Explore the vast range of titles available.

Although CT radiomics has shown promising results in the evaluation of vertebral fractures, the need for manual segmentation of fractured vertebrae limited the routine clinical implementation of radiomics. Therefore, automated segmentation of fractured vertebrae is needed for successful clinical use of radiomics. In this study, we aimed to develop and validate an automated algorithm for segmentation of fractured vertebral bodies on CT, and to evaluate the applicability of the algorithm in a radiomics prediction model to differentiate benign and malignant fractures. A convolutional neural network was trained to perform automated segmentation of fractured vertebral bodies using 341 vertebrae with benign or malignant fractures from 158 patients, and was validated on independent test sets (internal test, 86 vertebrae [59 patients]; external test, 102 vertebrae [59 patients]). Then, a radiomics model predicting fracture malignancy on CT was constructed, and the prediction performance was compared between automated and human expert segmentations. The algorithm achieved good agreement with human expert segmentation at testing (Dice similarity coefficient, 0.93–0.94; cross-sectional area error, 2.66–2.97%; average surface distance, 0.40–0.54 mm). The radiomics model demonstrated good performance in the training set (AUC, 0.93). In the test sets, automated and human expert segmentations showed comparable prediction performances (AUC, internal test, 0.80 vs 0.87, p = 0.044; external test, 0.83 vs 0.80, p = 0.37). In summary, we developed and validated an automated segmentation algorithm that showed comparable performance to human expert segmentation in a CT radiomics model to predict fracture malignancy, which may enable more practical clinical utilization of radiomics.

Conventional computer-aided diagnosis using convolutional neural networks (CNN) has limitations in detecting sensitive changes and determining accurate decision boundaries in spectral and structural diseases such as scoliosis. We devised a new method to detect and diagnose adolescent idiopathic scoliosis in chest X-rays (CXRs) employing the latent space's discriminative ability in the generative adversarial network (GAN) and a simple multi-layer perceptron (MLP) to screen adolescent idiopathic scoliosis CXRs. Our model was trained and validated in a two-step manner. First, we trained a GAN using CXRs with various scoliosis severities and utilized the trained network as a feature extractor using the GAN inversion method. Second, we classified each vector from the latent space using a simple MLP. The 2-layer MLP exhibited the best classification in the ablation study. With this model, the area under the receiver operating characteristic (AUROC) curves were 0.850 in the internal and 0.847 in the external datasets. Furthermore, when the sensitivity was fixed at 0.9, the model's specificity was 0.697 in the internal and 0.646 in the external datasets. We developed a classifier for Adolescent idiopathic scoliosis (AIS) through generative representation learning. Our model shows good AUROC under screening chest radiographs in both the internal and external datasets. Our model has learned the spectral severity of AIS, enabling it to generate normal images even when trained solely on scoliosis radiographs.

ObjectivesTo develop and validate a scoring system integrating MRI and laboratory findings to differentiate necrotizing fasciitis (NF) from non-necrotizing fasciitis (non-NF).MethodsThis retrospective study included 144 subjects who underwent surgery in one of three tertiary referral centers for NF or cellulitis with non-NF. The development cohort consisted of 96 subjects (NF = 47; non-NF = 49) from one center, and the validation cohort consisted of 48 subjects (NF = 23; cellulitis with non-NF = 25) from two different centers. The Laboratory Risk Indicator for Necrotizing Fasciitis (LRINEC) scoring system and five MRI findings (thickening of the intermuscular deep fascia ≥ 3 mm, extensive involvement of the deep fascia, multi-compartmental involvement in one extremity, presence of gas, and contrast-enhancement pattern) were included in univariate and multivariate logistic regression analysis to identify independent predictors of NF. An additive scoring system was developed using the coefficients of the final regression model. Model performance was assessed for discrimination and calibration. The scoring system was externally validated.ResultThe final scoring system consisted of three variables: thickening of the deep fascia ≥ 3 mm, multi-compartmental involvement, and LRINEC score. The new predictive model showed improved performance (area under the receiver operating characteristic curve [AUC], 0.862; positive and negative predictive values, 82% and 79%, respectively), compared with the LRINEC score alone (0.814, 77% and 67%, respectively). The model also showed good discrimination with the external validation dataset (AUC, 0.933).ConclusionsDifferentiation of NF from severe cellulitis with non-NF can be achieved with the new predictive scoring system.Key Points• The new predictive scoring system integrating two MRI findings with the LRINEC score can help in the differentiation of necrotizing fasciitis from severe cellulitis with non-necrotizing fasciitis.• Thickening of the deep fascia ≥ 3 mm and multi-compartmental involvement were the most important MRI findings for the differentiation.

