Cervical spondylosis is one of the most common degenerative diseases, seriously affecting life quality. Unlike diseases with explicit lesions like cancer, hydroncus, or fracture, the degeneration of the cervical spine cannot be explicitly detected from the appearance of medical images, requiring extensive experience of doctors to interpret subtle clues. However, the extremely high incidence of cervical spondylosis coincides with a serious shortage of experienced doctors and uneven distribution of medical resources, hindering early diagnosis. We propose a cascade-ensemble deep learning framework for cervical spondylosis diagnosis. The framework integrates vertebral body detection and degenerative diagnosis through a cascading architecture, and jointly trains an ensemble of degenerative indicators in a multi-task learning manner. We demonstrate that deep learning models are more sensitive to distance and position based indicators than angle based ones. In intervertebral stenosis analysis, our method achieves comparable performance to senior radiologists and clinicians, with much faster diagnostic speed.

To assess the potential of fast field echo resembling a CT using restricted echo-spacing (FRACTURE) sequence to enhance conventional MRI in detecting bone abnormalities of cervical spondylosis. 137 consecutive patients with cervical spondylosis who underwent clinically indicated paired CT and MRI within 2 weeks between January and June 2024. After routine MRI, the 3D-FRACTURE sequences were performed. Three radiologists independently evaluated the data during three sessions: (1) CT with consensus, (2) routine MRI, and (3) FRACTURE, with a 4-week interval between sessions. Assessments included osteophytes, bony foraminal stenosis, posterior longitudinal ligament ossification (OPLL), their anatomical location, and diagnostic confidence, using CT as the reference standard. Inter- and intra-reader reproducibility was assessed using multi-rater Fleiss κ and the intraclass correlation coefficient (ICC), respectively. The non-inferiority assessment compared routine MRI/FRACTURE and CT diagnoses using a relative reduction margin of 0.5. The study sample comprised 82 males and 55 females (age 56.9 ± 9.8 years). ICC indicated good to excellent inter-rater reliability for FRACTURE (osteophytes: ICC, 0.83–1.00; OPLL: ICC, 0.73–0.92; bony foraminal stenosis: ICC, 0.76–0.98), which was superior to conventional MRI (most ICC values < 0.7). The diagnostic confidence by FRACTURE sequences was significantly higher than by routine MRI (p < 0.001). Non-inferiority analysis demonstrated that FRACTURE and CT detection were similar for osteophyte, bony foraminal stenosis, and OPLL within a margin of 0.5. The FRACTURE sequence demonstrated comparable performance to CT in bone abnormalities detection in cervical spondylosis, superior to the routine MRI protocol. The FRACTURE sequence addresses the limitations of conventional MRI in evaluating bone abnormalities, potentially minimizing radiation exposure and streamlining the diagnostic process for patients.

BACKGROUND:Endometrial cancer (EC) is a common gynecologic malignancy; accurate assessment of key prognostic factors is important for treatment planning. PURPOSE:To develop a deep learning (DL) framework based on biparametric MRI for automated segmentation and multitask classification of EC key prognostic factors, including grade, stage, histological subtype, lymphovascular space invasion (LVSI), and deep myometrial invasion (DMI). STUDY TYPE:Retrospective. SUBJECTS:A total of 325 patients with histologically confirmed EC were included: 211 training, 54 validation, and 60 test cases. FIELD STRENGTH/SEQUENCE:T2-weighted imaging (T2WI, FSE/TSE) and diffusion-weighted imaging (DWI, SS-EPI) sequences at 1.5 and 3 T. ASSESSMENT:The DL model comprised tumor segmentation and multitask classification. Manual delineation on T2WI and DWI acted as the reference standard for segmentation. Separate models were trained using T2WI alone, DWI alone and combined T2WI + DWI to classify dichotomized key prognostic factors. Performance was assessed in validation and test cohorts. For DMI, the combined model's was compared with visual assessment by four radiologists (with 1, 4, 7, and 20 years' experience), each of whom independently reviewed all cases. STATISTICAL TESTS:Segmentation was evaluated using the dice similarity coefficient (DSC), Jaccard similarity coefficient (JSC), Hausdorff distance (HD95), and average surface distance (ASD). Classification performance was assessed using area under the receiver operating characteristic curve (AUC). Model AUCs were compared using DeLong's test. p < 0.05 was considered significant. RESULTS:In the test cohort, DSCs were 0.80 (T2WI) and 0.78 (DWI) and JSCs were 0.69 for both. HD95 and ASD were 7.02/1.71 mm (T2WI) versus 10.58/2.13 mm (DWI). The classification framework achieved AUCs of 0.78-0.94 (validation) and 0.74-0.94 (test). For DMI, the combined model performed comparably to radiologists (p = 0.07-0.84). CONCLUSIONS:The unified DL framework demonstrates strong EC segmentation and classification performance, with high accuracy across multiple tasks. EVIDENCE LEVEL:3. TECHNICAL EFFICACY:Stage 3.

