BackgroundPosterior lumbar interbody fusion (PLIF) with pedicle screw fixation is a standard treatment for degenerative spinal disorders. However, rigid fixation may accelerate adjacent segment degeneration (ASD). Cortical bone trajectory (CBT) fixation has been proposed as an alternative to traditional trajectory (TT) fixation in lumbar fusion surgery. However, the biomechanical characteristics of CBT, particularly in comparison with hybrid CBT-TT fixation, remain incompletely understood.MethodsA validated L4-L5 finite element model was developed to compare CBT, TT, and hybrid CBT-TT fixation constructs. Two simulation phases were performed. The immediate postoperative model was used to evaluate fusion-site micromotion and strain distribution, whereas the post-fusion model was used to assess adjacent-segment range of motion (ROM), intradiscal pressure (IDP), and facet joint force.ResultsAll fixation constructs provided sufficient segmental stability. Compared with TT fixation, CBT fixation produced greater but controlled fusion-site micromotion and a broader strain-stimulated region. The hybrid CBT-TT construct demonstrated intermediate biomechanical characteristics between CBT and TT fixation. Following simulated solid fusion, CBT and hybrid CBT-TT fixation reduced superior adjacent-segment ROM, IDP, and facet joint forces relative to TT fixation, indicating lower adjacent-segment biomechanical loading.ConclusionCBT fixation modified fusion-site mechanical conditions while maintaining overall construct stability. Both CBT and hybrid CBT-TT fixation reduced adjacent-segment biomechanical loading compared with TT fixation. These findings represent comparative biomechanical observations under simulated conditions and should not be interpreted as direct evidence of enhanced osseous fusion or prevention of ASD.

Thoracic spinal stenosis (TSS) is a severe spinal disorder that can lead to thoracic myelopathy, particularly in non-ambulatory patients. Thoracic ossification of the posterior longitudinal ligament (T-OPLL) and thoracic ossification of the ligamentum flavum (T-OLF) are the primary etiologies, but their differences in clinical characteristics and surgical outcomes remain unclear. This study aimed to compare the clinical features and surgical efficacy between non-ambulatory T-OPLL and T-OLF patients. A retrospective analysis was conducted on 126 non-ambulatory patients (70 in T-OPLL group, 56 in T-OLF group) who underwent decompression surgery between 2012 and 2023. Demographic data, surgical details, perioperative complications, and neurological outcomes were compared between groups. T-OPLL patients were younger (51.3 vs. 58.2 years, P < 0.001) and more frequently female (74.3

Study Design Retrospective cohort study.Objective To describe the clinical characteristics and surgical outcomes of non-ambulatory patients with thoracic ossification of the posterior longitudinal ligament (T-OPLL), and to identify predictors of independent walking recovery.Methods This retrospective study analyzed 70 non-ambulatory T-OPLL patients treated with either circumferential decompression (CD) or posterior decompression with fusion (PDF) surgery at a single center over 10 years (2012-2022). Outcomes included neurological recovery rate, independent walking recovery, and complications.Results Overall, 71.4% of patients regained independent walking. The CD group showed superior functional outcomes (90% vs 64% walking recovery, P = 0.030) but higher complication rates (70% vs 38% CSF leakage, P = 0.015). Shorter disease duration (P = 0.018) and lower BMI (P = 0.027) independently predicted better walking recovery.Conclusions For non-ambulatory T-OPLL patients, CD surgery provides better functional recovery while PDF surgery offers a safer alternative. Early surgical intervention and individualized approach selection based on disease duration and BMI are critical for optimizing outcomes.

Study DesignRetrospective Cohort Study.ObjectivesTo evaluate the prognostic value of preoperative increased signal intensity (ISI) grade on T2-weighted magnetic resonance imaging (MRI) and to identify risk factors associated with poor neurological recovery after surgery in adult cervical spinal cord injury without radiographic evidence of trauma (SCIWORET).MethodsA total of 128 consecutive SCIWORET patients who underwent surgical treatment between January 2016 and June 2023 were retrospectively analyzed. Patients were classified into 3 groups (Grade 0, 1, and 2) according to ISI grade on preoperative MRI. Neurological function was assessed using the Japanese Orthopaedic Association (JOA) score and recovery rate (RR). Multivariate logistic regression analysis identified predictors of poor RR (< 50%), and receiver operating characteristic (ROC) analysis determined the optimal age cutoff for prognosis.ResultsSignal intensity alterations were observed in 111 patients (86.7%). Higher ISI grades correlated with lower preoperative JOA scores (r = -0.303, P < 0.001) but not with postoperative RR (r = -0.067, P = 0.450). Multivariate analysis identified age as the only independent predictor of poor RR (OR = 1.10, 95% CI: 1.04-1.16, P < 0.001). ROC curve analysis yielded an optimal age cutoff of 58.5 years.ConclusionsPreoperative ISI grade reflected the severity of spinal cord injury but did not predict postoperative neurological recovery. Older age (> 58.5 years) independently predicted poorer outcomes, highlighting the importance of integrating patient age into prognostic counseling and individualized surgical decision-making in SCIWORET.

