Association between dichotomized VASARI feature and overall survival in glioblastoma patients: a single-institution propensity score matching analysis.

OBJECTIVES: This study aimed to investigate the intra- and inter-observer consistency of the Visually Accessible Rembrandt Images (VASARI) feature set before and after dichotomization, and the association between dichotomous VASARI features and the overall survival (OS) in glioblastoma (GBM) patients. METHODS: This retrospective study included 351 patients with pathologically confirmed IDH1 wild-type GBM between January 2016 and June 2022. Firstly, VASARI features were assessed by four radiologists with varying levels of experience before and after dichotomization. Cohen's kappa coefficient (κ) was calculated to measure the intra- and inter-observer consistency. Then, after adjustment for confounders using propensity score matching, Kaplan-Meier curves were used to compare OS differences for each dichotomous VASARI feature. Next, patients were randomly stratified into a training set (n = 211) and a test set (n = 140) in a 3:2 ratio. Based on the training set, Cox proportional hazards regression analysis was adopted to develop combined and clinical models to predict OS, and the performance of the models was evaluated with the test set. RESULTS: Eleven VASARI features with κ value of 0.61-0.8 demonstrated almost perfect agreement after dichotomization, with the range of κ values across all readers being 0.874-1.000. Seven VASARI features were correlated with GBM patient OS. For OS prediction, the combined model outperformed the clinical model in both training set (C-index, 0.762 vs. 0.723) and test set (C-index, 0.812 vs. 0.702). CONCLUSION: The dichotomous VASARI features exhibited excellent inter- and intra-observer consistency. The combined model outperformed the clinical model for OS prediction.

Risk factor analysis for major mediastinal vessel invasion in thymic epithelial tumors based on multi-slice CT Imaging.

Objective: To explore the characteristics and risk factors for major mediastinal vessel invasion in different risk grades of thymic epithelial tumors (TETs) based on computed tomography (CT) imaging, and to develop prediction models of major mediastinal artery and vein invasion. Methods: One hundred and twenty-two TET patients confirmed by histopathological analysis who underwent thorax CT were enrolled in this study. Clinical and CT data were retrospectively reviewed for these patients. According to the abutment degree between the tumor and major mediastinal vessels, the arterial invasion was divided into grade I, II, and III (< 25%, 25 - 49%, and ≥ 50%, respectively); the venous invasion was divided into grade I and II (< 50% and ≥ 50%). The degree of vessel invasion was compared among different defined subtypes or stages of TETs using the chi-square tests. The risk factors associated with TET vascular invasion were identified using multivariate logistic regression analysis. Results: Based on logistic regression analysis, male patients (ß = 1.549; odds ratio, 4.824) and the pericardium or pleural invasion (ß = 2.209; odds ratio, 9.110) were independent predictors of 25% artery invasion, and the midline location (ß = 2.504; odds ratio, 12.234) and mediastinal lymphadenopathy (ß = 2.490; odds ratio, 12.06) were independent predictors of 50% artery invasion. As for 50% venous invasion, the risk factors include midline location (ß = 2.303; odds ratio, 10.0), maximum tumor diameter larger than 5.9 cm (ß = 4.038; odds ratio, 56.736), and pericardial or pleural effusion (ß = 1.460; odds ratio, 4.306). The multivariate logistic model obtained relatively high predicting efficacy, and the area under the curve (AUC), sensitivity, and specificity were 0.944, 84.6%, and 91.7% for predicting 50% artery invasion, and 0.913, 81.8%, and 86.0% for 50% venous invasion in TET patients, respectively. Conclusion: Several CT features can be used as independent predictors of ≥50% artery or venous invasion. A multivariate logistic regression model based on CT features is helpful in predicting the vascular invasion grades in patients with TET.

Optimizing the radiomics-machine-learning model based on non-contrast enhanced CT for the simplified risk categorization of thymic epithelial tumors: A large cohort retrospective study.