Rationale and Objectives To generate virtual T1 contrast-enhanced (T1CE) sequences from plain spinal MRI sequences using the denoising diffusion probabilistic model (DDPM) and to compare its performance against one baseline model pix2pix and three advanced models. Methods A total of 1195 consecutive spinal tumor patients who underwent contrast-enhanced MRI at two hospitals were divided into a training set (n = 809, 49 ± 17 years, 437 men), an internal test set (n = 203, 50 ± 16 years, 105 men), and an external test set (n = 183, 52 ± 16 years, 94 men). Input sequences were T1- and T2-weighted images, and T2 fat-saturation images. The output was T1CE images. In the test set, one radiologist read the virtual images and marked all visible enhancing lesions. Results were evaluated using sensitivity (SE) and false discovery rate (FDR). We compared differences in lesion size and enhancement degree between reference and virtual images, and calculated signal-to-noise (SNR) and contrast-to-noise ratios (CNR) for image quality assessment. Results In the external test set, the mean squared error was 0.0038±0.0065, and structural similarity index 0.78±0.10. Upon evaluation by the reader, the overall SE of the generated T1CE images was 94% with FDR 2%. There was no difference in lesion size or signal intensity ratio between the reference and generated images. The CNR was higher in the generated images than the reference images (9.241 vs. 4.021; P＜0.001). Conclusion The proposed DDPM demonstrates potential as an alternative to gadolinium contrast in spinal MRI examinations of oncologic patients.

A deep learning (DL) model that can automatically detect and classify cervical canal and neural foraminal stenosis using cervical spine magnetic resonance imaging (MRI) can improve diagnostic accuracy and efficiency. A method comprising region-of-interest (ROI) detection and cascade prediction was formulated for diagnosing cervical spinal stenosis based on a DL model. First, three part-specific convolutional neural networks were employed to detect the ROIs in different parts of the cervical MR images. Cascade prediction of the stenosis categories was subsequently performed to record the stenosis level and position on each patient slice. Finally, the results were combined to obtain a patient-level diagnostic report. Performance was evaluated based on the accuracy (ACC), area under the curve (AUC), sensitivity, specificity, F1 Score, diagnosis time of the DL model, and recall rate for ROI detection localization. The average recall rate of the ROI localization was 89.3

Background Accurate prognostic assessment is vital for the personalized treatment of endometrial cancer (EC). Although radiomics models have demonstrated prognostic potential in EC, the impact of region of interest (ROI) delineation strategies and the clinical significance of peritumoral features remain uncertain. Our study thereby aimed to explore the predictive performance of varying radiomics models for the prediction of LVSI, DMI, and disease stage in EC. Methods Patients with 174 histopathology-confirmed EC were retrospectively reviewed. ROIs were manually delineated using the 2D and 3D approach on T2-weighted MRI images. Six radiomics models involving intratumoral (2D(intra) and 3D(intra)), peritumoral (2D(peri) and 3D(peri)), and combined models (2D(intra + peri) and 3D(intra + peri)) were developed. Models were constructed using the logistic regression method with five-fold cross-validation. Area under the receiver operating characteristic curve (AUC) was assessed, and was compared using the Delong's test. Results No significant differences in AUC were observed between the 2D(intra) and 3D(intra) models, or the 2D(peri) and 3D(peri) models in all prediction tasks (P > 0.05). Significant difference was observed between the 3D(intra) and 3D(peri) models for LVSI (0.738 vs. 0.805) and DMI prediction (0.719 vs. 0.804). The 3D(intra + peri) models demonstrated significantly better predictive performance in all 3 prediction tasks compared to the 3D(intra) model in both the training and validation cohorts (P < 0.05). Conclusions Comparable predictive performance was observed between the 2D and 3D models. Combined models significantly improved predictive performance, especially with 3D delineation, suggesting that intra- and peritumoral features can provide complementary information for comprehensive prognostication of EC.

