Investigating the correlation between the interlaminar space size on plain radiograph and lumbar spinal stenosis (LSS).
100 Patients with LSS and 100 normal participants without LSS were included in this study. Lumbar interlaminar parameters were measured on plain radiographs. Spinal canal parameters were measure on CT and MRI. These image parameters were compared between LSS and control group. The linear correlation among interlaminar parameters, spinal canal parameters and Oswestry Disability Index (ODI) were analyzed. The cut-off values of interlaminar parameters for diagnosing L3/4 and L4/5 symptomatic LSS were analyzed.
Excepting for L1/2 interlaminar width, all interlaminar and spinal canal related parameters in LSS group were significantly smaller than those in control group. Excepting for L1/2 interlaminar width and L5/S1 interlaminar height, all interlaminar parameters had significantly positive linear correlation with spinal canal parameters accordingly in each lumbar level. For diagnosis of symptomatic LSS, The cut off values of L4/5 interlaminar width, height and area were 18.46 mm, 11.37 mm and 134.05 mm2 while 15.78 mm, 13.59 mm and 157.98 mm2 in L3/4. Both spinal canal size and interlaminar size had no linear correlation with ODI in cases of L4/5 LSS.
Lumbar interlaminar space size on plain radiograph has positive linear correlation with developmental and degenerative LSS, excepting for L1/2 interlaminar width and L5/S1 interlaminar height. Lumbar plain radiograph can be a feasible way for predicting and helping to diagnose LSS through evaluating the interlaminar space size.
III.

BACKGROUND: To develop a deep learning model to classify primary bone tumors from preoperative radiographs and compare performance with radiologists.
METHODS: A total of 1356 patients (2899 images) with histologically confirmed primary bone tumors and pre-operative radiographs were identified from five institutions\' pathology databases. Manual cropping was performed by radiologists to label the lesions. Binary discriminatory capacity (benign versus not-benign and malignant versus not-malignant) and three-way classification (benign versus intermediate versus malignant) performance of our model were evaluated. The generalizability of our model was investigated on data from external test set. Final model performance was compared with interpretation from five radiologists of varying level of experience using the Permutations tests.
RESULTS: For benign vs. not benign, model achieved area under curve (AUC) of 0•894 and 0•877 on cross-validation and external testing, respectively. For malignant vs. not malignant, model achieved AUC of 0•907 and 0•916 on cross-validation and external testing, respectively. For three-way classification, model achieved 72•1% accuracy vs. 74•6% and 72•1% for the two subspecialists on cross-validation (p = 0•03 and p = 0•52, respectively). On external testing, model achieved 73•4% accuracy vs. 69•3%, 73•4%, 73•1%, 67•9%, and 63•4% for the two subspecialists and three junior radiologists (p = 0•14, p = 0•89, p = 0•93, p = 0•02, p < 0•01 for radiologists 1-5, respectively).
CONCLUSIONS: Deep learning can classify primary bone tumors using conventional radiographs in a multi-institutional dataset with similar accuracy compared to subspecialists, and better performance than junior radiologists.
BACKGROUND: The project described was supported by RSNA Research & Education Foundation, through grant number RSCH2004 to Harrison X. Bai.