Retrospective Cohort Study. Chondrosarcoma of mobile spine is a rare aggressive malignant tumor and postsurgical local recurrence rates remain high. En bloc resection is currently the preferred treatment. Resection that achieves tumor-free margin removal of the tumor may enable more complete removal of tumor tissue but significantly increases the complexity and risk of surgery and results in more postoperative complications. We sought to compare surgical outcomes, complications, and prognoses between patients who underwent en bloc resection with and without intralesional removal of the tumor. We reviewed 56 patients with spinal chondrosarcoma who underwent en bloc tumor resection and reconstructive surgery at our center between 2000 and 2024 with a minimum postoperative follow-up of 1 year. We collected and analyzed data regarding surgical procedures, complication characteristics, and local tumor control and recurrence. We included 56 patients. Of these, 36 patients underwent the first surgery, and 20 experienced recurrences. All patients underwent en bloc tumor resection; 36 and 20 underwent intralesional and tumor-free margin resections, respectively. We recorded 83 complications; the incidence and the number of major complications were significantly higher in the tumor-free margin surgery group. Thirty patients experienced tumor recurrence and 26 patients died. Tumor-free margin en bloc resection and conventional-type chondrosarcoma were predictive factors for reduced long-term postoperative recurrence and mortality risk. Tumor-free margin resection carries higher risks and is associated with a greater number of perioperative complications, but reduces the risk of local tumor recurrence and prolongs recurrence-free survival and overall survival, providing patients with better prognoses.

Study DesignRetrospective Cohort Study.ObjectivesTherapeutic strategies for Aneurysmal Bone Cysts (ABCs) of the spine remain controversial and encompass several modalities, including open surgery (ie, intralesional curettage and/or en bloc resection), localized injections, and pharmacotherapy. This study was designed to retrospectively analyze the clinical data from patients treated at our institution, with the objective of evaluating the clinical outcomes, safety profile, and efficacy of different treatments including open surgery, percutaneous injections of a doxycycline-albumin suspension, and Denosumab therapy in the management of these lesions, thereby establishing a basis for a comprehensive treatment algorithm.MethodsFrom January 2010 to December 2024, 27 patients who had no/minor neurological deficits (modified Frankel scale D or E) were included in the study, of whom 6 were treated with open surgery (surgery group), 14 were treated with percutaneous doxycycline/albumin injection (injection group) and 7 were treated with Denosumab (Denosumab group). The demographic and clinical information of these groups were recorded and compared.ResultsIn the surgery cohort (n = 6), complete neurological recovery was achieved in 5 patients (83.3%), while 1 patient experienced residual minor paresthesia; their mean Visual Analogue Scale (VAS) score decreased from 4.9 to 0.8. Two patients developed local recurrence over a mean follow-up period of 46.6 months (range, 15-113 months). In the injection cohort (n = 14), all patients demonstrated a significant reduction in lesion size and resumed normal daily activities. Complete symptomatic resolution was reported in 10 patients; 4 achieved partial resolution, with only mild, exertion-related local pain. Their mean VAS score decreased from 4.9 to 0.7. During a mean follow-up of 68.1 months (range, 7-117 months), no complications or recurrences were observed. In the Denosumab cohort (n = 7), all patients exhibited a significant reduction in lesion size and resumed normal activities. Complete symptomatic resolution was achieved in 6 patients, while 1 experienced partial resolution, characterized by intermittent, mild upper extremity weakness. Their mean VAS score decreased from 4.8 to 0.24. No complications or recurrences were documented over a median follow-up of 9.5 months (range, 1-88 months). No significant differences were observed among the cohorts regarding the rates of recurrence (P = 0.227) or complications (P = 0.304).ConclusionsFor patients presenting with spinal ABCs characterized by an absence of severe neurological deficits, both intralesional injection and systemic drug therapy are effective minimally invasive treatment options and can be considered as first-line choices. However, while both approaches demonstrate favorable short-term efficacy, their long-term outcomes require further investigation. Although not the preferred initial treatment, open surgery remains a crucial option for rapidly correcting spinal instability and reversing neurological deficits.