To evaluate the osseous anatomy of the proximal femur extracted from a 3D-MRI volumetric interpolated breath-hold (VIBE) sequence using either a Dixon or water excitation (WE) fat suppression method, and to measure the overall difference using CT as a reference standard. This retrospective study reviewed imaging of adult patients with hip pain who underwent 3D hip MRI and CT. A semi-automatically segmented CT model served as the reference standard, and MRI segmentation was performed manually for each unilateral hip joint. The differences between Dixon-VIBE-3D-MRI vs. CT, and WE-VIBE-3D-MRI vs. CT, were measured. Equivalence tests between Dixon-VIBE and WE-VIBE models were performed with a threshold of 0.1 mm. Bland-Altman plots and Lin's concordance-correlation coefficient were used to analyze the agreement between WE and Dixon sequences. Subgroup analyses were performed for the femoral head/neck, intertrochanteric, and femoral shaft areas. The mean and maximum differences between Dixon-VIBE-3D-MRI vs. CT were 0.2917 and 3.4908 mm, respectively, whereas for WE-VIBE-3D-MRI vs. CT they were 0.3162 and 3.1599 mm. The mean differences of the WE and Dixon methods were equivalent (P = 0.0292). However, the maximum difference was not equivalent between the two methods and it was higher in WE method. Lin's concordance-correlation coefficient showed poor agreement between Dixon and WE methods. The mean differences between the CT and 3D-MRI models were significantly higher in the femoral shaft area (P = 0.0004 for WE and P = 0.0015 for Dixon) than in the other areas. The maximum difference was greatest in the intertrochanteric area for both techniques. The difference between 3D-MR and CT models were acceptable with a maximal difference below 3.5mm. WE and Dixon fat suppression methods were equivalent. The mean difference was highest at the femoral shaft area, which was off-center from the magnetization field.

This study aimed to determine the prevalence of vertebral venous congestion (VVC) in patients with chemoport insertion, evaluate the imaging characteristics of nodular VVC, and identify the factors associated with VVC. This retrospective single-center study was based on follow-up contrast-enhanced chest computed tomography (CT) of 1412 adult patients who underwent chemoport insertion between January 2016 and December 2016. The prevalence of venous stenosis, reflux, and VVC were evaluated. The imaging features of nodular VVC, including specific locations within the vertebral body, were analyzed. To identify the factors associated with VVC, patients with VVC were compared with a subset of patients without VVC who had been followed up for > 3 years without developing VVC after chemoport insertion. Toward this, a multivariable logistic regression analysis was performed. After excluding 333 patients, 1079 were analyzed (mean age ± standard deviation, 62.3 ± 11.6 years; 540 females). The prevalence of VVC was 5.8% (63/1079), with all patients (63/63) demonstrating vertebral venous reflux and 67% (42/63) with innominate vein stenosis. The median interval between chemoport insertion and VVC was 515 days (interquartile range, 204-881 days). The prevalence of nodular VVC was 1.5% (16/1079), with a mean size of 5.9 ± 3.1 mm and attenuation of 784 ± 162 HU. Nodular VVC tended to be located subcortically. Forty-four patients with VVC underwent CT examinations with contrast injections in both arms; the VVC disappeared in 70% (31/44) when the contrast was injected in the arm contralateral to the chemoport site. Bevacizumab use was independently associated with VVC (odds ratio, 3.45; < 0.001). The prevalence of VVC and nodular VVC was low in patients who underwent chemoport insertion. Nodular VVC was always accompanied by vertebral venous reflux and tended to be located subcortically. To avoid VVC, contrast injection in the arm contralateral to the chemoport site is preferred.

