Pre-Operative Planning in Complex Deformities and Use of Patient-Specific UNiD™ Instrumentation

Deep Learning-Assisted Quantitative Measurement of Thoracolumbar Fracture Features on Lateral Radiographs

OBJECTIVE

This study aimed to develop and validate a deep learning (DL) algorithm for the quantitative measurement of thoracolumbar (TL) fracture features, and to evaluate its efficacy across varying levels of clinical expertise.

METHODS

Using the pretrained Mask Region-Based Convolutional Neural Networks model, originally developed for vertebral body segmentation and fracture detection, we fine-tuned the model and added a new module for measuring fracture metrics-compression rate (CR), Cobb angle (CA), Gardner angle (GA), and sagittal index (SI)-from lumbar spine lateral radiographs. These metrics were derived from six-point labeling by 3 radiologists, forming the ground truth (GT). Training utilized 1,000 nonfractured and 318 fractured radiographs, while validations employed 213 internal and 200 external fractured radiographs. The accuracy of the DL algorithm in quantifying fracture features was evaluated against GT using the intraclass correlation coefficient. Additionally, 4 readers with varying expertise levels, including trainees and an attending spine surgeon, performed measurements with and without DL assistance, and their results were compared to GT and the DL model.

RESULTS

The DL algorithm demonstrated good to excellent agreement with GT for CR, CA, GA, and SI in both internal (0.860, 0.944, 0.932, and 0.779, respectively) and external (0.836, 0.940, 0.916, and 0.815, respectively) validations. DL-assisted measurements significantly improved most measurement values, particularly for trainees.

CONCLUSION

The DL algorithm was validated as an accurate tool for quantifying TL fracture features using radiographs. DL-assisted measurement is expected to expedite the diagnostic process and enhance reliability, particularly benefiting less experienced clinicians.

A Standardized Curriculum Improves Trainee Rod Bending Proficiency in Spinal Deformity Surgery. Results of a Prospective Randomized Controlled Educational Study

INTRODUCTION

Surgical simulation is increasingly being accepted as a training platform to promote skill development and a safe surgical technique. Preliminary investigations in spine surgery show that simulation paired with educational intervention can markedly improve trainee performance. This study used a newly developed thoracolumbar fusion rod bending model to assess the effect of a novel educational curriculum and simulator training on surgical trainee rod bending speed and proficiency.

METHODS

Junior (PGY1 to 2) and senior (PGY3-fellow) surgical trainees at a single academic institution were prospectively enrolled in a rod bending simulation using a T7-pelvis spinal fusion model. Participants completed two simulations, with 1 month between first and second attempts. Fifty percent of surgeons in each training level were randomized to receive an educational curriculum (rod bending technique videos and unlimited simulator practice) between simulation attempts. Rod bending simulation proficiency was determined by the percentage of participants who completed the task (conclusion at 20 minutes), time to task completion or conclusion, and number of incomplete set screws at task conclusion. Participants completed a preparticipation and postparticipation survey. Univariate analysis compared rod bending proficiency and survey results between education and control cohorts.

RESULTS

Forty trainees (20 junior and 20 senior) were enrolled, with 20 participants randomized to the education and control cohorts. No notable differences were observed in the first simulation rod bending proficiency or preparticipation survey results between the education and control cohorts. In the second simulation, the education versus the control cohort demonstrated a significantly higher completion rate ( P = 0.01), shorter task time ( P = 0.009), fewer incomplete screws ( P = 0.003), and greater experience level ( P = 0.008) and comfort level ( P = 0.002) on postparticipation survey.

DISCUSSION

Trainees who participated in a novel educational curriculum and simulator training relative to the control cohort improved markedly in rod bending proficiency and comfort level. Rod bending simulation could be incorporated in existing residency and fellowship surgical skills curricula.

LEVEL OF EVIDENCE

I.