Study designRetrospective cross-sectional study.ObjectivesGiven the aggressive nature of primary spinal tumors, postsurgical local recurrence rates remain high. En bloc resection is currently the preferred treatment. However, the presence of a large thoracic cavity mass increases the surgical difficulty, risk, and likelihood of extensive complications. We report diagnostic and therapeutic characteristics, surgical strategies, and perioperative complications of such tumors treated with en bloc resection.MethodsWe reviewed 25 patients with primary spinal tumors and extensive thoracic cavity involvement who underwent en bloc resection at our center between 2012 and 2023 with a minimum postoperative follow-up of 1 year. We collected and analyzed data on surgical procedures, complication characteristics, and local tumor control and recurrence, and compared our findings with previous studies.ResultsWe included 25 patients (14 males and 11 females; mean age, 41.3 years). Of these, 14 patients underwent the first surgery, and 11 experienced recurrences. All patients underwent en bloc resection; 9 and 16 underwent intralesional and extralesional resections, respectively, 16 and 9 underwent posterior-only and combined approaches, respectively. The average surgery duration was 674 min, with an average estimated intraoperative blood loss of 2,388 mL. Eighty complications were recorded; 24 patients (96%) experienced at least one perioperative complication.ConclusionFor primary spinal tumors with huge thoracic cavity involvement, en bloc resection remains the optimal treatment for achieving local tumor control. Suitability for this procedure depends on the patient’s fitness for major surgery, the absence of distant metastases, and tumor resectability. Surgery can be performed via posterior-only or combined anteroposterior approaches.

OBJECTIVE: To describe a novel technique, posterior thoracic antidisplacement and fusion (PTAF), for a special type of ossification of the posterior longitudinal ligament in the thoracic spine (T-OPLL), and to evaluate its safety and efficacy. METHODS: From July to December 2020, 5 consecutive patients with beak-type T-OPLL located at the thoracic vertebral body level underwent PTAF surgery. Their demographic data, radiological parameters, perioperative complications, and surgery-related findings were recorded and analyzed. The surgical outcomes were assessed using a modified Japanese Orthopedic Association scale, and the recovery rate was calculated using the Hirabayashi's method. RESULTS: All patients were followed up for at least two years. The mean thickness of OPLL was 9.4 +/- 1.0 mm, and the OPLL spinal canal occupying ratio was 67.7% +/- 8.5%. Postoperatively, the mean antidisplacement distance of OPLL was 8.1 +/- 1.8 mm, and the average shortened distance of the spinal column was 6.0 +/- 1.13 mm. The mean operation time and blood loss were 158.2 +/- 26.3 minutes and 460 +/- 89.4 mL, respectively. Perioperative complications were cerebrospinal fluid leakage and instrument failure, 2 cases each. The mean modified Japanese Orthopedic Association score was increased from 3.6 +/- 2.9 before surgery to 9.4 +/- 3.0 at the last follow-up, and the average recovery rate was 84.2 +/- 30.5%. CONCLUSIONS: The preliminary clinical outcomes indicate that PTAF is a safe and effective method for the treatment of beak-type T-OPLL, which has its apex located at the vertebral body level and has a high spinal canal occupation ratio.

Study Design: Retrospective Cohort Study Objectives: Chondrosarcoma of mobile spine is a rare aggressive malignant tumor and postsurgical local recurrence rates remain high. En bloc resection is currently the preferred treatment. Resection that achieves non-intralesional removal of the tumor may enable more complete removal of tumor tissue but significantly increases the complexity and risk of surgery and results in more postoperative complications. We sought to compare surgical outcomes, complications, and prognoses between patients who underwent en bloc resection with and without intralesional removal of the tumor. Methods: We reviewed 56 patients with spinal chondrosarcoma who underwent en bloc tumor resection and reconstructive surgery at our center between 2000 and 2024 with a minimum postoperative follow-up of 1 year. We collected and analyzed data regarding surgical procedures, complication characteristics, and local tumor control and recurrence. Results: We included 56 patients. Of these, 36 patients underwent the first surgery, and 20 experienced recurrences. All patients underwent en bloc tumor resection; 36 and 20 underwent intralesional and non-intralesional resections, respectively. We recorded 83 complications; the incidence and the number of major complications were significantly higher in the non-intralesional surgery group. Thirty patients experienced tumor recurrence and 26 patients died. Non-intralesional en bloc resection and conventional-type chondrosarcoma were predictive factors for reduced long-term postoperative recurrence and mortality risk. Conclusions: Non-intralesional resection carries higher risks and is associated with a greater number of perioperative complications, but reduces the risk of local tumor recurrence and prolongs recurrence-free survival and overall survival, providing patients with better prognoses.