Background Whole-body MRI (WB-MRI) including diffusion-weighted image (DWI) have been widely used in patients with multiple myeloma. However, evidence for the value of WB-MRI in the evaluation of treatment response remains sparse. Therefore, we evaluated the role of WB-MRI in the response assessment. Methods In our WB-MRI registry, we searched multiple myeloma patients treated with chemotherapy who underwent both baseline and follow-up WB-MRI scans. Clinical responses were categorized as complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD), using IMWG criteria. Using RECIST 1.1, MD Anderson (MDA) criteria, and MDA-DWI criteria, imaging responses on WB-MRI were rated as CR, PR, SD, or PD by two radiologists independently. Then, discrepancy cases were resolved by consensus. Weighted Kappa analysis was performed to evaluate agreement between the imaging and clinical responses. The diagnostic accuracy of image responses in the evaluation of clinical CR, objective response (CR and PR), and PD was calculated. Results Forty-two eligible patients were included. There was moderate agreement between imaging and clinical responses (κ = 0.54 for RECIST 1.1, κ = 0.58 for MDA criteria, κ = 0.69 for MDA-DWI criteria). WB-MRI showed excellent diagnostic accuracy in assessment of clinical PD (sensitivity 88.9%, specificity 94.7%, positive predictive value [PPV] 84.2%, negative predictive value [NPV] 96.4% in all three imaging criteria). By contrast, WB-MRI showed low accuracy in assessment of clinical CR (sensitivity 4.5%, specificity 98.1%, PPV 50.0%, NPV 71.2% in all three imaging criteria). As to the clinical objective response, the diagnostic accuracy was higher in MDA-DWI criteria than RECIST 1.1 and MDA criteria (sensitivity/specificity/PPV/NPV, 84.2%/94.4%/98.0%/65.4, 54.4%/100%/100%/40.9, and 61.4%/94.4%/97.2%/43.6%, respectively). Conclusions In the imaging response assessment of multiple myeloma, WB-MRI showed excellent performance in the evaluation of PD, but not in the assessment of CR or objective response. When adding DWI to imaging response criteria, diagnostic accuracy for objective response was improved and agreement between imaging and clinical responses was increased.

Background Limited data exist on the optimal postoperative surveillance protocol for high-grade soft tissue sarcoma, particularly regarding the optimal imaging modality and imaging interval for detecting local recurrence. This study aimed to assess the benefit of short-term postoperative ultrasonography (USG) for detecting local recurrence in patients with high-grade soft tissue sarcoma. Methods Patients with newly diagnosed high-grade soft tissue sarcoma who underwent surgical resection between January 2010 and June 2020 were included. Short-term USG was added to the follow-up protocol as a surveillance tool alongside routine magnetic resonance imaging (MRI). The primary outcome was the additional detection rate of short-term USG compared with routine MRI surveillance for early local recurrence detection. Subgroup analysis was performed to evaluate factors influencing USG detection rate. The additional detection rate of short-term USG for detection of metastatic lymph nodes was also evaluated. The secondary outcome was the false referral rate of short-term USG. Results In total, 198 patients (mean age ± standard deviation: 52.1 ± 15.8 years; 94 women) were included. Local recurrence occurred in 20 patients (10.1%; 20/198). Short-term USG detected local recurrence in advance of routine MRI visits in 7 out of 198 patients, resulting in an additional detection rate of 3.5% (95% CI: 1.7–7.1%). Subgroup analysis showed no significant difference in the short-term USG detection rate based on initial tumor characteristics, and receipt of radiotherapy or chemotherapy. Short-term USG additionally detected five of seven patients with metastatic lymph nodes [2.5% (95% CI, 1.1–5.8%, 5/198)]. The false referral rate of short-term USG was 3.5% (95% CI: 1.7–7.1%; 7/198). Conclusions Short-term USG as part of postoperative surveillance for high-grade soft tissue sarcoma can enhance early detection of local tumor recurrence and metastatic lymphadenopathy. Early detection of local tumor recurrence could lead to a prompt surgical resection and aid in local disease control.