PURPOSE: This study aimed to establish and compare the radiomics machine learning (ML) models based on non-contrast enhanced computed tomography (NECT) and clinical features for predicting the simplified risk categorization of thymic epithelial tumors (TETs). EXPERIMENTAL DESIGN: A total of 509 patients with pathologically confirmed TETs from January 2009 to May 2018 were retrospectively enrolled, consisting of 238 low-risk thymoma (LRT), 232 high-risk thymoma (HRT), and 39 thymic carcinoma (TC), and were divided into training (n = 433) and testing cohorts (n = 76) according to the admission time. Volumes of interest (VOIs) covering the whole tumor were manually segmented on preoperative NECT images. A total of 1218 radiomic features were extracted from the VOIs, and 4 clinical variables were collected from the hospital database. Fourteen ML models, along with varied feature selection strategies, were used to establish triple-classification models using the radiomic features (radiomic models), while clinical-radiomic models were built after combining with the clinical variables. The diagnostic accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curve (AUC) of radiologist assessment, the radiomic and clinical-radiomic models were evaluated on the testing cohort. RESULTS: The Support Vector Machine (SVM) clinical-radiomic model demonstrated the highest AUC of 0.841 (95% CI 0.820 to 0.861) on the cross-validation result and reached an AUC of 0.844 (95% CI 0.793 to 0.894) in the testing cohort. For the one-vs-rest question of LRT vs HRT + TC, the sensitivity, specificity, and accuracy reached 80.00%, 63.41%, and 71.05%, respectively. For HRT vs LRT + TC, they reached 60.53%, 78.95%, and 69.74%. For TC vs LRT + HRT they reached 33.33%, 98.63%, and 96.05%, respectively. Compared with the radiomic models, superior diagnostic efficacy was demonstrated for most clinical-radiomics models, and the AUC of the Bernoulli Naive Bayes model was significantly improved. Radiologist2's assessment achieved a higher AUC of 0.813 (95% CI: 0.756-0.8761) than other radiologists, which was slightly lower than the SVM clinical-radiomic model. Combined with other evaluation indicators, SVM, as the best ML model, demonstrated the potential of predicting the simplified risk categorization of TETs with superior predictive performance to that of radiologists' assessment. CONCLUSION: Most of the ML models are promising in predicting the simplified TETs risk categorization with superior efficacy to that of radiologists' assessment, especially the SVM models, demonstrated the integration of ML with NECT may be valuable in aiding the diagnosis and treatment planning.

Differentiation Between Primary Central Nervous System Lymphoma and Atypical Glioblastoma Based on MRI Morphological Feature and Signal Intensity Ratio: A Retrospective Multicenter Study.

OBJECTIVES: To investigate the value of morphological feature and signal intensity ratio (SIR) derived from conventional magnetic resonance imaging (MRI) in distinguishing primary central nervous system lymphoma (PCNSL) from atypical glioblastoma (aGBM). METHODS: Pathology-confirmed PCNSLs (n = 93) or aGBMs (n = 48) from three institutions were retrospectively enrolled and divided into training cohort (n = 98) and test cohort (n = 43). Morphological features and SIRs were compared between PCNSL and aGBM. Using linear discriminant analysis, multiple models were constructed with SIRs and morphological features alone or jointly, and the diagnostic performances were evaluated via receiver operating characteristic (ROC) analysis. Areas under the curves (AUCs) and accuracies (ACCs) of the models were compared with the radiologists' assessment. RESULTS: Incision sign, T2 pseudonecrosis sign, reef sign and peritumoral leukomalacia sign were associated with PCNSL (training and overall cohorts, P < 0.05). Increased T1 ratio, decreased T2 ratio and T2/T1 ratio were predictive of PCNSL (all P < 0.05). ROC analysis showed that combination of morphological features and SIRs achieved the best diagnostic performance for differentiation of PCNSL and aGBM with AUC/ACC of 0.899/0.929 for the training cohort, AUC/ACC of 0.794/0.837 for the test cohort and AUC/ACC of 0.869/0.901 for the overall cohort, respectively. Based on the overall cohort, two radiologists could distinguish PCNSL from aGBM with AUC/ACC of 0.732/0.724 for radiologist A and AUC/ACC of 0.811/0.829 for radiologist B. CONCLUSION: MRI morphological features can help differentiate PCNSL from aGBM. When combined with SIRs, the diagnostic performance was better than that of radiologists' assessment.

Predicting Subtype of Growth Hormone Pituitary Adenoma based on Magnetic Resonance Imaging Characteristics.

PURPOSE: This study aimed to investigate the value of magnetic resonance (MR) characteristics in differentiating the subtypes of growth hormone pituitary adenomas. MATERIALS AND METHODS: The clinical and MR imaging data of 70 patients with growth hormone pituitary adenoma confirmed by surgery and pathology were retrospectively analyzed. The tumors were divided into dense granular (DG; 36 cases) and sparse granular subtypes (SG; 34 cases). The tumors' MR features were analyzed, including the mean and maximum diameters, T2 signal intensity, T2 relative signal intensity (rSI), homogeneity, enhancement degree, and invasiveness (Knosp grade). Mann-Whitney U test and χ2 test were used to analyze MR characteristics between the 2 groups. The independent predictors and predictive probabilities of tumor subtypes were obtained via a logistic regression model, and the efficacy was compared by receiver operating characteristic curve. RESULTS: The mean and maximum diameters of growth hormone adenoma in DG and SG were 1.77 versus 2.45 and 1.95 versus 3.00 cm (median, P < 0.05), respectively. There was a significant difference between the 2 groups in T2 signal intensity and rSI (P values were 0.02 and 0.001, respectively). Most DG adenomas (86.1%) appeared as hypointense on T2 images, and 38.2% of SG adenomas were hyperintense. There was no significant difference in tumor homogeneity (P = 0.622). A significant difference was found in the Knosp grade between the 2 subtypes (P = 0.004). In addition, the enhancement degree of SG adenomas was significantly higher than that of DG adenomas (P = 0.001). Logistic regression analysis showed that high T2 rSI value and marked contrast enhancement were independent predictors of the 2 subtypes, and the odds ratios were 4.811 and 4.649, respectively. The multivariate logistic model obtained relatively high predicting efficacy, and the area under the curve, sensitivity, and specificity were 0.765, 0.882, and 0.500, respectively. CONCLUSIONS: There are significant differences in tumor size, T2 signal intensity, T2 rSI, enhancement degree, and invasiveness between DG and SG adenomas. The logistic model based on the marked contrast enhancement and high T2 rSI value has an important value in predicting the subtype of growth hormone adenoma.