The deep learning method is an efficient solution for improving the quality of undersampled magnetic resonance (MR) image reconstruction while reducing lengthy data acquisition. Most deep learning methods neglect the mutual constraints between the real and imaginary components of complex-valued k-space data. In this paper, a new complex-valued convolutional neural network, namely, Dense-U-Dense Net (DUD-Net), is proposed to interpolate the undersampled k-space data and reconstruct MR images. The proposed network comprises dense layers, U-Net, and other dense layers in sequence. The dense layers are used to simulate the mutual constraints between real and imaginary components, and U-Net performs feature sparsity and interpolation estimation for k-space data. Two MRI datasets were used to evaluate the proposed method: brain magnitude-only MR images and knee complex-valued k-space data. Several operations were conducted for data preprocessing. First, the complex-valued MR images were synthesized by phase modulation on magnitude-only images. Second, a radial trajectory based on the golden angle was used for k-space under-sampling, whereby a reversible normalization method was proposed to balance the distribution of positive and negative values in k-space data. The optimal performance of DUD-Net was demonstrated based on a quantitative evaluation of inter-method and intra-method comparisons. When compared with other methods, significant improvements were achieved, PSNRs were increased by 10.78 and 5.74dB, whereas RMSEs were decreased by 71.53% and 30.31% for magnitude and phase image, respectively. It is concluded that DUD-Net significantly improves the performance of MR image reconstruction.

目的:探索肾透明细胞癌脊柱转移的手术治疗效果以及生存期相关因素.方法:回顾2008年1月～2019年12月于我科行手术治疗的肾透明细胞癌脊柱转移患者45例,其中男性34例,女性11例,平均年龄60.1±7.6岁,平均病程为4.3±4.4个月(1～24个月).术前Frankel分级E级25例,D级17例,C级1例,B级2例.术前伴有内脏转移10例,伴有其他骨转移16例.按手术方式分为肿瘤切除组和姑息减压组,肿瘤切除组包括肿瘤减灭术和全脊椎切除术.统计手术出血量及手术相关并发症,对所有患者进行随访,根据患者的生存状态分为终点组(已去世)和随访组(仍存活),记录随访组患者的疼痛缓解情况、生活自理能力、神经功能评估及随访前半年内的影像学复查结果.通过Kaplan-Meier法估算中位生存期并获得生存曲线.将可能影响患者生存期的因素进行单因素分析,将单因素分析中有统计学差异的因素纳入多因素COX回归分析.结果:随访时间均为12个月以上,末次随访时仍有30例患者存活(随访组),平均随访时间为36.7±31.6个月,手术后的中位生存期为81.0个月,术后1年生存率为84.4％,术后2年生存率为78.6％,术后3年生存率为70.2％.随访组患者术前和随访时的NRS评分平均值分别为6.0±2.4分(1～10分)和2.9±2.7分(0～8分),19例患者与术前相比疼痛明显缓解(NRS评分降低3分以上).术前和随访时的KPS评分平均值分别为71.3±14.3分(30～90分)和70.0±12.2分(40～90分).29例患者随访时的Frankel分级为D级以上,其中6例获得改善,14例保持不变.单因素分析中颈椎转移(P=0.040)、内脏转移(P=0.018)、术后放疗(P=0.027)和术后靶向治疗(P=0.018)对术后生存期的影响有统计学意义,肿瘤切除组的患者预后优于姑息减压组,多因素分析的结果无统计学意义.结论:手术后局部放疗和全身靶向治疗的多学科综合治疗可以提高肾透明细胞癌脊柱转移患者的生存期,手术治疗可以缓解疼痛,保护神经功能,从而提高生活质量.

