Katsos, Konstantinos D.

OBJECTIVE This study aimed to develop an artificial intelligence (AI) model for automatically detecting Hounsfield unit (HU) values at the L1 vertebra in preoperative thoracolumbar CT scans. This model serves as a screening tool for osteoporosis in patients undergoing spine surgery, offering an alternative to traditional bone mineral density measurement methods like dual-energy x-ray absorptiometry. METHODS The authors utilized two CT scan datasets, comprising 501 images, which were split into training, validation, and test subsets. The nnU-Net framework was used for segmentation, followed by an algorithm to calculate HU values from the L1 vertebra. The model’s performance was validated against manual HU calculations by expert raters on 56 CT scans. Statistical measures included the Dice coefficient, Pearson correlation coefficient, intraclass correlation coefficient (ICC), and Bland-Altman plots to assess the agreement between AI and human-derived HU measurements. RESULTS The AI model achieved a high Dice coefficient of 0.91 for vertebral segmentation. The Pearson correlation coefficient between AI-derived HU and human-derived HU values was 0.96, indicating strong agreement. ICC values for interrater reliability were 0.95 and 0.94 for raters 1 and 2, respectively. The mean difference between AI and human HU values was 7.0 HU, with limits of agreement ranging from −21.1 to 35.2 HU. A paired t-test showed no significant difference between AI and human measurements (p = 0.21). CONCLUSIONS The AI model demonstrated strong agreement with human experts in measuring HU values, validating its potential as a reliable tool for automated osteoporosis screening in spine surgery patients. This approach can enhance preoperative risk assessment and perioperative bone health optimization. Future research should focus on external validation and inclusion of diverse patient demographics to ensure broader applicability.

OBJECTIVE The aims of this study were to 1) define the incidence of transforaminal lumbar interbody fusion (TLIF) interbody subsidence; 2) determine the relative importance of preoperative and intraoperative patient- and instrumentation-specific risk factors predictive of postoperative subsidence using CT-based assessment; and 3) determine the impact of TLIF subsidence on postoperative complications and fusion rates. METHODS All adult patients who underwent one- or two-level TLIF for lumbar degenerative conditions at a multi-institutional academic center between 2017 and 2019 were retrospectively identified. Patients with traumatic injury, infection, malignancy, previous fusion at the index level, combined anterior-posterior procedures, surgery with greater than two TLIF levels, or incomplete follow-up were excluded. Interbody subsidence at the superior and inferior endplates of each TLIF level was directly measured on the endplate-facing surface of both coronal and sagittal CT scans obtained greater than 6 months postoperatively. Patients were grouped based on the maximum subsidence at each operative level classified as mild, moderate, or severe based on previously documented < 2-mm, 2- to 4-mm, and ≥ 4-mm thresholds, respectively. Univariate and regression analyses compared patient demographics, medical comorbidities, preoperative bone quality, surgical factors including interbody cage parameters, and fusion and complication rates across subsidence groups. RESULTS A total of 67 patients with 85 unique fusion levels met the inclusion and exclusion criteria. Overall, 28% of levels exhibited moderate subsidence and 35% showed severe subsidence after TLIF with no significant difference in the superior and inferior endplate subsidence. Moderate (≥ 2-mm) and severe (≥ 4-mm) subsidence were significantly associated with decreases in cage surface area and Taillard index as well as interbody cages with polyetheretherketone (PEEK) material and sawtooth surface geometry. Severe subsidence was also significantly associated with taller preoperative disc spaces, decreased vertebral Hounsfield units (HU), the absence of bone morphogenetic protein (BMP) use, and smooth cage surfaces. Regression analysis revealed decreases in Taillard index, cage surface area, and HU, and the absence of BMP use predicted subsidence. Severe subsidence was found to be a predictor of pseudarthrosis but was not significantly associated with revision surgery. CONCLUSIONS Patient-level risk factors for TLIF subsidence included decreased HU and increased preoperative disc height. Intraoperative risk factors for TLIF subsidence were decreased cage surface area, PEEK cage material, bullet cages, posterior cage positioning, smooth cage surfaces, and sawtooth surface designs. Severe subsidence predicted TLIF pseudarthrosis; however, the causality of this relationship remains unclear.

Spinal cord tumors constitute a diverse group of rare neoplasms associated with significant mortality and morbidity that pose unique clinical and surgical challenges. Diagnostic accuracy and outcome prediction are critical for informed decision making and can promote personalized medicine and facilitate optimal patient management. Machine learning has the ability to analyze and combine vast amounts of data, allowing the identification of patterns and the establishment of clinical associations, which can ultimately enhance patient care. Although artificial intelligence techniques have been explored in other areas of spine surgery, such as spinal deformity surgery, precise machine learning models for spinal tumors are lagging behind. Current applications of machine learning in spinal cord tumors include algorithms that improve diagnostic precision by predicting genetic, molecular, and histopathological profiles. Furthermore, artificial intelligence-based systems can assist surgeons with preoperative planning and surgical resection, potentially reducing the risk of recurrence and consequently improving clinical outcomes. Machine learning algorithms promote personalized medicine by enabling prognostication and risk stratification based on accurate predictions of treatment response, survival, and postoperative complications. Despite their promising potential, machine learning models require extensive validation processes and quality assessments to ensure safe and effective translation to clinical practice.