Differentiating thymoma, thymic carcinoma and lymphoma based on collagen fibre patterns with T2- and diffusion-weighted magnetic resonance imaging.

OBJECTIVES: The amount and distribution of intratumoural collagen fibre vary among different thymic tumours, which can be clearly detected with T2- and diffusion-weighted MR images. To explore the incidences of collagen fibre patterns (CFPs) among thymomas, thymic carcinomas and lymphomas on imaging, and to evaluate the efficacy and reproducibility of CFPs in differential diagnosis of thymic tumours. MATERIALS AND METHODS: Three hundred and ninety-eight patients with pathologically diagnosed thymoma, thymic carcinoma and lymphoma who underwent T2- and diffusion-weighted MR imaging were retrospectively enrolled. CFPs were classified into four categories: septum sign, patchy pattern, mixed pattern and no septum sign. The incidences of CFPs were compared among different thymic tumours, and the efficacy and reproducibility in differentiating the defined tumour types were analysed. RESULTS: There were significant differences in CFPs among thymomas, thymic squamous cell carcinomas (TSCCs), other thymic carcinomas and neuroendocrine tumours (OTC&NTs) and thymic lymphomas. Septum signs were found in 209 (86%) thymomas, which differed between thymomas and any other thymic neoplasms (all p < 0.005). The patchy, mixed patterns and no septum sign were mainly seen in TSCCs (80.3%), OTC&NTs (78.9%) and thymic lymphomas (56.9%), respectively. The consistency of different CFP evaluation between two readers was either good or excellent. CFPs achieved high efficacy in identifying the thymic tumours. CONCLUSION: The CFPs based on T2- and diffusion-weighted MR imaging were of great value in the differential diagnosis of thymic tumours. KEY POINTS: • Significant differences are found in intratumoural collagen fibre patterns among thymomas, thymic squamous cell carcinomas, other thymic carcinomas and neuroendocrine tumours and thymic lymphomas. • The septum sign, patchy pattern, mixed pattern and no septum sign are mainly seen in thymomas (86%), thymic squamous cell carcinomas (80.3%), other thymic carcinomas and neuroendocrine tumours (79%) and thymic lymphomas (57%), respectively. • The collagen fibre patterns have high efficacy and reproducibility in differentiating thymomas, thymic squamous cell carcinomas and thymic lymphomas.

Usefulness of the Texture Signatures Based on Multiparametric MRI in Predicting Growth Hormone Pituitary Adenoma Subtypes.

OBJECTIVE: To explore the usefulness of texture signatures based on multiparametric magnetic resonance imaging (MRI) in predicting the subtypes of growth hormone (GH) pituitary adenoma (PA). METHODS: Forty-nine patients with GH-secreting PA confirmed by the pathological analysis were included in this retrospective study. Texture parameters based on T1-, T2-, and contrast-enhanced T1-weighted images (T1C) were extracted and compared for differences between densely granulated (DG) and sparsely granulated (SG) somatotroph adenoma by using two segmentation methods [region of interest 1 (ROI1), excluding the cystic/necrotic portion, and ROI2, containing the whole tumor]. Receiver operating characteristic (ROC) curve analysis was performed to determine the differentiating efficacy. RESULTS: Among 49 included patients, 24 were DG and 25 were SG adenomas. Nine optimal texture features with significant differences between two groups were obtained from ROI1. Based on the ROC analyses, T1WI signatures from ROI1 achieved the highest diagnostic efficacy with an AUC of 0.918, the accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were 85.7, 72.0, 100.0, 100.0, and 77.4%, respectively, for differentiating DG from SG. Comparing with the T1WI signature, the T1C signature obtained relatively high efficacy with an AUC of 0.893. When combining the texture features of T1WI and T1C, the radiomics signature also had a good performance in differentiating the two groups with an AUC of 0.908. In addition, the performance got in all the signatures from ROI2 was lower than those in the corresponding signature from ROI1. CONCLUSION: Texture signatures based on MR images may be useful biomarkers to differentiate subtypes of GH-secreting PA patients.

