Machine Learning Predicts the 3D Outcomes of Adolescent Idiopathic Scoliosis Surgery Using Patient-Surgeon Specific Parameters

Patient-Specific Surgical Correction of Adolescent Idiopathic Scoliosis: A Systematic Review

The restoration of sagittal alignment is fundamental to the surgical correction of adolescent idiopathic scoliosis (AIS). Despite established techniques, some patients present with inadequate postoperative thoracic kyphosis (TK), which may increase the risk of proximal junctional kyphosis (PJK) and imbalance. There is a lack of knowledge concerning the effectiveness of patient-specific rods (PSR) with measured sagittal curves in achieving a TK similar to that planned in AIS surgery, the factors influencing this congruence, and the incidence of PJK after PSR use. This is a systematic review of all types of studies reporting on the PSR surgical correction of AIS, including research articles, proceedings, and gray literature between 2013 and December 2023. From the 28,459 titles identified in the literature search, 81 were assessed for full-text reading, and 7 studies were selected. These included six cohort studies and a comparative study versus standard rods, six monocentric and one multicentric, three prospective and four retrospective studies, all with a scientific evidence level of 4 or 3. They reported a combined total of 355 AIS patients treated with PSR. The minimum follow-up was between 4 and 24 months. These studies all reported a good match between predicted and achieved TK, with the main difference ranging from 0 to 5 degrees, p > 0.05, despite the variability in surgical techniques and the rods' properties. There was no proximal junctional kyphosis, whereas the current rate from the literature is between 15 and 46% with standard rods. There are no specific complications related to PSR. The exact role of the type of implants is still unknown. The preliminary results are, therefore, encouraging and support the use of PSR in AIS surgery.

Rod Angulation Relationship with Thoracic Kyphosis after Adolescent Idiopathic Scoliosis Posterior Instrumentation

INTRODUCTION

Surgery to correct spinal deformities in scoliosis involves the use of contoured rods to reshape the spine and correct its curvatures. It is crucial to bend these rods appropriately to achieve the best possible correction. However, there is limited research on how the rod bending process relates to spinal shape in adolescent idiopathic scoliosis surgery.

METHODS

A retrospective study was conducted using a prospective multicenter scoliosis database. This study included adolescent idiopathic scoliosis patients from the database who underwent surgery with posterior instrumentation covering the T4 to T12 segments. Standing global spine X-rays were used in the analysis. The sagittal Cobb angles between T5 and T11 were measured on the spine. Additionally, the curvature of the rods between T5 and T11 was measured using the tangent method. To assess the relationship between these measurements, the difference between the dorsal kyphosis (TK) and the rod kyphosis (RK) was calculated (ΔK = TK - RK). This study aimed to analyze the correlation between ΔK and various patient characteristics. Both descriptive and statistical analyses were performed to achieve this goal.

RESULTS

This study encompassed a cohort of 99 patients, resulting in a total of 198 ΔK measurements for analysis. A linear regression analysis was conducted, revealing a statistically significant positive correlation between the kyphosis of the rods and that of the spine (r = 0.77, p = 0.0001). On average, the disparity between spinal and rod kyphosis averaged 5.5°. However, it is noteworthy that despite this modest mean difference, there was considerable variability among the patients. In particular, in 84% of cases, the concave rod exhibited less kyphosis than the spine, whereas the convex rod displayed greater kyphosis than the spine in 64% of cases. It was determined that the primary factor contributing to the flattening of the left rod was the magnitude of the coronal Cobb angle, both before and after the surgical procedure. These findings emphasize the importance of considering individual patient characteristics when performing rod bending procedures, aiming to achieve the most favorable outcomes in corrective surgery.

CONCLUSIONS

Although there is a notable and consistent correlation between the curvature of the spine and the curvature of the rods, it is important to acknowledge the substantial heterogeneity observed in this study. This heterogeneity suggests that individual patient factors play a significant role in shaping the outcome of spinal corrective surgery. Furthermore, this study highlights that more severe spinal curvatures in the frontal plane have an adverse impact on the shape of the rods in the sagittal plane. In other words, when the scoliosis curve is more pronounced in the frontal plane, it tends to influence the way the rods are shaped in the sagittal plane. This underscores the complexity of spinal deformities and the need for a tailored approach in surgical interventions to account for these variations among patients.

Developing a new tool for scoliosis screening in a tertiary specialistic setting using artificial intelligence: a retrospective study on 10,813 patients: 2023 SOSORT award winner

PURPOSE

The study aims to assess if the angle of trunk rotation (ATR) in combination with other readily measurable clinical parameters allows for effective non-invasive scoliosis screening.

