum follow-up of 24 (24-69) months. On the basis of preoperative plain radiographs in upright standing position, patients were retrospectively assigned to 4 groups according to the individual sagittal profile type (SPT). In patients with persistent back pain, a facet infiltration at the index level was performed. For all patients, an SPT could be defined. Global SP did not change compared with the preoperative state. All groups improved clinically over follow-up. At the last follow-up, types 1 and 4 demonstrated significantly inferior scores for pain and function. TDR-induced changes at the superior adjacent segment and the posterior disk height at the index level were also correlated to inferior clinical results. Infiltration test was positive in type 1-4: 67%, 40%, 33%, and 75%, respectively, of the symptomatic patients. We suggest SPTs 1 and 4 to represent a contraindication for lumbar TDR of levels L4/5 or L5/S1. Local adaptation in the adjacent segment to TDR may influence the clinical outcome as well.

Radiographic Comparison between Cervical Spine Lateral and Whole-Spine Lateral Standing Radiographs.

Study Design Retrospective radiologic study. Objective The sagittal alignment of the cervical spine can be evaluated using either a lateral cervical radiograph or a whole-spine lateral view on which the cervical spine is included. To our knowledge, however, no report has compared the two. The purpose of this work is to identify the difference in radiographic parameters between the cervical spine lateral view and the whole-spine lateral view. Methods We retrospectively analyzed 59 adult patients suffering from neck pain with cervical spine lateral radiographs and whole-spine lateral radiographs from November 2007 to December 2011. The radiographs were measured using standard techniques to obtain the following parameters from the two different radiographs: occipital–C2 angle, C2–C7 angle, C7–sternal angle, sternal slope, T1 slope, C2 central offset distance, the distance between C2 and C7 plumb lines, C4 anteroposterior (AP) diameter, the ratio of C2 central off distance to C4 AP diameter, the ratio of plumb lines' distance to C4 AP diameter. Results We found that the occipital–C2 angle, sternal slope, and C4 AP diameter were similar, but the C2–C7 angle, C7–sternal angle, T1 slope, C2 central offset distance, distance between C2 and C7 plumb lines, ratio of C2 central off distance to C4 AP diameter, and ratio of plumb lines' distance to C4 AP diameter were different. However, the error of measurement was greater than the small angular and linear differences between the two views. Conclusions Most numerical values of the measured radiographic parameters appear to be different between the two views. However, the two views are comparable because the numerical differences were smaller than the errors of measurement.

Validation of a simple computerized tool for measuring spinal and pelvic parameters

Recent studies have emphasized measuring the sagittal vertical axis (SVA) and pelvic parameters (pelvic incidence, sacral slope, and pelvic tilt) when evaluating spinal disorders. An accurate and reproducible measurement is important for a reliable result. Although computerized measurement is more consistent than manual measurement, computerized measurement requires an expensive software program, the need to transfer images to a workstation, and additional education for users. An inexpensive and convenient computerized measurement program is desirable and necessary. The object of this study was to propose a computerized tool for measuring spinal and pelvic parameters and to evaluate the efficacy of this new tool compared with manual measurement. The authors devised a tool that provides computerized measurements of the SVA and pelvic parameters in a picture archiving and communication system (PACS) without transferring images to another program. This tool was created by merging functions in the PACS. The resulting tool is easy to implement by merging functions (indicate the center of 2 points, plot a vertical and a horizontal line from a point, and measure the angles between lines) in any image viewer. The tool was made into icons on a toolbar in the PACS. Measurements of distance and angle were computerized by identifying crucial points after selecting the icon. For SVA, 4 points were identified around each corner of the C-7 body and a fifth point at the superior/posterior corner of the S-1 body. For pelvic parameters, 4 points were identified at the centers of each femoral head and at the anterior/superior and posterior/superior corners of S-1. Thirty-three whole-spine lateral radiographs were randomly selected from the radiographic database. To evaluate inter- and intraobserver variability between observers and method, skilled (2 years of experience) and unskilled (1 week of experience) observers measured SVA and pelvic parameters 3 times with a 7-day interval between each time using both computerized and manual measurement methods. The reliability was measured using the intraclass correlation coefficient. The computerized method showed better congruity than the manual method in both skilled and unskilled observers (p < 0.05), and the intraclass correlation coefficients were > 0.9. The skilled observer showed better agreement than the unskilled observer with both computerized and manual methods, and this difference was prominent in measuring pelvic parameters (p < 0.05). The computerized method required less time than the manual method, especially for the unskilled observer (p < 0.05). A computerized measurement of pelvic parameters may be a more reliable and efficacious approach than manual measurements. This benefit is more prominent in the unskilled observer, and adding this simple function to an image viewer may be recommended in future studies.

Remodeling of adjacent spinal alignments following cervical arthroplasty and anterior discectomy and fusion

To evaluate the effects of cervical artificial disc replacement (ADR) and anterior discectomy and fusion (ACDF) on adjacent spinal alignments. The cohort consisted of 33 patients who undergone single-level cervical ADR (15 patients) and ACDF (18 patients) for radiculopathy, who had not had any previous spine surgery, and who had a minimum follow-up of 2 years. Whole-spine lateral radiographs were taken at the pre-operative and follow-up consultations. Cervical lordosis, thoracic kyphosis, lumbar lordosis, and sagittal balance were measured each time. The patients filled out pre-operative and follow-up functional evaluation forms including visual analogue scale (VAS) of neck and arm. The mean follow-up durations of patients who had cervical ADR and ACDF were 28 ± 5.0 and 30 ± 5.8 months, respectively. The patients having ACDF had the higher mean age (53 ± 9.0 years) than that of patients with cervical ADR (45 ± 11.7 years). The cervical lordosis and thoracic kyphosis in cervical ADR group increased significantly more than those of the ACDF group in follow-up assessment (P = 0.011 and 0.012). There was no significant change of lumbar lordosis in intra- and inter-group analyses. The follow-up sagittal balances for the cervical ADR and ACDF groups moved towards a neutral value. Although the follow-up neck and arm VAS of the both groups improved than those of the pre-operative status, the groups did not differ significantly except for a difference in neck VAS, which improved more after ADR. The remodeling of cervical and thoracic curves after cervical ADR and ACDF was coupled and complementary. Cervical ADR contributed the restorations of angulations of cervical and thoracic spines. The neck VAS improved more after cervical ADR than after ACDF.