Classification of Prostate Transitional Zone Cancer and Hyperplasia Using Deep Transfer Learning From Disease-Related Images.

Purpose The diagnosis of prostate transition zone cancer (PTZC) remains a clinical challenge due to their similarity to benign prostatic hyperplasia (BPH) on MRI. The Deep Convolutional Neural Networks (DCNNs) showed high efficacy in diagnosing PTZC on medical imaging but was limited by the small data size. A transfer learning (TL) method was combined with deep learning to overcome this challenge. Materials and methods A retrospective investigation was conducted on 217 patients enrolled from our hospital database (208 patients) and The Cancer Imaging Archive (nine patients). Using T2-weighted images (T2WIs) and apparent diffusion coefficient (ADC) maps, DCNN models were trained and compared between different TL databases (ImageNet vs. disease-related images) and protocols (from scratch, fine-tuning, or transductive transferring). Results PTZC and BPH can be classified through traditional DCNN. The efficacy of TL from natural images was limited but improved by transferring knowledge from the disease-related images. Furthermore, transductive TL from disease-related images had comparable efficacy to the fine-tuning method. Limitations include retrospective design and a relatively small sample size. Conclusion Deep TL from disease-related images is a powerful tool for an automated PTZC diagnostic system. In developing regions where only conventional MR scans are available, the accurate diagnosis of PTZC can be achieved via transductive deep TL from disease-related images.

Differentiation of Pseudoprogression from True Progressionin Glioblastoma Patients after Standard Treatment: A Machine Learning Strategy Combinedwith Radiomics Features from T1-weighted Contrast-enhanced Imaging.

BACKGROUND: Based on conventional MRI images, it is difficult to differentiatepseudoprogression from true progressionin GBM patients after standard treatment, which isa critical issue associated with survival. The aim of this study was to evaluate the diagnostic performance of machine learning using radiomics modelfrom T1-weighted contrast enhanced imaging(T1CE) in differentiating pseudoprogression from true progression after standard treatment for GBM. METHODS: Seventy-sevenGBM patients, including 51 with true progression and 26 with pseudoprogression,who underwent standard treatment and T1CE, were retrospectively enrolled.Clinical information, including sex, age, KPS score, resection extent, neurological deficit and mean radiation dose, were also recorded collected for each patient. The whole tumor enhancementwas manually drawn on the T1CE image, and a total of texture 9675 features were extracted and fed to a two-step feature selection scheme. A random forest (RF) classifier was trained to separate the patients by their outcomes.The diagnostic efficacies of the radiomics modeland radiologist assessment were further compared by using theaccuracy (ACC), sensitivity and specificity. RESULTS: No clinical features showed statistically significant differences between true progression and pseudoprogression.The radiomic classifier demonstrated ACC, sensitivity, and specificity of 72.78%(95% confidence interval [CI]: 0.45,0.91), 78.36%(95%CI: 0.56,1.00) and 61.33%(95%CI: 0.20,0.82).The accuracy, sensitivity and specificity of three radiologists' assessment were66.23%(95% CI: 0.55,0.76), 61.50%(95% CI: 0.43,0.78) and 68.62%(95% CI: 0.55,0.80); 55.84%(95% CI: 0.45,0.66),69.25%(95% CI: 0.50,0.84) and 49.13%(95% CI: 0.36,0.62); 55.84%(95% CI: 0.45,0.66), 69.23%(95% CI: 0.50,0.84) and 47.06%(95% CI: 0.34,0.61), respectively. CONCLUSION: T1CE-based radiomics showed better classification performance compared with radiologists' assessment.The radiomics modelwas promising in differentiating pseudoprogression from true progression.

MR imaging of thymomas: a combined radiomics nomogram to predict histologic subtypes.