Patient-specific spinal rods in adult spinal deformity surgery reduce proximal junctional failure: a review of patient outcomes and surgical technique in a prospective observational cohort

Background

Spinal rods used for adult spinal deformity (ASD) correction are usually manufactured straight and bent manually during surgery. Pre-bent patient-specific spinal rods (PSSR) developed with software provide the surgeon with an intraoperative deformity correction consistent with the surgical plan. Our aim was to report clinical and radiological outcomes using PSSR. We investigated rates of junctional complications both proximally [kyphosis (PJK) and failure (PJF)] and distally [failure (DJF)].

Methods

Prospective case series of 20 consecutive patients who underwent ≥4 level ASD surgery with PSSR at a single institution between January 2019 and December 2022. Preoperative, 6-week, 6-month, 12-month, 24-month, and final follow-ups assessed patient satisfaction (Ottawa decision regret questionnaire) and patient reported outcome measures (PROMs) [visual analogue scale (VAS; Back/Leg), Oswestry disability index (ODI), and 12-Item Short Form Survey (SF-12)]. Sagittal spinopelvic parameters [sagittal vertical axis (SVA), pelvic tilt (PT), and pelvic incidence and lumbar lordosis mismatch (PI-LL)] measured by serial EOS scans were performed preoperatively then compared to planned correction and postoperative measures. Interoperative cages (narrow/wide) were placed for interbody support. PJK risk score assessed likelihood of developing kyphosis. Serial computed tomography (CT) imaging assessed complication (fusion/subsidence).

Results

The mean age of the patients (75% female) was 71.9±6.9 years, and the mean follow-up was 25.2±8.6 [7-40] months. Preoperative mean PROMs showed statistically significant overall improvement (P<0.001) postoperatively to final follow-up. Four patients without wide footprint cages at L4/5 or L5/S1, suffered DJF and reported regret undergoing surgery. Statistically significant difference (P<0.001) between preoperative and surgical plan in SVA and PI-LL but not in PT (P=0.058). No statistically significant difference in surgical plan versus the postoperative SVA, PI-LL, and PT (due to difficulty achieving the surgical plan, and also to maintaining the correction). One patient suffered PJF. There was a mean proximal kyphotic angle (PKA) of 17.8±13.0 degrees and PJK risk score of 3.7±1.0 with 40% who experienced PJK. No rod breakages were observed.

Conclusions

In this series, PSSR improved PROMs and treated ASD. Sagittal parameters planned preoperatively correlated with postoperative correction. PJF was reduced, compared to the literature (35%), but PJK was observed over time. DJF occurred and was related to the absence of interbody cages at the lumbosacral junction and decisional regret.

Rod angulation does not reflect sagittal curvature in adult spinal deformity surgery: comparison of lumbar lordosis and rod contouring

STUDY DESIGN

A retrospective study.

OBJECTIVE

Relationship between rod and spinal shape in the sagittal plane in adult spinal deformity (ASD) surgery.

BACKGROUND

Corrective surgery for adult spinal deformity (ASD) involves the use of contoured rods to correct and modify the spinal curvatures. Adequate rod bending is crucial for achieving optimal correction. The correlation between rods and spinal shape in long constructs has not been reported previously.

METHODS

We conducted a retrospective analysis of a prospective, multicenter database of patients who underwent surgery for ASD. The inclusion criteria were patients who underwent pelvic fixation and had an upper instrumented vertebra at or above T12. Pre- and post-operative standing radiographs were used to assess lumbar lordosis at the L4S1 and L1S1 levels. The angle between the tangents to the rod at the L1, L4, and S1 pedicles was calculated to determine the L4S1 and L1S1 rod lordosis. The difference between the lumbar lordosis (LL) and the rod lordosis (RL) was calculated as ΔL = LL-RL. The correlation between this difference (ΔL) and various characteristics was analyzed using descriptive and statistical methods.

