Spine surgeon versus AI algorithm full-length radiographic measurements: a validation study of complex adult spinal deformity patients

INTRODUCTION

Spinal measurements play an integral role in surgical planning for a variety of spine procedures. Full-length imaging eliminates distortions that can occur with stitched images. However, these images take radiologists significantly longer to read than conventional radiographs. Artificial intelligence (AI) image analysis software that can make such measurements quickly and reliably would be advantageous to surgeons, radiologists, and the entire health system.

MATERIALS AND METHODS

Institutional Review Board approval was obtained for this study. Preoperative full-length standing anterior-posterior and lateral radiographs of patients that were previously measured by fellowship-trained spine surgeons at our institution were obtained. The measurements included lumbar lordosis (LL), greatest coronal Cobb angle (GCC), pelvic incidence (PI), coronal balance (CB), and T1-pelvic angle (T1PA). Inter-rater intra-class correlation (ICC) values were calculated based on an overlapping sample of 10 patients measured by surgeons. Full-length standing radiographs of an additional 100 patients were provided for AI software training. The AI algorithm then measured the radiographs and ICC values were calculated.

RESULTS

ICC values for inter-rater reliability between surgeons were excellent and calculated to 0.97 for LL (95% CI 0.88-0.99), 0.78 (0.33-0.94) for GCC, 0.86 (0.55-0.96) for PI, 0.99 for CB (0.93-0.99), and 0.95 for T1PA (0.82-0.99). The algorithm computed the five selected parameters with ICC values between 0.70 and 0.94, indicating excellent reliability. Exemplary for the comparison of AI and surgeons, the ICC for LL was 0.88 (95% CI 0.83-0.92) and 0.93 for CB (0.90-0.95). GCC, PI, and T1PA could be determined with ICC values of 0.81 (0.69-0.87), 0.70 (0.60-0.78), and 0.94 (0.91-0.96) respectively.

CONCLUSIONS

The AI algorithm presented here demonstrates excellent reliability for most of the parameters and good reliability for PI, with ICC values corresponding to measurements conducted by experienced surgeons. In future, it may facilitate the analysis of large data sets and aid physicians in diagnostics, pre-operative planning, and post-operative quality control.

Prevalence of total joint arthroplasty in the adult spine deformity population

PURPOSE

The spine, hip, and knee are anatomically and biomechanically connected. Total hip arthroplasty (THA) and total knee arthroplasty (TKA) are commonly employed to treat degenerative changes in the hip and knee, while fusion is used for spine degeneration. Spine deformity correction results in changes in sagittal alignment and pelvic parameters, and patients with stiff spines have higher rates of THA dislocation and revision due to instability. The goal of this study was to determine the prevalence of total joint arthroplasty (TJA) in adult spinal deformity (ASD) patients at our institution.

METHODS

Following Institutional Review Board approval, we retrospectively reviewed a list of cases performed by the senior author from 4/2017 to 5/2021. Patients > 18 years old undergoing preoperative evaluation for symptomatic lumbar degeneration or ASD were included. Patients < 18 years old, those diagnosed with adolescent idiopathic scoliosis, and non-fusion cases were excluded. Perioperative full-length standing EOS images were examined for the presence or absence of THA, TKA, or both. Demographic data was collected from patient electronic medical records, and statistical analyses were completed.

RESULTS

572 consecutive cases were reviewed, and 322 were excluded. 250 cases (97M:153F) were included in the final analysis, with a mean age of 61.8 ± 11.2 years. A total of 74 patients had a TJA (29.4%). THA was present in 41 patients (16.4%), and TKA was present in 49 patients (19.6%). Males had a higher prevalence of TJA, THA, and TKA (29.9%, 16.5%, and 21.6%) than females (29.4%, 16.3%, and 18.3%).

CONCLUSIONS

This study revealed a high prevalence TJA rate of 29.4% in ASD at our institution. This rate surpasses the prevalence rate reported among the general population in previous studies. High prevalence of patients with ASD and TJA may merit special surgical consideration.

Lumbar degenerative spondylolisthesis: role of sagittal alignment

PURPOSE

To evaluate the sagittal alignment of the lumbar spine in patients with degenerative spondylolisthesis at the L4-5 level.