To study the relationship between inflammatory and structural lesions in the sacroiliac joints (SIJs) on MRI and spinal progression observed on conventional radiographs in patients with axial spondyloarthritis (axSpA). One hundred and ten patients who fulfilled the ASAS axSpA criteria were enrolled. All underwent SIJ MRI at baseline and lumbar spine radiographs at baseline and after 2 years. Inflammatory and structural lesions on SIJ MRI were scored using the SPondyloArthritis Research Consortium of Canada (SPARCC) method. Spinal radiographs were scored using the Stoke AS Spinal Score (SASSS). Multivariate logistic regression analysis was performed to identify predictors of spinal progression. Among the 110 patients, 25 (23%) showed significant radiographic progression (change of SASSS≥2) over 2 years. There was no change in the SASSS over 2 years according to the type of inflammatory lesion. Patients with fat metaplasia or ankyloses on baseline MRI showed a significantly higher SASSS at 2 years than those without (p<0.001). According to univariate logistic regression analysis, age at diagnosis, HLA-B27 positivity, the presence of fat metaplasia, erosion, and ankyloses on SIJ MRI, increased baseline CRP levels, and the presence of syndesmophytes at baseline were associated with spinal progression over 2 years. Multivariate analysis identified syndesmophytes and severe fat metaplasia on baseline SIJ MRI as predictive of spinal radiographic progression (OR, 14.74 and 5.66, respectively). Inflammatory lesions in the SIJs on baseline MRI were not associated with spinal radiographic progression. However, fat metaplasia at baseline was significantly associated with spinal progression after 2 years.

ObjectiveAlthough many types of ankle fracture can be combined with syndesmosis injury, preoperative imaging studies rarely reveal instability of the syndesmosis. This study assessed the use of magnetic resonance imaging (MRI) for syndesmotic instability in patients with unstable ankle fracture.MethodsA total of 74 patients who were treated for Lauge-Hansen supination external rotation/Weber B type fracture or pronation external rotation/Weber C type fracture and who underwent MRI for preoperative assessment were enrolled. The MRI findings of the syndesmotic ligament and the results of an intraoperative stress test were evaluated.ResultsTwenty-six patients had a positive result on the intraoperative stress test for syndesmotic instability. The MRI findings of the syndesmotic ligaments revealed that complete tear of the posterior inferior tibiofibular ligament (PITFL) was the most reliable predictor of syndesmotic instability (sensitivity, 74%; specificity, 78%; positive predictive value, 54%). Interobserver agreement for the intraoperative stress test and MRI assessment was excellent, except for the MRI findings of the interosseous ligament (62% agreement; kappa, 0.3).ConclusionsComplete tear of the PITFL on MRI has additional diagnostic value for syndesmotic instability in ankle fracture. However, because the sensitivity might not be sufficient to justify the costs associated with MRI, cost-effectiveness should be considered.

Objectives To develop and validate a combined radiomics-clinical model to predict malignancy of vertebral compression fractures on CT. Methods One hundred sixty-five patients with vertebral compression fractures were allocated to training ( n = 110 [62 acute benign and 48 malignant fractures]) and validation ( n = 55 [30 acute benign and 25 malignant fractures]) cohorts. Radiomics features ( n = 144) were extracted from non-contrast-enhanced CT images. Radiomics score was constructed by applying least absolute shrinkage and selection operator regression to reproducible features. A combined radiomics-clinical model was constructed by integrating significant clinical parameters with radiomics score using multivariate logistic regression analysis. Model performance was quantified in terms of discrimination and calibration. The model was internally validated on the independent data set. Results The combined radiomics-clinical model, composed of two significant clinical predictors (age and history of malignancy) and the radiomics score, showed good calibration (Hosmer-Lemeshow test, p > 0.05) and discrimination in both training (AUC, 0.970) and validation (AUC, 0.948) cohorts. Discrimination performance of the combined model was higher than that of either the radiomics score (AUC, 0.941 in training cohort and 0.852 in validation cohort) or the clinical predictor model (AUC, 0.924 in training cohort and 0.849 in validation cohort). The model stratified patients into groups with low and high risk of malignant fracture with an accuracy of 98.2% in the training cohort and 90.9% in the validation cohort. Conclusions The combined radiomics-clinical model integrating clinical parameters with radiomics score could predict malignancy in vertebral compression fractures on CT with high discriminatory ability. Key Points • A combined radiomics-clinical model was constructed to predict malignancy of vertebral compression fractures on CT by combining clinical parameters and radiomics features. • The model showed good calibration and discrimination in both training and validation cohorts. • The model showed high accuracy in the stratification of patients into groups with low and high risk of malignant vertebral compression fractures.