METHODS

We analysed 10,813 patients (4-18 years old) who underwent clinical and radiological evaluation for scoliosis in a tertiary clinic specialised in spinal deformities. We considered as predictors ATR, Prominence (mm), visible asymmetry of the waist, scapulae and shoulders, familiarity, sex, BMI, age, menarche, and localisation of the curve. We implemented a Logistic Regression model to classify the Cobb angle of the major curve according to thresholds of 15, 20, 25, 30, and 40 degrees, by randomly splitting the dataset into 80-20% for training and testing, respectively.

RESULTS

The model showed accuracies of 74, 81, 79, 79, and 84% for 15-, 20-, 25-, 30- and 40-degrees thresholds, respectively. For all the thresholds ATR, Prominence, and visible asymmetry of the waist were the top five most important variables for the prediction. Samples that were wrongly classified as negatives had always statistically significant (p ≪ 0.01) lower values of ATR and Prominence. This confirmed that these two parameters were very important for the correct classification of the Cobb angle. The model showed better performances than using the 5 and 7 degrees ATR thresholds to prescribe a radiological examination.

CONCLUSIONS

Machine-learning-based classification models have the potential to effectively improve the non-invasive screening for AIS. The results of the study constitute the basis for the development of easy-to-use tools enabling physicians to decide whether to prescribe radiographic imaging.

Patient and surgical predictors of 3D correction in posterior spinal fusion: a systematic review

BACKGROUND

Restoration of three-dimensional (3D) alignment is critical in correcting patients with adolescent idiopathic scoliosis using posterior spinal fusion (PSF). However, current studies mostly rely on 2D radiographs, resulting in inaccurate assessment of surgical correction and underlying predictive factors. While 3D reconstruction of biplanar radiographs is a reliable and accurate tool for quantifying spinal deformity, no study has reviewed the current literature on its use in evaluating surgical prognosis.

PURPOSE

To summarize the current evidence on patient and surgical factors affecting sagittal alignment and curve correction after PSF based on 3D parameters derived from reconstruction of biplanar radiographs.

METHODS

A comprehensive search was conducted by three independent investigators on Medline, PubMed, Web of Science, and Cochrane Library to obtain all published information on predictors of postoperative alignment and correction after PSF. Search items included "adolescent idiopathic scoliosis," "stereoradiography," "three-dimensional," "surgical," and "correction." The inclusion and exclusion criteria were carefully defined to include clinical studies. Risk of bias was assessed with the Quality in Prognostic Studies tool, and level of evidence for each predictor was rated with the Grading of Recommendations, Assessment, Development, and Evaluations approach. 989 publications were identified, with 444 unique articles subjected to full-text screening. Ultimately, 41 articles were included.

RESULTS

Strong predictors of better curve correction included preoperative normokyphosis (TK > 15°), a corresponding rod contour, intraoperative vertebral rotation and translation, and upper and lower instrumented vertebrae selected based on sagittal and axial inflection points. For example, for Lenke 1 patients with junctional vertebrae above L1, fusion to NV-1 (1 level above the neutral vertebra) achieved optimal curve correction while preserving motion segments. Pre-op coronal Cobb angle and axial rotation, distal junctional kyphosis, pelvic incidence, sacral slope, and type of instrument were identified as predictors with moderate evidence. For Lenke 1C patients, > 50% LIV rotation was found to increase spontaneous lumbar curve correction. Pre-op thoracolumbar apical translation and lumbar lordosis, Ponte osteotomies, and rod material were found to be predictors with low evidence.

CONCLUSIONS

Rod contouring and UIV/LIV selection should be based on preoperative 3D TK in order to achieve normal postoperative alignment. Specifically, Lenke 1 patients with high-lying rotations should be fused distally at NV-1, while hypokyphotic patients with large lumbar curves and truncal shift should be fused at NV to improve lumbar alignment. Lenke 1C curves should be corrected using > 50% LIV rotation counterclockwise to the lumbar rotation. Further investigation should compare surgical correction between pedicle-screw and hybrid constructs using matched cohorts. DJK and overbending rods are potential predictors of postoperative alignment.

Use of machine learning in pediatric surgical clinical prediction tools: A systematic review

PURPOSE

Clinical prediction tools (CPTs) are decision-making instruments utilizing patient data to predict specific clinical outcomes, risk-stratify patients, or suggest personalized diagnostic or therapeutic options. Recent advancements in artificial intelligence have resulted in a proliferation of CPTs created using machine learning (ML)-yet the clinical applicability of ML-based CPTs and their validation in clinical settings remain unclear. This systematic review aims to compare the validity and clinical efficacy of ML-based to traditional CPTs in pediatric surgery.