Objectives: To develop a deep learning (DL) model for segmentation of the suprapatellar capsule (SC) and infrapatellar fat pad (IPFP) based on sagittal proton density-weighted images and to distinguish between three common types of knee synovitis. Materials and Methods: This retrospective study included 376 consecutive patients with pathologically confirmed knee synovitis (rheumatoid arthritis, gouty arthritis, and pigmented villonodular synovitis) from two institutions. A semantic segmentation model was trained on manually annotated sagittal proton density-weighted images. The segmentation results of the regions of interest and patients' sex and age were used to classify knee synovitis after feature processing. Classification by the DL method was compared to the classification performed by radiologists. Results: Data of the 376 patients (mean age, 42 +/- 15 years; 216 men) were separated into a training set ( n = 233), an internal test set ( n = 93), and an external test set ( n = 50). The automated segmentation model showed good performance (mean accuracy: 0.99 and 0.99 in the internal and external test sets). On the internal test set, the DL model performed better than the senior radiologist (accuracy: 0.86 vs. 0.79; area under the curve [AUC]: 0.83 vs. 0.79). On the external test set, the DL diagnostic model based on automatic segmentation performed as well or better than senior and junior radiologists (accuracy: 0.79 vs. 0.79 vs. 0.73; AUC: 0.76 vs. 0.77 vs. 0.70). Conclusion: DL models for segmentation of SC and IPFD can accurately classify knee synovitis and aid radiologic diagnosis.

Objectives To evaluate the image quality and diagnostic performance of AI-assisted compressed sensing (ACS) accelerated two-dimensional fast spin-echo MRI compared with standard parallel imaging (PI) in clinical 3.0T rapid knee scans. Methods This prospective study enrolled 130 consecutive participants between March and September 2022. The MRI scan procedure included one 8.0-min PI protocol and two ACS protocols (3.5 min and 2.0 min). Quantitative image quality assessments were performed by evaluating edge rise distance (ERD) and signal-to-noise ratio (SNR). Shapiro-Wilk tests were performed and investigated by the Friedman test and post hoc analyses. Three radiologists independently evaluated structural disorders for each participant. Fleiss κ analysis was used to compare inter-reader and inter-protocol agreements. The diagnostic performance of each protocol was investigated and compared by DeLong’s test. The threshold for statistical significance was set at p < 0.05. Results A total of 150 knee MRI examinations constituted the study cohort. For the quantitative assessment of four conventional sequences with ACS protocols, SNR improved significantly ( p < 0.001), and ERD was significantly reduced or equivalent to the PI protocol. For the abnormality evaluated, the intraclass correlation coefficient ranged from moderate to substantial between readers ( κ = 0.75–0.98) and between protocols ( κ = 0.73–0.98). For meniscal tears, cruciate ligament tears, and cartilage defects, the diagnostic performance of ACS protocols was considered equivalent to PI protocol (Delong test, p > 0.05). Conclusions Compared with the conventional PI acquisition, the novel ACS protocol demonstrated superior image quality and was feasible for achieving equivalent detection of structural abnormalities while reducing acquisition time by half. Clinical relevance statement Artificial intelligence–assisted compressed sensing (ACS) providing excellent quality and a 75% reduction in scanning time presents significant clinical advantages in improving the efficiency and accessibility of knee MRI for more patients. Key Points • The prospective multi-reader study showed no difference in diagnostic performance between parallel imaging and AI-assisted compression sensing (ACS) was found. • Reduced scan time, sharper delineation, and less noise with ACS reconstruction. • Improved efficiency of the clinical knee MRI examination by the ACS acceleration.