METHODS

Patients with untreated degenerative spondylolisthesis at L4-5 were retrospectively identified from the clinical practice of spine surgeons at an academic medical center. All patients had standing X-rays that were reviewed by the senior surgeon to confirm the presence of degenerative spondylolisthesis at L4-5. Radiographs were analyzed for the following: lumbar lordosis (LL), lower lumbar lordosis (L4-S1; LLL), L5-S1 lordosis, pelvic incidence (PI), and pelvic tilt (PT). From these measurements, lumbar distribution index (LLL/LL × 100; LDI), ideal lumbar lordosis (PI × 0.62 + 29; ILL), PI-LL mismatch, and relative lumbar lordosis (LL-ILL; RLL) were calculated. These parameters were used to evaluate the sagittal alignment of the lumbar spine. Normal alignment was defined based on previous studies and clinical experience.

RESULTS

117 participants met inclusion criteria, with an average age of 67.2 years. The majority of the cohort demonstrated hypolordotic sagittal alignment of the L5-S1 segment when assessed in relation to ILL, PI, and LL (73.5%, 61.5%, and 50.4% respectively). Evaluation of the lower lumbar spine (L4-S1) demonstrated normal sagittal alignment when evaluated via LDI and LLL (65%, 52.1%, respectively), suggesting the presence of compensatory hyperextension at L4-5 in response to the L5-S1 hypolordosis. Consequently, normal sagittal alignment of the regional lumbar spine was maintained when evaluated using LL, PI-LL mismatch, and RLL (51.3%, 47%, and 62.4% respectively).

CONCLUSIONS

This study demonstrates that there is a high incidence of relative hypolordosis at the L5-S1 level among patients who present with degenerative spondylolisthesis at L4-5. The L5-S1 hypolordosis is associated with L4-5 hyperlordosis, such that the lower lumbar lordosis (L4-S1; LLL) and regional lumbar lordosis (LL) are still within normal range. It is probable that L5-S1 hypolordosis was the initial pathologic event that incited compensatory L4-5 hyperlordosis, which in turn may have led to facet degeneration and laxity, and eventually to development of spondylolisthesis.

A Novel Surgical Indication for Scheuermann's Kyphosis

Scheuermann kyphosis can be treated surgically to restore proper sagittal alignment. Thoracic curves >70° are typically indicated for surgical intervention. However, patients who have reached their natural limit of compensatory lumbar hyperlordosis are at risk of accelerated degeneration. This can be determined by comparing lumbar lordosis on standing neutral radiographs and supine extension radiographs. Minimal additional lordosis in extension compared with neutral, abutment of the spinous processes, or greater lumbar lordosis standing than with attempted extension suggest the patient is maximally compensated. We present a case of an adolescent boy with Scheuermann kyphosis who had reached the limit of his hyperlordosis compensation reserve. He subsequently underwent a T4 to L2 posterior spinal fusion with T7 to T11 Ponte Smith-Petersen grade two osteotomies. He tolerated the procedure well with no intraoperative complications or neuromonitoring changes. The patient has continued to do well and progressed to normal activity at 5-month follow-up.

Postoperative use and early discontinuation of intravenous lidocaine in spine patients

PURPOSE

Our institution employs a multimodal approach to manage postoperative pain after spine surgery. It involves continuous intravenous (IV) lidocaine until the morning of postoperative day two. This study aimed to determine the rate and reasons for early discontinuation of IV lidocaine in our spine patients.

METHODS

We conducted a retrospective chart review and included pediatric patients who underwent ≥ 3-level spine surgery and received postoperative IV lidocaine from November 2019 to September 2022. For each case, we recorded the side effects of IV lidocaine, adverse events, time to discontinuation, and discontinuation rate. Subsequently, we used the same methodology to generate an adult cohort for comparison.

RESULTS

We included 52 pediatric (18M:34F) and 50 (21M:29F) adult patients. The pediatric cohort's mean age was 14 years (8-18), and BMI 23.9 kg/m2 (13.0-42.8). The adult cohort's mean age was 61 years (29-82), and BMI 28.8 kg/m2 (17.2-44.1). IV lidocaine was discontinued prematurely in 21/52 (40.4%) of the pediatric cases and 26/50 (52.0%) of the adult cases (RR = 0.78, p = 0.2428). The side effects noted in the pediatric cases vary, including numbness, visual disturbance, and obtundation, but no seizures. The most common adverse events were fever and motor dysfunction.

CONCLUSION

The early discontinuation rate of IV lidocaine use after spine surgery for children in our institution does not differ significantly from that of adults. The nature of the side effects and the reasons for discontinuation between the groups were similar. Thus, the safety profile of IV lidocaine for pediatric spine patients is comparable to adults.