Objectives To evaluate the survival and medium to long term health-related quality of life (HRQoL) of patients with primary cervical spinal tumors in a cross-sectional study and to identify any significant associations with demographic or clinical characteristics. Methods Patients diagnosed with primary cervical spinal tumors were retrospectively enrolled and their clinical, radiologic, and follow-up data (specifically the EQ-5D questionnaire) were collected. Univariate and multivariate Cox time-dependent regression analyses were performed to examine the significance of certain variables on overall survival. Univariate and multivariate logistic regression analyses were conducted to identify variables significant for overall HRQoL and each dimension of the EQ-5D. Results A total of 341 patients were enrolled in the study with a mean follow-up of 70 months. The diagnosis was benign in 246 cases, malignant in 84, and unconfirmed in 11. The 5-year overall survival rate was 86% and the 10-year overall survival rate was 65%. Multivariate analysis suggested that surgical treatment ( P = 0.002, hazard ratio [HR] = 0.431, 95% CI. [0.254, 0.729]), benign and malignant tumors [ P < 0.001, HR = 2.788, 95% CI. (1.721, 4.516)], tumor and surrounding normal tissue boundary [ P = 0.010, HR = 1.950, 95% CI. (1.171, 3.249)], and spinal instability [ P = 0.031, HR = 1.731, 95% CI. (1.051, 2.851)] still had significant effects on survival. Conclusions In this cross-sectional study, we evaluated the survival period and medium and long-term health-related quality of life of patients with primary tumors of the cervical spine, and analyzed the significant related factors of tumor clinical characteristics. Surgery, myelopathy, malignancy, spinal pain relieved by lying down or supine position, and tumor infiltration on MRI were significant predictors for overall survival. Enneking stage and age were significant predictors for HRQoL.

Abstract Purpose : We aimed to compare the characteristics of anti-loosening and the pullout force between cortical bone trajectory (CBT) and conventional trajectory (CT) screws under different bone qualities. Methods : Finite element models were established for CBT screws and CT screws in a single-segment L4 vertebra. Compression loads were applied to the models, followed by cyclic loading to simulate the screw loosening process resulting from nail channel expansion. The pullout force was then applied along the screw axis to simulate screw pullout, and pull-out force under different bone quality was calculated and compared. Results : In the loosening test, screw channel expansion was significantly smaller in the CBT group. The CT screw channel volume expanded by 20.34% and 20.60% in healthy and osteoporotic bone, whereas the CBT screw channel volume only expanded by 0.20% and 0.19%. The pullout force of CBT screws was higher than that of CT screws in both healthy and osteoporotic bone (CBT, 1457.51 N and 1040.65 N; CT, 761.68 N and 461.36 N). Conclusions : The loosening model combined with pullout testing can better simulate in vivo screw holding capacity. In different bone qualities, especially in osteoporotic bone, CBT screws have better anti-loosening and higher pullout strength than CT screws.

The oncogenetic function of inhibitory member of the apoptosis stimulating protein of p53 family (iASPP) in chordoma is unclear and remains to elucidate. The expression of iASPP in chordoma tissues and cells, its correlation to clinicopathological parameters and the effect on the patients' prognosis were evaluated. Cellular proliferation, invasion and cisplatin-response were observed after the iASPP knockdown or overexpression in vitro. Co-Immunoprecipitation assay was used to explore the interaction between iASPP and p53. The regulation of miRNA-124 on the expression and apoptotic function of iASPP was explored after transiently transfecting cells with miRNA-124 mimics or inhibitor. Results indicated that iASPP overexpressed in chordoma tissues and cells. Its overexpression was associated with tumor invasion and local recurrence, and was predictive of patients' poor prognosis. Cells with iASPP-silence showed a decreased ability of proliferation and invasion, but an increasing sensitivity to cisplatin. Besides, iASPP could combine with p53 in either endogenous or exogenous detection. Post-transcriptionally, miRNA-124 negatively regulated the expression of iASPP, which further led to the changes of apoptosis-related proteins. Thus, iASPP overexpression is associated with the clinical outcome in spinal chordoma and influences cellular proliferation, invasion, and the sensitivity to cisplatin.