Objectives: Developing a Deep learning (DL) model to automatically detect and classify cervical canal and neural foraminal stenosis on cervical spine MRI can improve the accuracy and efficiency of its diagnostic. Methods: A method for cervical spinal stenosis was proposed based on the DL model, consisting of region of interest (ROI) detection and cascade prediction. First, three part-specific convolutional neural networks were used to detect the ROIs in different parts of cervical MRI images. Then, the cascade prediction of stenosis categories was performed to obtain the results of stenosis level and position on each slice of the patients. Finally, in the testing, the results were fused to obtain a patient-level diagnostic report. The performance was evaluated with the metrics of accuracy (ACC), area under curve (AUC), sensitivity, specificity, F1 Score, and diagnosis time of the DL model, as well as recall rate for ROI detection localization. Results: The average recall rate of ROIs localization reached 89.3% (neural foramen), and 99.7% (central canal) under the five-fold cross-validation of our DL model. In dichotomous classification (normal or mild vs moderate or severe), ACC and AUC of the DL model were very close to the level of radiologists, and the F1 score (84.8%) of the DL model was slightly higher than that of radiologists (83.8%) at central canal. Conclusion: The DL model showed comparable performance with subspecialist radiologists for detection and classification of the central canal and neural foraminal stenosis at cervical spine MRI with significant time-saving.

BACKGROUND:Primary leiomyosarcoma of the spine is extremely rare and lacks specific clinical symptoms. This study investigated the imaging manifestations and clinicopathological findings of primary leiomyosarcoma of the spine, aiming to improve the radiologists' understanding of the disease and reduce misdiagnoses.METHODS:The clinical, imaging, and pathological manifestations in eleven patients with pathologically confirmed primary leiomyosarcoma of the spine were retrospectively analyzed. The imaging features analyzed included lesion location, shape, border, size, and density/intensity, and adjacent bone destruction status, residual bone trabeculae, vertebral compression, and contrast enhancement.RESULTS:The patients' primary clinical symptom was usually focal pain. Primary leiomyosarcoma of the spine was mostly a solitary lesion and tended to occur in the posterior elements. The tumors had a lobulated shape with osteolytic bone destruction, ill-defined borders, and could involve multiple segments. Computed tomography (CT) examination showed isodense masses. Six patients showed residual bone trabeculae. Two patients had miscellany T2-weighted imaging (T2WI) signals, while the tumor and spinal cord of the remaining patients were isointense on T1-weighted imaging (T1WI) and T2WI. Among the seven patients who underwent contrast-enhanced scanning, six displayed homogeneous enhancement. Eight patients underwent gross-total tumor resection with no recurrence.CONCLUSIONS:Primary leiomyosarcoma of the spine tends to be a solitary lesion in the posterior elements and appears as a lobulated mass with osteolytic bone destruction and an ill-defined border. The tumor and spinal cord can be isointense on T1WI and T2WI. Contrast-enhanced scanning displays homogeneous enhancement. The lesion tends not to recur after surgical gross-total tumor resection.

Objectives To propose a deep learning-based classification framework, which can carry out patient-level benign and malignant tumors classification according to the patient’s multi-plane images and clinical information. Methods A total of 430 cases of spinal tumor, including axial and sagittal plane images by MRI, of which 297 cases for training (14072 images), and 133 cases for testing (6161 images) were included. Based on the bipartite graph and attention learning, this study proposed a multi-plane attention learning framework, BgNet, for benign and malignant tumor diagnosis. In a bipartite graph structure, the tumor area in each plane is used as the vertex of the graph, and the matching between different planes is used as the edge of the graph. The tumor areas from different plane images are spliced at the input layer. And based on the convolutional neural network ResNet and visual attention learning model Swin-Transformer, this study proposed a feature fusion model named ResNetST for combining both global and local information to extract the correlation features of multiple planes. The proposed BgNet consists of five modules including a multi-plane fusion module based on the bipartite graph, input layer fusion module, feature layer fusion module, decision layer fusion module, and output module. These modules are respectively used for multi-level fusion of patient multi-plane image data to realize the comprehensive diagnosis of benign and malignant tumors at the patient level. Results The accuracy (ACC: 79.7%) of the proposed BgNet with multi-plane was higher than that with a single plane, and higher than or equal to the four doctors’ ACC (D1: 70.7%, p=0.219; D2: 54.1%, p<0.005; D3: 79.7%, p=0.006; D4: 72.9%, p=0.178). Moreover, the diagnostic accuracy and speed of doctors can be further improved with the aid of BgNet, the ACC of D1, D2, D3, and D4 improved by 4.5%, 21.8%, 0.8%, and 3.8%, respectively. Conclusions The proposed deep learning framework BgNet can classify benign and malignant tumors effectively, and can help doctors improve their diagnostic efficiency and accuracy. The code is available at https://github.com/research-med/BgNet.