OBJECTIVES: Accurately predicting the WHO classification of thymomas is urgently needed to optimize individualized therapeutic strategies. We aimed to develop and validate a combined radiomics nomogram for personalized prediction of histologic subtypes in patients with thymomas. METHODS: A total of 182 thymoma patients were divided into training (n = 128) and test (n = 54) cohorts. Radiomics features were extracted from T2-weighted, T2-weighted fat suppression, and diffusion-weighted images to establish a radiomics signature in the training cohort. Multivariate logistic regression analysis was used to develop a combined radiomics nomogram that incorporated clinical, conventional MR imaging variables, apparent diffusion coefficient (ADC) value, and radiomics signature. The efficacy of clinical, conventional MR imaging, or ADC model was also evaluated respectively. The performances of different models were compared by receiver operating characteristic analysis and Delong test. The discrimination, calibration, and clinical usefulness of the combined radiomics nomogram were assessed. RESULTS: The radiomics signature, consisting of 14 features, achieved favorable predictive efficacy in differentiating low-risk from high-risk thymomas, outperforming clinical, conventional MR imaging, and ADC models. The combined radiomics nomogram incorporating tumor shape, ADC value, and radiomics signature yielded the best performance (training cohort: area under the curve [AUC] = 0.946, test cohort: AUC = 0.878). The calibration curve and decision curve analysis indicated the clinical utility of the combined radiomics nomogram. CONCLUSIONS: The radiomics signature is a useful tool that can be used to predict histologic subtypes of thymomas. The combined radiomics nomogram improved the individualized subtype prediction in patients with thymomas. KEY POINTS: • Fourteen robust features were selected to develop a radiomics signature for preoperative prediction of thymoma subtype. • MRI-based radiomics signature can differentiate low-risk thymomas from high-risk thymomas with favorable predictive efficacy compared with clinical, conventional MR imaging, and ADC models. • Combined radiomics nomogram based on tumor shape, ADC value, and radiomics signature could improve the individualized subtype prediction in patients with thymomas.

Distinguishing brain inflammation from grade II glioma in population without contrast enhancement: a radiomics analysis based on conventional MRI.

PURPOSE: In populations without contrast enhancement, the imaging features of atypical brain parenchyma inflammations can mimic those of grade II gliomas. The aim of this study was to assess the value of the conventional MR-based radiomics signature in differentiating brain inflammation from grade II glioma. METHODS: Fifty-seven patients (39 patients with grade II glioma and 18 patients with inflammation) were divided into primary (n = 44) and validation cohorts (n = 13). Radiomics features were extracted from T1-weighted images (T1WI) and T2-weighted images (T2WI). Two-sample t-test and least absolute shrinkage and selection operator (LASSO) regression were adopted to select features and build radiomics signature models for discriminating inflammation from glioma. The predictive performance of the models was evaluated via area under the receiver operating characteristic curve (AUC) and compared with the radiologists' assessments. RESULTS: Based on the primary cohort, we developed T1WI, T2WI and combination (T1WI + T2WI) models for differentiating inflammation from glioma with 4, 8, and 5 radiomics features, respectively. Among these models, T2WI and combination models achieved better diagnostic efficacy, with AUC of 0.980, 0.988 in primary cohort and that of 0.950, 0.925 in validation cohort, respectively. The AUCs of radiologist 1's and 2's assessments were 0.661 and 0.722, respectively. CONCLUSION: The signature based on radiomics features helps to differentiate inflammation from grade II glioma and improved performance compared with experienced radiologists, which could potentially be useful in clinical practice.

Amide Proton Transfer Imaging in Predicting Isocitrate Dehydrogenase 1 Mutation Status of Grade II/III Gliomas Based on Support Vector Machine.

BACKGROUND: To compare the efficacies of univariate and radiomics analyses of amide proton transfer weighted (APTW) imaging in predicting isocitrate dehydrogenase 1 (IDH1) mutation of grade II/III gliomas. METHODS: Fifty-nine grade II/III glioma patients with known IDH1 mutation status were prospectively included (IDH1 wild type, 16; IDH1 mutation, 43). A total of 1044 quantitative radiomics features were extracted from APTW images. The efficacies of univariate and radiomics analyses in predicting IDH1 mutation were compared. Feature values were compared between two groups with independent t-test and receiver operating characteristic (ROC) analysis was applied to evaluate the predicting efficacy of each feature. Cases were randomly assigned to either the training (n = 49) or test cohort (n = 10) for the radiomics analysis. Support vector machine with recursive feature elimination (SVM-RFE) was adopted to select the optimal feature subset. The adverse impact of the imbalance dataset in the training cohort was solved by synthetic minority oversampling technique (SMOTE). Subsequently, the performance of SVM model was assessed on both training and test cohort. RESULTS: As for univariate analysis, 18 features were significantly different between IDH1 wild-type and mutant groups (P < 0.05). Among these parameters, High Gray Level Run Emphasis All Direction offset 8 SD achieved the biggest area under the curve (AUC) (0.769) with the accuracy of 0.799. As for radiomics analysis, SVM model was established using 19 features selected with SVM-RFE. The AUC and accuracy for IDH1 mutation on training set were 0.892 and 0.952, while on the testing set were 0.7 and 0.84, respectively. CONCLUSION: Radiomics strategy based on APT image features is potentially useful for preoperative estimating IDH1 mutation status.

Better efficacy in differentiating WHO grade II from III oligodendrogliomas with machine-learning than radiologist's reading from conventional T1 contrast-enhanced and fluid attenuated inversion recovery images.