BackgroundSpinal giant cell tumor (SGCT) is a relatively rare primary tumor. En bloc resection is the preferred surgical procedure for it due to its aggressiveness, meanwhile leading to more complications. We reported the characteristics of perioperative complications and local control of total tumor resection including en bloc resection and piecemeal resection for primary thoracic and lumbar spinal giant cell tumors in a single center over 10 years.MethodsThis is a retrospective cross-sectional and cohort study. Forty-one consecutive patients with SGCTs who underwent total tumor resection from 2010 to 2020 at our institution and were followed up for at least 24 months were reviewed. Surgery data, complication characteristics and local tumor control were collected and compared by different surgical procedure.ResultsForty-one patients were included, consisting of 18 males and 23 females, with a mean age of 34.2 years. Thirty-one had thoracic vertebra lesions, and 10 had lumbar vertebra lesions. Thirty-five patients were primary cases, and 6 patients were recurrent cases. Eighteen patients were treated by total en bloc spondylectomy (TES), 12 patients underwent en bloc resection according to WBB surgical system, and 11 patients underwent piecemeal resection. The average surgical time was 498 min, and the mean estimated blood loss was 2145 ml. A total of 58 complications were recorded, and 30 patients (73.2%) had at least one perioperative complication. All patients were followed up after surgery for at least 2 years. A total of 6 cases had postoperative internal fixation failure, and 4 cases presented local tumor recurrence (9.8%).ConclusionsAlthough the surgical technique is difficult and accompanied by a high rate of perioperative complications, en bloc resection can achieve favorable local control in SGCT. When it is too difficult to complete en bloc resection, thoroughly piecemeal resection without residual is also acceptable, given the relatively low recurrence rate.

Abstract Purpose: We aimed to compare the characteristics of anti-loosening and the pullout force between cortical bone trajectory (CBT) and conventional trajectory (CT) screws under different bone qualities. Methods: Finite element models were established for CBT screws and CT screws in a single-segment L4 vertebra. Compression loads were applied to the models, followed by cyclic loading to simulate the screw loosening process resulting from nail channel expansion. The pullout force was then applied along the screw axis to simulate screw pullout, and pull-out force under different bone quality was calculated and compared. Results: In the loosening test, screw channel expansion was significantly smaller in the CBT group. The CT screw channel volume expanded by 20.34% and 20.60% in healthy and osteoporotic bone, whereas the CBT screw channel volume only expanded by 0.20% and 0.19%. The pullout force of CBT screws was higher than that of CT screws in both healthy and osteoporotic bone (CBT, 1457.51 N and 1040.65 N; CT, 761.68 N and 461.36 N). Conclusions: The loosening model combined with pullout testing can better simulate in vivo screw holding capacity. In different bone qualities, especially in osteoporotic bone, CBT screws have better anti-loosening and higher pullout strength than CT screws.

To compare total en bloc spondylectomy with marginal margins against piecemeal spondylectomy with intralesional margins in the surgical treatment of Enneking stage III spinal giant cell tumor (GCT) in terms of local recurrence. A retrospective survival analysis of patients with Enneking stage III GCT who underwent TES with marginal margins or total piecemeal spondylectomy with intralesional margins was performed between January 2006 and April 2020. Local recurrence-free survival (LRFS) was the time between the date of surgery and recurrence. Factors with p-values < 0.05 in the univariate analysis were included in the multivariate analysis using proportional hazard analysis. Sixty patients (25 men and 35 women) with a mean age of 35.6 (range 11–71) years were included. The mean follow-up duration was 93 (range 24–198) months. Two patients were lost to follow-up 6 and 14 years after the procedure. Over a 10-year period, the recurrence rate was 13.3