Abstract Background Epithelioid hemangioendothelioma (EHE) is a low-grade malignant vascular neoplasm with the potential to metastasize. Primary EHE of the spine is very rare and an accurate diagnosis is crucial to treatment planning. We aim to investigate the imaging and clinical data of spinal EHE to improve the understanding of the disease. Methods We retrospectively analyzed the imaging manifestations and clinical data of 12 cases with pathologically confirmed spinal EHE. The imaging features analyzed included number, locations, size, border, density, signal, majority of the lesions, expansile osteolysis, residual bone trabeculae, sclerotic rim, vertebral compression, enhancement. Results Patients included 5 female and 7 male patients (mean age: 43.0 ± 19.6 years; range 15–73 years). Multiple lesions were noted in 1 case and single lesion was noted in 11 cases. The lesions were located in the thoracic, cervical, lumbar, and sacral vertebrae in 7, 3, 1, and 1 cases, respectively. They were centered in the vertebral body and posterior elements in 9 and 3 cases, respectively. Residual bone trabeculae, no sclerotic margin, and surrounding soft-tissue mass were noted in 11 cases, each, and mild expansile osteolysis and vertebral compression were noted in 10 and 6 cases, respectively. MRI was performed for 11 patients, all of whom showed isointensity on T1WI, hyperintensity or slight hyperintensity on T2WI, and hyperintensity on fat-suppressed T2WI. A marked enhancement pattern was noted in 10 cases. Conclusion Spinal EHE tend to develop in the thoracic vertebrae. EHE should be considered when residual bone trabeculae can be seen in the bone destruction area, accompanied by pathological compression fracture, no sclerotic rim, and high signal intensity for a vascular tumor on T2WI.

Quantitative comparison of diffusion parameters from various models of diffusion-weighted (DWI) and diffusion kurtosis (DKI) imaging for distinguishing spinal metastases and chordomas. DWI and DKI examinations were performed in 31 and 13 cases of spinal metastases and chordomas, respectively. DWI derived apparent diffusion coefficient (ADC), true diffusion coefficient (D), pseudo diffusion coefficient (D*), perfusion fraction (f), water molecular distributed diffusion coefficient (DDC), and intravoxel water diffusion heterogeneity (α). DKI derived mean diffusivity (MD) and mean kurtosis (MK). Independent sample t-testing compared statistical differences among parameters. Sensitivity, specificity, and area under the receiver operating characteristic (ROC) curve were determined. Pearson correlation analysis evaluated the parameters’ correlations. ADC, D, f, DDC, α, and MD were significantly lower in spinal metastases than chordomas (all P < 0.05). MK was significantly higher in spinal metastases than chordomas (P < 0.05). D had the highest area under the ROC curve (AUC) of 0.886, greater than MD (AUC = 0.706) or DDC (AUC = 0.742) in differentiating the two tumors (both P < 0.05). Combining D with f and α statistically significantly increased the AUC for diagnosis (to 0.995) relative to D alone (P < 0.05). There was a certain correlation among DDC, ADC, and D (all P < 0.05). Monoexponential, biexponential, and stretched-exponential models of DWI and DKI can potentially differentiate spinal metastases and chordomas. D combined with f and α performed best.