RESULTS

Eighty-three patients were included in the study, resulting in 166 analyzed differences (ΔL) between the rod and spinal lordosis. The values for rod lordosis were found to be both greater and lesser than those of the spine but were mostly lower. The range for total ΔL was -24 °-30.9 °, with a mean absolute ΔL of 7.8 ° for L1S1 (standard deviation (SD) = 6.0) and 9.1 ° for L4S1 (SD = 6.8). In 46% of patients, both rods had a ΔL of over 5 °, and over 60% had at least one rod with a ΔL difference of over 5 °. Factors found to be related to a higher ΔL included postoperative higher lumbar lordosis, presence of osteotomies, higher corrected degrees, older age, and thinner rods. Multivariate analysis correlated only higher postoperative L1S1 lordosis with higher ΔL. No correlation was found between a higher ΔL and sagittal imbalance.

CONCLUSIONS

Variations between spinal and rod curvatures were observed despite the linear regression correlation. The shape of the rod does not seem to be predictive of the shape of the spine in the sagittal plane in ASD long-construct surgeries. Several factors, other than rod contouring, are involved in explaining the postoperative shape of the spine. The observed variation calls into question the fundamentals of the ideal rod concept.

Personalized Medicine in Orthopaedic Surgery: The Case of Spine Surgery

Personalized medicine has made a tremendous impact on patient care. Although initially, it revolutionized pharmaceutical development and targeted therapies in oncology, it has also made an important impact in orthopaedic surgery. The field of spine surgery highlights the effect of personalized medicine because the improved understanding of spinal pathologies and technological innovations has made personalized medicine a key component of patient care. There is evidence for several of these advancements to support their usage in improving patient care. Proper understanding of normative spinal alignment and surgical planning software has enabled surgeons to predict postoperative alignment accurately. Furthermore, 3D printing technologies have demonstrated the ability to improve pedicle screw placement accuracy compared with free-hand techniques. Patient-specific, precontoured rods have shown improved biomechanical properties, which reduces the risk of postoperative rod fractures. Moreover, approaches such as multidisciplinary evaluations tailored to specific patient needs have demonstrated the ability to decrease complications. Personalized medicine has shown the ability to improve care in all phases of surgical management, and several of these approaches are now readily available to orthopaedic surgeons.

Clinical Validation of a Novel Musculoskeletal Modeling Framework to Predict Postoperative Sagittal Alignment

STUDY DESIGN

A retrospective radiographic and biomechanical analysis of 108 thoracolumbar fusion patients from two clinical centers.

OBJECTIVE

This study aimed to determine the validity of a computational framework for predicting postoperative patient posture based on preoperative imaging and surgical data in a large clinical sample.

SUMMARY OF BACKGROUND DATA

Short-term and long-term studies on thoracolumbar fusion patients have discussed that a preoperative predictive model would benefit surgical planning and improve patient outcomes. Clinical studies have shown that postoperative alignment changes at the pelvis and intact spine levels may negatively affect postural balance and quality of life. However, it remains challenging to predict such changes preoperatively because of confounding surgical and patient factors.

MATERIALS AND METHODS

Patient-specific musculoskeletal models incorporated weight, height, body mass index, age, pathology-associated muscle strength, preoperative sagittal alignment, and surgical treatment details. The sagittal alignment parameters predicted by the simulations were compared with those observed radiographically at a minimum of three months after surgery.

RESULTS

Pearson correlation coefficients ranged from r=0.86 to 0.95, and mean errors ranged from 4.1° to 5.6°. The predictive accuracies for postoperative spinopelvic malalignment (pelvic incidence minus lumbar lordosis>10°) and sagittal imbalance parameters (TPA>14°, T9PA>7.4°, or LPA>7.2°) were between 81% and 94%. Patients treated with long fusion (greater than five segments) had relatively lower prediction errors for lumbar lordosis and spinopelvic mismatch than those in the local and short groups.

CONCLUSIONS

The overall model performance with long constructs was superior to those of the local (one to two segments) and short (three to four segments) fusion cases. The clinical framework is a promising tool in development to enhance clinical judgment and to help design treatment strategies for predictable surgical outcomes.

LEVEL OF EVIDENCE

3.