BACKGROUND: The medical imaging to differentiate World Health Organization (WHO) grade II (ODG2) from III (ODG3) oligodendrogliomas still remains a challenge. We investigated whether combination of machine leaning with radiomics from conventional T1 contrast-enhanced (T1 CE) and fluid attenuated inversion recovery (FLAIR) magnetic resonance imaging (MRI) offered superior efficacy. METHODS: Thirty-six patients with histologically confirmed ODGs underwent T1 CE and 33 of them underwent FLAIR MR examination before any intervention from January 2015 to July 2017 were retrospectively recruited in the current study. The volume of interest (VOI) covering the whole tumor enhancement were manually drawn on the T1 CE and FLAIR slice by slice using ITK-SNAP and a total of 1072 features were extracted from the VOI using 3-D slicer software. Random forest (RF) algorithm was applied to differentiate ODG2 from ODG3 and the efficacy was tested with 5-fold cross validation. The diagnostic efficacy of radiomics-based machine learning and radiologist's assessment were also compared. RESULTS: Nineteen ODG2 and 17 ODG3 were included in this study and ODG3 tended to present with prominent necrosis and nodular/ring-like enhancement (P < 0.05). The AUC, ACC, sensitivity, and specificity of radiomics were 0.798, 0.735, 0.672, 0.789 for T1 CE, 0.774, 0.689, 0.700, 0.683 for FLAIR, as well as 0.861, 0.781, 0.778, 0.783 for the combination, respectively. The AUCs of radiologists 1, 2 and 3 were 0.700, 0.687, and 0.714, respectively. The efficacy of machine learning based on radiomics was superior to the radiologists' assessment. CONCLUSIONS: Machine-learning based on radiomics of T1 CE and FLAIR offered superior efficacy to that of radiologists in differentiating ODG2 from ODG3.

MRI Radiomics Analysis for Predicting the Pathologic Classification and TNM Staging of Thymic Epithelial Tumors: A Pilot Study.

OBJECTIVE. The purpose of this study was to explore the performance of MRI radiomics in predicting the pathologic classification and TNM staging of thymic epithelial tumors (TETs). MATERIALS AND METHODS. Clinical and MRI data for 189 patients with TETs were retrospectively collected. A total of 2088 radiomics features were extracted from T2-weighted images and T2-weighted fat-suppressed (FS) images. With the use of a support vector machine with recursive feature elimination, the optimal feature subsets were selected and used to construct two predictive models for pathologic classification and TNM staging. In multivariable logistic regression analysis, we incorporated the radiomics model, conventional MRI findings, and clinical variables to develop a radiomics nomogram for predicting risk stratification of advanced TETs. RESULTS. Of the extracted features, 125 features were selected to construct the radiomics model for predicting pathologic classification, and 69 features were selected to construct the radiomics model for predicting TNM staging. The models achieved AUC values of 0.880 and 0.948 in the training cohort and 0.771 and 0.908 in the test cohort, respectively, for distinguishing among low-risk thymomas, high-risk thymomas, and thymic carcinomas and differentiating between early-stage and advanced-stage TETs. The radiomics model, symptom, and pericardial effusion constituted a radiomics nomogram, with an AUC value of 0.967 (95% CI, 0.891-0.989) in the training cohort and 0.957 (95% CI, 0.842-0.974) in the test cohort. CONCLUSION. MRI radiomics analysis has the potential to differentiate the pathologic classification and TNM staging of TETs. A radiomics nomogram provides a useful tool for in dividualized prediction of the risk of advanced-stage TET before a patient undergoes treatment.

Perfusion, Diffusion, Or Brain Tumor Barrier Integrity: Which Represents The Glioma Features Best?