METHODS

Nine databases were searched from 2000 until July 9, 2021 to retrieve articles reporting on CPTs and ML for pediatric surgical conditions. PRISMA standards were followed, and screening was performed by two independent reviewers in Rayyan, with a third reviewer resolving conflicts. Risk of bias was assessed using the PROBAST.

RESULTS

Out of 8300 studies, 48 met the inclusion criteria. The most represented surgical specialties were pediatric general (14), neurosurgery (13) and cardiac surgery (12). Prognostic (26) CPTs were the most represented type of surgical pediatric CPTs followed by diagnostic (10), interventional (9), and risk stratifying (2). One study included a CPT for diagnostic, interventional and prognostic purposes. 81% of studies compared their CPT to ML-based CPTs, statistical CPTs, or the unaided clinician, but lacked external validation and/or evidence of clinical implementation.

CONCLUSIONS

While most studies claim significant potential improvements by incorporating ML-based CPTs in pediatric surgical decision-making, both external validation and clinical application remains limited. Further studies must focus on validating existing instruments or developing validated tools, and incorporating them in the clinical workflow.

TYPE OF STUDY

Systematic Review LEVEL OF EVIDENCE: Level III.

Simultaneous Hypercorrection of Lowest Instrumented Vertebral Tilt and Main Thoracic Curve is Associated With Progression of Residual Lumbar Curve in Adolescent Idiopathic Scoliosis

STUDY DESIGN

A retrospective cohort study.

OBJECTIVE

To determine radiographic parameters, including the lowest instrumented vertebral (LIV) tilt, related to the postoperative magnitude and progression of residual lumbar curves (LCs) in adolescent idiopathic scoliosis patients who underwent posterior spinal fusion with LIV at or above L1.

SUMMARY OF BACKGROUND DATA

Although several guidelines have been proposed for thoracic curve fusion, factors related to the postoperative magnitude and potential progression of unfused LCs remained undetermined. The effect of the LIV tilt on residual LCs is also unclear.

MATERIALS AND METHODS

Patients with Lenke type 1 to 4 curves who underwent posterior spinal fusion with LIV at or above L1 with a minimum follow-up period of 2 years were evaluated. Prediction models for residual LCs were developed using multivariate linear regressions with selected radiographic parameters. Subgroup analyses, followed by sensitivity tests, were then performed for variables best predicting the progression of residual LCs.

RESULTS

A total of 130 patients were included. Multivariate linear regression analysis showed that the immediate postoperative LIV-tilt angle was associated with the immediate postoperative LCs and the prediction model for residual LCs, with high accuracy ( R =0.93 and 0.77, respectively). Sensitivity tests revealed immediate postoperative LIV-tilt angle <10° and correction rate of main thoracic curve Cobb angle >53% as predictors for progression of residual LCs, and they reached moderate discrimination when combined together as one criterion (odds ratio=16.3, 95% confidence interval=5.3-50.1; sensitivity=89%, specificity=67%, positive predicted value=51%, negative predicted value=94%).

CONCLUSION

The current study revealed that LIV tilt, as an operable factor during surgery, is not only a determinant in prediction models showing high correlation with the magnitude of postoperative LCs but a predictor for progression of residual LCs. "Immediate postoperative LIV-tilt angle <10° and correction rate of main thoracic curve Cobb angle >53%," as a united criterion, could serve as a predictor for progression of residual LCs.

Machine Learning Applications of Surgical Imaging for the Diagnosis and Treatment of Spine Disorders: Current State of the Art

Recent developments in machine learning (ML) methods demonstrate unparalleled potential for application in the spine. The ability for ML to provide diagnostic faculty, produce novel insights from existing capabilities, and augment or accelerate elements of surgical planning and decision making at levels equivalent or superior to humans will tremendously benefit spine surgeons and patients alike. In this review, we aim to provide a clinically relevant outline of ML-based technology in the contexts of spinal deformity, degeneration, and trauma, as well as an overview of commercial-level and precommercial-level surgical assist systems and decisional support tools. Furthermore, we briefly discuss potential applications of generative networks before highlighting some of the limitations of ML applications. We conclude that ML in spine imaging represents a significant addition to the neurosurgeon's armamentarium-it has the capacity to directly address and manifest clinical needs and improve diagnostic and procedural quality and safety-but is yet subject to challenges that must be addressed before widespread implementation.