PurposeThis project aimed to assess the significance of vascular endothelial growth factor (VEGF) and p53 for predicting progression-free survival (PFS) in patients with spinal giant cell tumor of bone (GCTB) and to construct models for predicting these two biomarkers based on clinical and computer tomography (CT) radiomics to identify high-risk patients for improving treatment.Material and MethodsA retrospective study was performed from April 2009 to January 2019. A total of 80 patients with spinal GCTB who underwent surgery in our institution were identified. VEGF and p53 expression and clinical and general imaging information were collected. Multivariate Cox regression models were used to verify the prognostic factors. The radiomics features were extracted from the regions of interest (ROIs) in preoperative CT, and then important features were selected by the SVM to build classification models, evaluated by 10-fold crossvalidation. The clinical variables were processed using the same method to build a conventional model for comparison.ResultsThe immunohistochemistry of 80 patients was obtained: 49 with high-VEGF and 31 with low-VEGF, 68 with wild-type p53, and 12 with mutant p53. p53 and VEGF were independent prognostic factors affecting PFS found in multivariate Cox regression analysis. For VEGF, the Spinal Instability Neoplastic Score (SINS) was greater in the high than low groups, p < 0.001. For p53, SINS (p = 0.030) and Enneking stage (p = 0.017) were higher in mutant than wild-type groups. The VEGF radiomics model built using 3 features achieved an area under the curve (AUC) of 0.88, and the p53 radiomics model built using 4 features had an AUC of 0.79. The conventional model built using SINS, and the Enneking stage had a slightly lower AUC of 0.81 for VEGF and 0.72 for p53.Conclusionp53 and VEGF are associated with prognosis in patients with spinal GCTB, and the radiomics analysis based on preoperative CT provides a feasible method for the evaluation of these two biomarkers, which may aid in choosing better management strategies.

Objectives To explore the predictive value of preoperative imaging in patients with spinal giant cell tumor of bone (GCTB) for postoperative recurrence and risk stratification. Methods Clinical data for 62 cases of spinal GCTB diagnosed and treated at our hospital from 2008 to 2018 were identified. All patients were followed up for more than 2 years according to the clinical guidelines after surgery. Medical history data including baseline demographic and clinical characteristics, computed tomography (CT) and magnetic resonance imaging (MRI) findings of recurrent and non‐recurrent patients were compared. Two musculoskeletal radiologists read the images and were blinded to the clinical data. The imaging features associated with postoperative recurrence were analyzed by multivariate logistic regression, and receiver operating characteristic (ROC) curve analysis was performed to determine the optimal cutoff value of the largest lesion diameter predicting recurrence after surgery. Results According to whether the disease recurred within the follow‐up period, patients were divided into the recurrence group and the non‐recurrence group. Of 62 patients (29 males and 33 females), 17 had recurrence and 45 did not. The recurrence rate was 27.4%. The mean follow‐up time was 73.66 (± 32.92) months. The three major treatments were total en bloc spondylectomy (n = 26), intralesional spondylectomy (n = 20), and curettage(n = 16). A total of 16 CT and MRI features were analyzed. A univariate analysis showed no significant difference in age, sex, treatment, multi‐vertebral body involvement, location, boundary, expansile mass, residual bone crest, paravertebral soft tissue mass, CT value, and MRI signal on T1‐weighted imaging (WI), T2‐WI, and T2‐WI fat suppression (FS) sequences (P > 0.05). The largest lesion diameter [(4.68 ± 1.79) vs (5.92 ± 2.17) cm, t = 2.287, P = 0.026] and the vertebral compression fracture (51% vs 82%, χ2 = 5.005, P = 0.025) were significantly different between the non‐recurrence and recurrence groups. Logistic regression analysis showed that both largest lesion diameter (odds ratio [OR], 1.584; 95% confidence interval [CI], 1.108–2.264; P = 0.012) and compression fracture (OR, 8.073; 95%CI, 1.481–11.003; P = 0.016) were independent predictors of postoperative recurrence. When we set the cutoff value for the largest lesion diameter at 4.2 cm, the sensitivity and specificity for distinguishing the recurrence and non‐recurrence of GCTB were 94.1% and 42.2%, respectively, and the area under the curve (AUC) was 0.671. The combined model achieved a sensitivity, specificity and accuracy of 47.1%, 97.8% and 83.9%, respectively. Conclusions In spinal GCTB, maximum lesion diameter and the vertebral compression fracture are associated with tumor recurrence after surgery, which may provide helpful information for planning personalized treatment.