Does A Hinged Operating Table Facilitate Sagittal Correction in Transforaminal Lumbar Interbody Fusion With Smith-Peterson Osteotomy? A Radiographic Analysis

BACKGROUND

Osteotomies allow the restoration of appropriate sagittal alignment; however, closure of osteotomies can be challenging. Typical closure involves compressing pedicle screw heads across the rods, potentially causing screw loosening and failure. Motorized hinged operating tables are often used to assist with controlled closure of osteotomies without manual compression, but there is no published research quantifying the amount of correction provided solely by changes in the table angle.

QUESTION/PURPOSE

What is the incremental amount of correction achieved by change in the table angle versus instrumented manipulation during osteotomy closure in transforaminal lumbar interbody fusion (TLIF) with Smith-Petersen osteotomy?

METHODS

Sixty-one patients undergoing Smith-Peterson osteotomy and bilateral TLIF using a motorized hinged table from October 2019 to March 2022 were prospectively enrolled. Two patients did not undergo surgery, two did not have table extension, and seven did not have data collected intraoperatively because of disruptions in research protocols owing to the coronavirus-19 pandemic. Fifty patients (24 male, 26 female) who underwent a total of 73 osteotomies were included in the final analysis. The mean ± standard deviation age was 61 ±11 years, and the mean BMI was 31 ± 6 kg/m2. Patients were positioned prone on the table and flexed to 10° for decompression, Smith-Petersen osteotomy, and TLIF. The table was then extended in 5° increments, and radiographs were taken until 10° of extension was achieved or the osteotomy was fully closed. Changes in segmental lordosis across the operative site for each 5° increment were measured to the nearest degree by two reviewers. Intraclass correlation coefficients for segmental lordosis measurements at each table angle change were calculated as 0.97 to 0.98, with all p values < 0.001, indicating excellent agreement.

RESULTS

Table change from 10° to 5° yielded a mean segmental lordosis change of 1.9° ± 1.5° (73 osteotomies), 5° to 0° yielded a change of 1.3° ± 0.9° (73 osteotomies), 0° to -5° yielded a change of 1.3° ± 1.0° (69 osteotomies), and -5° to -10° yielded a change of 1.1° ± 1.3° (61 osteotomies). Rod placement and compression yielded a mean 1.8° ± 2.0° of additional segmental lordosis.

CONCLUSION

Using a motorized hinged table facilitated an average of 5.6° of total segmental lordosis correction during controlled Smith-Peterson osteotomy closure without the need for cantilevering forces across spinal instrumentation. Surgeons can use this technique to reduce the compression forces needed to close osteotomies, which could eliminate a potential source of complications.Level of Evidence Level II, therapeutic study.

Development of a sacral fracture model to demonstrate effects on sagittal alignment

PURPOSE

To develop a modeling framework to predict the secondary consequences on spinal alignment following correction and to demonstrate the impact of pedicle subtraction osteotomy (PSO) location on sagittal alignment.

METHODS

Six patients were included, and pelvic incidence (PI) was measured. Full-length standing radiographs were uploaded into PowerPoint and manipulated to model S1-S2 joint line sacral fractures at 15°, 20°, 25°, and 30°. PSO corrections with hinge points at the anterior superior corner and vertical midpoint of the L3-5 vertebral bodies were modeled. Anterior translation (AT) and vertical shortening (VS) were calculated for the six PSO locations in the four fracture angle (FA) models.

RESULTS

PI had a strong effect in the mixed AT and VS models (P < 0.001). Both AT and VS were significantly different from zero at all FA (p < 0.001), and pairwise comparisons revealed all FA were different from each other with respect to both AT and VS after adjusting for PSO location (p < 0.001), increasing as FA increased. Varying PSO location resulted in significant differences in AT when comparing all locations (p < 0.001). AT was greatest for all FA in all patients when the PSO correction was performed at the L3-AS (p < 0.001). There were significant differences in VS when comparing the L5-Mid PSO location to the L3-AS, L3-Mid, L4-AS, and L4-Mid PSO locations (p < 0.034).

CONCLUSION

PSO correction superior to a sacral fracture resulted in AT and VS of the spine. It is crucial that these changes in spinal measures be predicted and accounted for to optimize patient sagittal alignment and outcomes.