An Evolution Gaining Momentum—The Growing Role of Artificial Intelligence in the Diagnosis and Treatment of Spinal Diseases

In recent years, applications using artificial intelligence have been gaining importance in the diagnosis and treatment of spinal diseases. In our review, we describe the basic features of artificial intelligence which are currently applied in the field of spine diagnosis and treatment, and we provide an orientation of the recent technical developments and their applications. Furthermore, we point out the possible limitations and challenges in dealing with such technological advances. Despite the momentary limitations in practical application, artificial intelligence is gaining ground in the field of spine treatment. As an applying physician, it is therefore necessary to engage with it in order to benefit from those advances in the interest of the patient and to prevent these applications being misused by non-medical partners.

Artificial Intelligence-Driven Prediction Modeling and Decision Making in Spine Surgery Using Hybrid Machine Learning Models

Healthcare systems worldwide generate vast amounts of data from many different sources. Although of high complexity for a human being, it is essential to determine the patterns and minor variations in the genomic, radiological, laboratory, or clinical data that reliably differentiate phenotypes or allow high predictive accuracy in health-related tasks. Convolutional neural networks (CNN) are increasingly applied to image data for various tasks. Its use for non-imaging data becomes feasible through different modern machine learning techniques, converting non-imaging data into images before inputting them into the CNN model. Considering also that healthcare providers do not solely use one data modality for their decisions, this approach opens the door for multi-input/mixed data models which use a combination of patient information, such as genomic, radiological, and clinical data, to train a hybrid deep learning model. Thus, this reflects the main characteristic of artificial intelligence: simulating natural human behavior. The present review focuses on key advances in machine and deep learning, allowing for multi-perspective pattern recognition across the entire information set of patients in spine surgery. This is the first review of artificial intelligence focusing on hybrid models for deep learning applications in spine surgery, to the best of our knowledge. This is especially interesting as future tools are unlikely to use solely one data modality. The techniques discussed could become important in establishing a new approach to decision-making in spine surgery based on three fundamental pillars: (1) patient-specific, (2) artificial intelligence-driven, (3) integrating multimodal data. The findings reveal promising research that already took place to develop multi-input mixed-data hybrid decision-supporting models. Their implementation in spine surgery may hence be only a matter of time.

Feature Importance Analysis for Postural Deformity Detection System Using Explainable Predictive Modeling Technique

This study aimed to analyze feature importance by applying explainable artificial intelligence (XAI) to postural deformity parameters extracted from a computer vision-based posture analysis system (CVPAS). Overall, 140 participants were screened for CVPAS and enrolled. The main data analyzed were shoulder height difference (SHD), wrist height difference (WHD), and pelvic height difference (PHD) extracted using a CVPAS. Standing X-ray imaging and radiographic assessments were performed. Predictive modeling was implemented with XGBoost, random forest regressor, and logistic regression using XAI techniques for global and local feature analyses. Correlation analysis was performed between radiographic assessment and AI evaluation for PHD, SHD, and Cobb angle. Main global features affecting scoliosis were analyzed in the order of importance for PHD (0.18) and ankle height difference (0.06) in predictive modeling. Outstanding local features were PHD, WHD, and KHD that predominantly contributed to the increase in the probability of scoliosis, and the prediction probability of scoliosis was 94%. When the PHD was >3 mm, the probability of scoliosis increased sharply to 85.3%. The paired t-test result for AI and radiographic assessments showed that the SHD, Cobb angle, and scoliosis probability were significant (p < 0.05). Feature importance analysis using XAI to postural deformity parameters extracted from a CVPAS is a useful clinical decision support system for the early detection of posture deformities. PHD was a major parameter for both global and local analyses, and 3 mm was a threshold for significantly increasing the probability of local interpretation of each participant and the prediction of postural deformation, which leads to the prediction of participant-specific scoliosis.

Discovering the association between the pre- and post-operative 3D spinal curve patterns in adolescent idiopathic scoliosis

BACKGROUND

The advantage of considering the three-dimensional curve patterns, including the patterns of the vertebral position and alignment, in classification of adolescent idiopathic scoliosis (AIS) patients and whether such classification system relates to the surgical outcomes are not fully determined.