To discuss whether radiologic parameters are closely related to posterior ligamentous complex (PLC) injury identified by magnetic resonance imaging (MRI).
One hundred and five thoracolumbar fracture (T11-L2) patients were retrospectively analyzed in the study. The patients were divided into different groups by the status of the PLC on MRI: intact, incompletely ruptured and ruptured. The radiographic parameters included the anterior edge-inferior endplate angle (AEIEA), the anterior edge displacement (AED), the Cobb angle (CA), the region angle (RA), the sagittal index (SI), local kyphosis (LK), the anterior/posterior vertebral height ratio (A/P ratio), the anterior vertebral height ratio (AVH ratio), and bony fragment in front of the fractured vertebra (BFOFV). T test, Pearson\'s Chi-square and multivariate logistic regression were calculated for the variables.
Supraspinous ligament (SSL) rupture versus intact was not only associated with the occurrence of AEIEA <70°, LK >25° and BFOFV, but also with increased AED (9.89 ± 3.12 mm and 9.34 ± 3.36 mm, P = 0.034), RA (9.52 ± 3.93° versus 7.91 ± 3.99°, P = 0.042), and LK (23.98 ± 5.88° versus 15.55 ± 5.28°, P = 0.021). The indications for interspinous ligament (ISL) injury included AEIEA <75°, AEIEA <70° (P = 0.004 and P < 0.001, respectively), increased AED (P = 0.010), LK >25° (P = 0.024), AVH (P < 0.001), and BFOFV (P < 0.001). Multivariate logistic regression analysis revealed that AEIEA <70° and BFOFV were high risk factors for SSL rupture [standard partial regression coefficients (betas) were 0.439 and 0.408, P = 0.003 and 0.001, respectively] and ISL rupture (betas were 0.548 and 0.494, P = 0.028 and 0.001, respectively). Increased AED and LK >25° were also related to either ISL rupture (P = 0.035 and 0.001, respectively) or SSL rupture (P = 0.014 and 0.008, respectively).
Our data may prove useful in a preliminary assessment of the PLC integrity based on plain radiographic imaging. We show that radiologic indications, such as AEIEA <70°, BFOFV, LK >25°, and increased AED, are correlated with ISL or SSL rupture, while RA, CA, SI, A/P ratio, and AVH ratio are not.

OBJECTIVE: The aim of this study was to prove that it is feasible to diagnose discoid lateral meniscus in radiographs. Plain radiographic findings of discoid lateral menisci with matched controls were analysed and compared in a quantitative method.
METHODS: Sixty consecutive patients (60 knees) who were diagnosed with discoid lateral meniscus (discoid group) by magnetic resonance imaging (MRI) were included. Another 60 age- and sex-matched controls with normal medial and lateral menisci on the basis of MRI findings were included as the control group. Each plain radiograph was evaluated from the anteroposterior view for the following variables: height of the fibular head (HFH), lateral joint space distance (LJSD), height of the lateral tibial spine (HLTS), obliquity of the lateral tibial plateau, obliquity of the lateral femoral condyle, distance from the lateral tibial spine to the lateral femoral condyle, height of the medial tibial spine, chordal distance of the femoral condyle (CDLF, CDMF), the HFH/LJSD, LJSD/HLTS and the CDLF/CDMF.
RESULTS: A significant difference was found in the HFH, LJSD, HLTS, DLC, CDLF, HFH/LJSD and LJSD/HLTS between the two groups. The cut-off values of the HFH, LJSD, HLTS, DLC, CDLF, HFH/LJSD and LJSD/HLTS were 12.9 mm, 6.6 mm, 7.8 mm, 3.0 mm, 2.7 mm, 2.0 and 0.9, respectively. Among the cut-off values in diagnosing discoid lateral meniscus, the sensitivity, specificity and ROC curve area of LJSD/HLTS were as high as 73.6 %, 83.0 % and 0.8, respectively. The corresponding values of the HFH/LJSD were as high as 66.0 %, 86.8 % and 0.8. For the first two indicators, the results of the HFH/LJSD and LJSD/HLTS were higher than that of most other parameters. At the same time, the ROC curve area of the HFH/LJSD and LJSD/HLTS ranked highest among all the results.
CONCLUSIONS: There were significant differences in the HFH, LJSD, HLTS, DLC, CDLF, HFH/LJSD and LJSD/HLTS, especially the HFH/LJSD and the LJSD/HLTS, between plain radiographic findings of discoid lateral meniscus patients and normal controls. The results of the HFH/LJSD and the LJSD/HLTS showed a positive impact on the diagnosis of discoid lateral meniscus in this research. These findings enable radiographs to screen for discoid lateral meniscus.
METHODS: II.