PURPOSE: This study aims to incorporate informative histogram indicator analyses and advanced multimodal MRI parameters to differentiate low-grade gliomas (LGGs) from high-grade gliomas (HGGs) and to explore the features associated with patients' survival. PATIENTS AND METHODS: A total of 120 patients with pathologically confirmed LGGs or HGGs receiving conventional and advanced MRI such as three-dimensional arterial spin labeling (3D-ASL), intravoxel incoherent motion-diffusion weighted imaging (IVIM-DWI), and dynamic contrast-enhanced MRI (DCE-MRI) were included. The mean and histogram indicators from advanced MRI were calculated from the entire tumor. The efficacies of a single indicator or multiple parameters were tested in distinguishing HGGs from LGGs and predicting patients' survival. Receiver operating characteristic (ROC) curve and multivariable stepwise logistic regression were used to evaluate the diagnostic efficacies. Leave-one-out cross-validation was further used to validate the accuracy of the parameter sets in glioma grading. Log-rank test using the Kaplan-Meier curve was utilized to predict patients' survival. RESULTS: Overall, parameters from DCE-MRI performed better than those from 3D-ASL or IVIM-DWI in both glioma grading and survival prediction. The histogram metrics of Ve were demonstrated to have higher accuracies (the accuracies for Extended Tofts_Ve mean and Extended Tofts_Ve median were 68.33% and 71.67%, respectively, while those for the Incremental_Ve mean and Incremental_Ve 75th were 68.33% and 72.50%, respectively) in grading LGGs from HGGs. The combination of Tofts_Ve histogram metrics was the one with the highest accuracy (81.67%) and area under ROC curve (AUC = 0.840). On the other hand, Patlak_Ktrans 95th (AUC = 0.9265) and Extended Tofts_Ve 95th (AUC = 0.9154) performed better than their corresponding means (Patlak_Ktrans mean: AUC = 0.9118 and Extended Tofts_Ve mean: AUC = 0.9044) in predicting patients' overall survival (OS) at 18-month follow-up. CONCLUSION: DCE-MRI-derived histogram features from the entire tumor were promising metrics for glioma grading and OS prediction. Combining single modal histogram features improved glioma grading. TRIAL REGISTRATION: NCT02622620.

Incidence and radiological pattern of eosinophilic granuloma: a retrospective study in a Chinese tertiary hospital.

BACKGROUND: The incidence and radiological patterns of eosinophilic granuloma (EG) in China is not clear. We described the incidence, presentation, and imaging characteristics of Chinese EG patients in a tertiary hospital. METHODS: A retrospective chart review was performed from January 2004 to October 2017 at a single tertiary general hospital. Seventy-six patients were pathologically identified as EG. Besides, 60 patients with preoperative imaging diagnosis of "EG" were analyzed to reveal the radiological patterns and their diagnostic power. RESULTS: Fifty-three male and 23 female EG patients with a mean age of 18.1 ± 16.7 years (range 1-58 years) were retrospectively included. Significant differences were observed in gender (male to female = 2.3:1) and age (the highest incidence at the age of 0~5 years) for EG. EG predominantly involved the skeletal system: flat bones (31.43%) > irregular bones (24.76%) > long bones (22.86%) > other organs (20.95%). No obvious relationships between season, biochemical markers, and EG incidence were observed. The common presenting symptoms were pain followed with local mass, and most patients underwent surgical resection. Among 60 imagingly diagnosed "EG" patients from April 2009 to October 2017, only 22 were with histological confirmation. The correct diagnosis rates were 37.1% (13 out of 35), 16.7% (5 out of 30), and 22.2% (8 out of 36) for plain radiography, computed tomography (CT), and magnetic resonance imaging (MRI), respectively. CONCLUSIONS: Chinese EG has a varied presentation, age distribution, and gender difference. EG diagnosis is still based on biopsy or histopathology instead of imaging techniques.

Predicting pathological subtypes and stages of thymic epithelial tumors using DWI: value of combining ADC and texture parameters.

OBJECTIVES: To explore the value of combining apparent diffusion coefficients (ADC) and texture parameters from diffusion-weighted imaging (DWI) in predicting the pathological subtypes and stages of thymic epithelial tumors (TETs). METHODS: Fifty-seven patients with TETs confirmed by pathological analysis were retrospectively enrolled. ADC values and optimal texture feature parameters were compared for differences among low-risk thymoma (LRT), high-risk thymoma (HRT), and thymic carcinoma (TC) by one-way ANOVA, and between early and advanced stages of TETs were tested using the independent samples t test. Receiver operating characteristic (ROC) curve analysis was performed to determine the differentiating efficacy. RESULTS: The ADC values in LRT and HRT were significantly higher than the values in TC (p = 0.004 and 0.001, respectively), also in early stage, values were significantly higher than ones in advanced stage of TETs (p < 0.001). Among all texture parameters analyzed in order to differentiate LRT from HRT and TC, the V312 achieved higher diagnostic efficacy with an AUC of 0.875, and combination of ADC and V312 achieved the highest diagnostic efficacy with an AUC of 0.933, for differentiating the LRT from HRT and TC. Furthermore, combination of ADC and V1030 achieved a relatively high differentiating ability with an AUC of 0.772, for differentiating early from advanced stages of TETs. CONCLUSIONS: Combination of ADC and DWI texture parameters improved the differentiating ability of TET grades, which could potentially be useful in clinical practice regarding the TET evaluation before treatment. KEY POINTS: • DWI texture analysis is useful in differentiating TET subtypes and stages. • Combination of ADC and DWI texture parameters may improve the differentiating ability of TET grades. • DWI texture analysis could potentially be useful in clinical practice regarding the TET evaluation before treatment.

Optimizing Texture Retrieving Model for Multimodal MR Image-Based Support Vector Machine for Classifying Glioma.