METHODS

A total of 371 AIS patients who received posterior spinal fusion surgery with 2-year follow-up were included retrospectively and consecutively. The 3D positions and orientations of the T1-L5 vertebrae were calculated from the 3D reconstructions of the spines at pre-operative and 2-year follow-up, a total of 102 variables per patient. A probabilistic clustering method was used to cluster the pre-operative and 2-year follow-up 3D spinal curve patterns separately. The distributions of the Lenke types and 3D pre-operative clusters in the post-operative clusters were determined.

RESULTS

A total of nine pre-operative clusters including, four right thoracic types, three left thoracolumbar/lumbar types, one low apex right thoracic/thoracolumbar, and one left thoracic/right lumbar were determined. Three post-operative 3D curve patterns were Type 1 with higher residual proximal Cobb angle, Type 2 with lower T5-T12 kyphosis and highest pelvic incidence-lordosis mismatch, and Type 3 with larger lumbar curve magnitude and rotation compared to the other two groups. More than 50% of patients in each of the 3D pre-operative clusters had the same post-operative group.

CONCLUSION

We developed a 3D classification of the AIS patients before and two-year after spinal fusion surgery. The link between the pre- and post-operative clusters lends itself to application of this classification system in developing predictive models of the AIS surgical outcomes.

True 3D parameters of the spinal deformity in adolescent idiopathic scoliosis

BACKGROUND

Spinal deformities in adolescent idiopathic scoliosis (AIS) are measured on 2D radiographs. Due to the 3D nature of the curve in AIS, such 2D measurements fail to differentiate between the true curve patterns, which in turn may adversly impact the clinical care and surgical planning. The use of 3D models of the spinal radiographs largely remains limited to the 3D measurements of the 2D parameters. The use of the true 3D variables of the spinal curves in describing the differences between the AIS patients is not fully explored.

METHODS

A cohort of 141 Lenke 1 AIS with two-view spinal stereoradiographs and 3D models of the spines were included. The 3D model of the spine was used to determine the spinal centerlines. The writhe and torsion of the 3D centerlines, which, respectively, quantify the coiling and twist of the curve, were calculated using differential geometry. Patients were clustered based on the writhe and torsion values to determine the patient groups with significantly different 3D curve characteristics. The relationship between the writhe and torsion was statistically determined. The distribution of the writhe and torsion groups between the lumbar modifier types was determined.

RESULTS

Two writhe and two torsion clusters were determined. Lumbar orientation of plane of maximum curvature (PMC) was significantly different between the torsion clusters and thoracic and lumbar PMC and thoracic Cobb angles were significantly different between the writhe groups, p < 0.05. More than 50% of the patients had high writhe and low torsion except for Lumbar modifier C that mainly belonged to the low writhe group.

DISCUSSION

Two geometrical parameters of the spinal centerline determine true 3D characteristics of the scoliotic curves. The parameters were complimentary and weakly correlated, quantifying different characteristics of the scoliotic spines.

Three-dimensional classification of the Lenke 1 adolescent idiopathic scoliosis using coronal and lateral spinal radiographs

BACKGROUND

Classification of the spinal deformity in adolescent idiopathic scoliosis (AIS) remains two-dimensional (2D) as the spinal radiographs remain the mainstay in clinical evaluation of the disease. 3D classification systems are proposed, however are time consuming. We here aim to evaluate the clinical application of a 3D classification system by the use of only posterior-anterior and lateral radiographs in Lenke 1 adolescent idiopathic scoliosis (AIS).

METHODS

Forty Lenke 1 AIS were classified by five observers following a three-step flowchart, developed based on our previous 3D classification system. This 3D classification characterizes the curve in the frontal and sagittal views and infers the third dimension with rules based on prior data to determine the 3D subtypes of the curve. Repeated rating was performed for 20 randomly selected patients in the same cohort. In addition to the classification by the raters, the 3D model of the spines were generated to determine the actual curve subtype based on the algorithm that was originally used to develop the 3D classification system. The interobserver and intraobserver reliability and the classification accuracy were determined for both 3D and axial classifications of the cohort.

RESULTS

The interobserver reliability was moderate to strong with a kappa value between 0.61-0.89 for 3D and axial classifications. Comparing the mathematical classification and the raters' classification, the classification accuracy among all raters ranged between 56 and 89%.

CONCLUSION

We evaluated the reliability of a previously developed 3D classification system for Lenke 1 AIS patients when only two-view spinal radiographs are available. Radiologists and orthopedic surgeons were able to identify the 3D subtypes of Lenke 1 AIS from the patients' radiographs with moderate to strong reliability. The new 3D classification has the potential to identify the subtypes of the Lenke 1 AIS without a need for quantitative 3D image post-processing.