Management of High-Grade Dysplastic Spondylolisthesis

The Meyerding classification grades the degree of slippage in the sagittal plane on lateral standing neutral imaging: 0% to 25% Grade I, 25% to 50% Grade II, 50% to 75% Grade III, 75% to 100% Grade IV, and greater than 100% Grade V (Spondyloptosis). Grades I and II are considered low-grade and Grades III-V are considered high-grade. There are several etiologies of spondylolisthesis. A classification system of the most common causes: Type I - Dysplastic, Type II - Isthmic (including subtypes: A - Lytic, B - Elongation, and C - Acute fracture), Type III - Degenerative, Type IV - Traumatic, Type V - Pathologic, and Type VI - Iatrogenic. Dysplastic spondylolisthesis is a type of spondylolisthesis that occurs at L5-S1 when dysplastic lumbosacral anatomy is present, and is associated with high-grade slip and spina bifida occulta.

Chance Fracture Pattern Presenting in Proximal Junctional Failure

INTRODUCTION

We present a case series of proximal junctional failure due to a Chance-type fracture.

METHODS

This is a retrospective review of patients who developed proximal junctional kyphosis because of Chance-type proximal junctional failure after spinal fusion for adult spinal deformity.

RESULTS

Fifteen patients were identified (4M:11F). The average age was 61.4 years (range, 39 to 77). The mean time to fracture identification was 25.4 days (range, 3 to 65). The average number of levels instrumented was 6.7 (range, 2 to 17). No patients had antecedent trauma before fracture onset. In 67% of cases with a lumbar upper instrumented vertebra (UIV), there was overcorrection of lumbar lordosis (LL) and/or lower LL. The five cases with a lower thoracic UIV had undergone notable correction of preoperative thoracolumbar junction kyphosis. 14 of 15 patients were treated with extension of fusion. Pedicle screws at the fracture level were salvaged by changing to an anatomic trajectory.

CONCLUSION

Continued pain at 6 to 12 weeks with radiographs showing an increased proximal junctional angle and cephalocaudal pedicle widening at the UIV should raise suspicion for this unique fracture pattern. A CT scan is recommended. Low bone density, LL and/or lower LL overcorrection, and selection of lower thoracic UIV in the setting of notable thoracolumbar junction correction may contribute to fracture risk.

Halo Formation and Resolution at 7-Year Follow-Up After Sacroiliac Joint Fusion Revision: A Case Report

CASE

A 49-year-old woman presented with low back pain after a work-related injury. She failed 5 months of conservative management and subsequently underwent minimally invasive (MI) left sacroiliac joint (SIJ) fusion with 3 triangular titanium implants. Four months postoperatively, she developed recurrence of symptoms and radiographic halo phenomenon about the implants. The cephalad and caudal implants were replaced with threaded self-tapping implants, and the middle implant was unable to be removed. At 7-year follow-up, the halo phenomenon had resolved.

CONCLUSION

This is an unusual case of radiographic halo phenomenon formation after MI SIJ fusion and halo resolution after subsequent revision.

Sacroiliac joint fusion navigation: how accurate is pin placement?

OBJECTIVE

Sacroiliac joint (SIJ) fusion utilizing intraoperative navigation requires a standard reference frame, which is often placed using a percutaneous pin. Proper placement ensures the correct positioning of SIJ fusion implants. There is currently no grading scheme for evaluation of pin placement into the pelvis. The purpose of this study was to evaluate the occurrence of ideal percutaneous pin placement into the posterior ilium during navigated SIJ fusion.

METHODS

After IRB approval was obtained, electronic medical records and intraoperative computed tomography images of patients who underwent navigated SIJ fusion by the senior author between October 2013 and January 2020 were reviewed. A pin placement grading scheme and the definition of "ideal" placement were developed by the authors and deemed acceptable by fellow attending surgeons. Six attending surgeons completed two rounds of pin placement grading, and statistical analysis was conducted.

RESULTS

Of 90 eligible patients, 73.3% had ideal pin placement, 17.8% medial/lateral breach, and 8.9% complete miss. Male patients were 3.7 times more likely to have ideal placement than females (p < 0.05). There was no relationship between BMI, SIJ fusion laterality, or pin placement laterality and ideal placement. Interobserver reliability was 0.72 and 0.70 in the first and second rounds, respectively, and defined as "substantial agreement." Intraobserver reliability ranged from 0.74 (substantial agreement) to 0.92 (almost perfect agreement).

CONCLUSIONS

Nonideal pin placement occurred in 26.7% of cases, but a true "miss" into the sacrum was rare. Ideal pin placement was more likely in males and was not associated with BMI, SIJ fusion laterality, or pin placement laterality. The grading scheme developed has high intraobserver and interobserver reliability, indicating that it is reproducible and can be used for future studies. When placing percutaneous pins, surgeons must be aware of factors that can decrease placement accuracy, regardless of location.