Objectives: To determine if radiomics analysis based on preoperative computed tomography (CT) can predict early postoperative recurrence of giant cell tumor of bone (GCTB) in the spine. Methods: In a retrospective review, 62 patients with pathologically confirmed spinal GCTB from March 2008 to February 2018, with a minimum follow-up of 24 months, were identified. The mean follow-up was 73.7 months (range, 28.7-152.1 months). The clinical information including age, gender, lesion location, multi-vertebral involvement, and surgical methods, were obtained. CT images acquired before the operation were retrieved for radiomics analysis. For each case, the tumor regions of interest (ROI) was manually outlined, and a total of 107 radiomics features were extracted. The features were selected via the sequential selection process by using the support vector machine (SVM), then used to construct classification models with Gaussian kernels. The differentiation between recurrence and non-recurrence groups was evaluated by ROC analysis, using 10-fold cross-validation. Results: Of the 62 patients, 17 had recurrence with a recurrence rate of 27.4%. None of the clinical information was significantly different between the two groups. Patients receiving curettage had a higher recurrence rate (6/16 = 37.5%) compared to patients receiving TES (6/26 = 23.1%) or intralesional spondylectomy (5/20 = 25%). The final radiomics model was built using 10 selected features, which achieved an accuracy of 89% with AUC of 0.78. Conclusions: The radiomics model developed based on pre-operative CT can achieve a high accuracy to predict the recurrence of spinal GCTB. Patients who have a high risk of early recurrence should be treated more aggressively to minimize recurrence. (c) 2021 The Authors. Published by Elsevier GmbH. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

OBJECTIVE:Solitary fibrous tumours (SFTs) occurring in the spine are rare. Herein, we review the clinical and imaging data of spinal SFT.METHODS:We retrospectively analysed eight cases of pathologically confirmed spinal SFT imaging and clinical data, pathological manifestations, surgical methods, and follow-up results.RESULTS:Five SFTs cases occurred in the cervical spine, two in the thoracic spine, and one in the lumbosacral spine. Five cases showed a dumbbell-shaped or lobulated soft tissue mass that grew across the intervertebral foramen, two cases showed an expansive intraosseous mass formation in the vertebral body and/or posterior element, and one case showed a long-spindle shaped intraspinal canal mass growing along the spinal canal. Seven caused local invasion and destruction of the vertebral body and posterior element. Benign SFTs displayed a good prognosis, whereas malignant SFTs were prone to recurrence and metastasis (3/4).CONCLUSION:Spinal SFTs are difficult to characterise with imaging and required pathological and immunohistochemical investigation. Prolonged follow-up is recommended once a diagnosis of spinal SFTs has been established because of the unclear biology.KEY POINTS:• Spinal solitary fibrous tumours are extremely rare. • SFTs should be showed the differential of masses developing though the foramen. • Combing imaging with pathology and immunochemistry assesses the diagnosis and establish nature.

PurposeTo characterize the morphological and dynamic-contrast-enhanced (DCE) MRI features of chordoma and giant cell tumor (GCT) of bone occurring in the axial skeleton. Materials and MethodsA total of 13 patients with chordoma and 26 patients with GCT who received conventional T1, T2, and DCE-MRI on 3 Tesla MR scanners were retrospectively identified and analyzed. Two radiologists evaluated morphological features independently, including the lesion location, expansile bone changes, vertebral compression, presence of paraspinal soft tissue mass, fibrous septa, and the signal intensity on T1WI and T2WI. The inter-observer agreement was evaluated by kappa test. The DCE kinetics was measured to obtain the initial area under curve (IAUC) and the wash-out slope; also the two-compartmental pharmacokinetic model was applied to obtain K-trans and k(ep). The diagnostic accuracy was evaluated by CHAID decision tree and ROC analysis. ResultsChordomas were more likely to show soft tissue mass than GCTs (13/13=100% versus 15/26=58%; P=0.007), as well as fibrous septa (9/13=69% versus 0; P<0.001). In decision tree analysis, presence of fibrous septa and lesion location yield 31/39=79% accuracy. The DCE-MRI pharmacokinetic parameters K-trans and k(ep) of GCTs were significantly higher than those of chordomas, 0.130.65 versus 0.060.04 (1/min) for K-trans, 0.620.22 versus 0.17 +/- 0.12 (1/min) for k(ep), P<0.001 for both. If using kep=0.43/min as the cut-off value, it achieved 100% sensitivity and 92% specificity to differentiate chordoma from GCT, with an overall accuracy of 37/39=95%. The IAUC was highly correlated with K-trans (r=0.94), and the slope was highly correlated with k(ep) (r=0.95). ConclusionSeveral morphological features were significantly different between chordoma and GCT, but their diagnostic performance was inferior to that of DCE-MRI.