BackgroundSpinal tuberculosis (TB) has the highest incidence in remote plateau areas, particularly in Tibet, China, due to inadequate local healthcare services, which not only facilitates the transmission of TB bacteria but also increases the burden on grassroots hospitals. Computer-aided diagnosis (CAD) is urgently required to improve the efficiency of clinical diagnosis of TB using computed tomography (CT) images. However, classical machine learning with handcrafted features generally has low accuracy, and deep learning with self-extracting features relies heavily on the size of medical datasets. Therefore, CAD, which effectively fuses multimodal features, is an alternative solution for spinal TB detection. MethodsA new deep learning method is proposed that fuses four elaborate image features, specifically three handcrafted features and one convolutional neural network (CNN) feature. Spinal TB CT images were collected from 197 patients with spinal TB, from 2013 to 2020, in the People's Hospital of Tibet Autonomous Region, China; 3,000 effective lumbar spine CT images were randomly screened to our dataset, from which two sets of 1,500 images each were classified as tuberculosis (positive) and health (negative). In addition, virtual data augmentation is proposed to enlarge the handcrafted features of the TB dataset. Essentially, the proposed multimodal feature fusion CNN consists of four main sections: matching network, backbone (ResNet-18/50, VGG-11/16, DenseNet-121/161), fallen network, and gated information fusion network. Detailed performance analyses were conducted based on the multimodal features, proposed augmentation, model stability, and model-focused heatmap. ResultsExperimental results showed that the proposed model with VGG-11 and virtual data augmentation exhibited optimal performance in terms of accuracy, specificity, sensitivity, and area under curve. In addition, an inverse relationship existed between the model size and test accuracy. The model-focused heatmap also shifted from the irrelevant region to the bone destruction caused by TB. ConclusionThe proposed augmentation effectively simulated the real data distribution in the feature space. More importantly, all the evaluation metrics and analyses demonstrated that the proposed deep learning model exhibits efficient feature fusion for multimodal features. Our study provides a profound insight into the preliminary auxiliary diagnosis of spinal TB from CT images applicable to the Tibetan area.

Background Moyamoya disease (MMD) is a rare cerebrovascular occlusive disease with progressive stenosis of the terminal portion of internal cerebral artery (ICA) and its main branches, which can cause complications, such as high risks of disability and increased mortality. Accurate and timely diagnosis may be difficult for physicians who are unfamiliar to MMD. Therefore, this study aims to achieve a preoperative deep-learning-based evaluation of MMD by detecting steno-occlusive changes in the middle cerebral artery or distal ICA areas. Methods A fine-tuned deep learning model was developed using a three-dimensional (3D) coordinate attention residual network (3D CA-ResNet). This study enrolled 50 preoperative patients with MMD and 50 controls, and the corresponding time of flight magnetic resonance angiography (TOF-MRA) imaging data were acquired. The 3D CA-ResNet was trained based on sub-volumes and tested using patch-based and subject-based methods. The performance of the 3D CA-ResNet, as evaluated by the area under the curve (AUC) of receiving-operator characteristic, was compared with that of three other conventional 3D networks. Results With the resulting network, the patch-based test achieved an AUC value of 0.94 for the 3D CA-ResNet in 480 patches from 10 test patients and 10 test controls, which is significantly higher than the results of the others. The 3D CA-ResNet correctly classified the MMD patients and normal healthy controls, and the vascular lesion distribution in subjects with the disease was investigated by generating a stenosis probability map and 3D vascular structure segmentation. Conclusions The results demonstrated the reliability of the proposed 3D CA-ResNet in detecting stenotic areas on TOF-MRA imaging, and it outperformed three other models in identifying vascular steno-occlusive changes in patients with MMD.

Study design Retrospective cohort study. Objective To identify risk factors and predictive models for proximal junctional kyphosis (PJK) in a long-term follow-up of patients with adult degenerative scoliosis (ADS) following posterior corrective surgeries. Materials and Methods A consecutive 113 ADS patients undergoing posterior corrective surgery between January 2008 and April 2019 with minimum 2-year follow-up were included. All patients underwent preoperative, postoperative, and final follow-up by X-ray imaging. Multivariate logistic analysis was performed on various risk factors and radiological predictor models. Results PJK was identified radiographically in 46.9% of patients. Potential risk factors for PJK included postoperative thoracic kyphosis (TK) (P < .05), final follow-up Pelvic Tilt (PT) (P < .05), PT changes at final follow-up (P < .05), age over 55 years old at the surgery (P < .05), theoretical thoracic kyphosis–actual thoracic kyphosis mismatch (TK mismatch) (P < .05) and theoretical lumbar lordosis–acutal lumbar lordosis mismatch (LL mismatch) (P < .05). As for the predictive models, PJK was predictive by the following indicators: preoperative global sagittal alignment ≥45° (Model 1), postoperative pelvic incidence–lumbar lordosis mismatch (PI–LL)≤10° and postoperative PI–LL overcorrection (Model 2), and TK+LL≥0° (Model 3) (P < .05). Postoperative TK mismatch (OR = 1.064) was independent as risk factors for PJK, with the cut-off values respectively set at −28.56° to predict occurrence of PJK. Conclusion The risk of radiographic PJK increases with an age over 55 years old and higher postoperative TK. In addition, postoperative TK mismatch is an independent risk factor for developing PJK. All three predictive models could effectively indicate the occurrence of PJK.