BACKGROUND: Accurate glioma grading plays an important role in patient treatment. PURPOSE: To investigate the influence of varied texture retrieving models on the efficacy of grading glioma with support vector machine (SVM). STUDY TYPE: Retrospective. POPULATION: In all, 117 glioma patients including 25, 29, and 63 grade II, III, and IV gliomas, respectively, based on WHO 2007. FIELD STRENGTH/SEQUENCE: 3.0T MRI/ T1 WI, T2 fluid-attenuated inversion recovery, contrast enhanced T1 , arterial spinal labeling, diffusion-weighted imaging (0, 30, 50, 100, 200, 300, 500, 800, 1000, 1500, 2000, 3000, and 3500 sec/mm2 ), and dynamic contrast-enhanced. ASSESSMENT: Texture attributes from 30 parametric maps were retrieved using four models, including Global, gray-level co-occurrence matrix (GLCM), gray-level run-length matrix (GLRLM), and gray-level size-zone matrix (GLSZM). Attributes derived from varied models were input into radial basis function SVM (RBF-SVM) combined with attribute selection using SVM-recursive feature elimination (SVM-RFE). The SVM model was trained and established with 80% randomly selected data of each category using 10-fold crossvalidation. The model performance was further tested using the remaining 20% data. STATISTICAL TESTS: Ten-fold crossvalidation was used to validate the model performance. RESULTS: Based on 30 parametric maps, 90, 240, 390, or 390 texture attributes were retrieved using the Global, GLCM, GLRLM, or GLSZM model, respectively. SVM-RFE was able to reduce attribute redundancy as well as improve RBF-SVM performance. Training data were oversampled by applying the Synthetic Minority Oversampling Technique (SMOTE) method to overcome the data imbalance problem; test results were able to further demonstrate the classifying performance of the final models. GLSZM using gray-level 64 was the optimal model to retrieve powerful image texture attributes to produce enough classifying power with an accuracy / area under the curve of 0.760/0.867 for the training and 0.875/0.971 for the independent test. Fifteen attributes were selected with SVM-RFE to provide comparable classifying efficacy. DATA CONCLUSION: When using image textures-based SVM classification of gliomas, the GLSZM model in combination with gray-level 64 and attribute selection may be an optimized solution. LEVEL OF EVIDENCE: 2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;49:1263-1274.

Glioma Grading on Conventional MR Images: A Deep Learning Study With Transfer Learning.

Background: Accurate glioma grading before surgery is of the utmost importance in treatment planning and prognosis prediction. But previous studies on magnetic resonance imaging (MRI) images were not effective enough. According to the remarkable performance of convolutional neural network (CNN) in medical domain, we hypothesized that a deep learning algorithm can achieve high accuracy in distinguishing the World Health Organization (WHO) low grade and high grade gliomas. Methods: One hundred and thirteen glioma patients were retrospectively included. Tumor images were segmented with a rectangular region of interest (ROI), which contained about 80% of the tumor. Then, 20% data were randomly selected and leaved out at patient-level as test dataset. AlexNet and GoogLeNet were both trained from scratch and fine-tuned from models that pre-trained on the large scale natural image database, ImageNet, to magnetic resonance images. The classification task was evaluated with five-fold cross-validation (CV) on patient-level split. Results: The performance measures, including validation accuracy, test accuracy and test area under curve (AUC), averaged from five-fold CV of GoogLeNet which trained from scratch were 0.867, 0.909, and 0.939, respectively. With transfer learning and fine-tuning, better performances were obtained for both AlexNet and GoogLeNet, especially for AlexNet. Meanwhile, GoogLeNet performed better than AlexNet no matter trained from scratch or learned from pre-trained model. Conclusion: In conclusion, we demonstrated that the application of CNN, especially trained with transfer learning and fine-tuning, to preoperative glioma grading improves the performance, compared with either the performance of traditional machine learning method based on hand-crafted features, or even the CNNs trained from scratch.

Iodine Quantification Using Dual-Energy Computed Tomography for Differentiating Thymic Tumors.

The aim of the study was to explore the efficacy of iodine quantification with dual-energy computed tomography (DECT) in differentiating thymoma, thymic carcinoma, and thymic lymphoma. MATERIALS AND METHODS: Fifty-seven patients with pathologically confirmed low-risk thymoma (n = 16), high-risk thymoma (n = 15), thymic carcinoma (n = 14), and thymic lymphoma (n = 12) underwent chest contrast-enhanced DECT scan were enrolled in this study. Tumor DECT parameters including iodine-related Hounsfield unit (IHU), iodine concentration (IC), mixed HU (MHU), and iodine ratio in dual phase, slope of energy spectral HU curve (λ), and virtual noncontrast (VNC) were compared for differences among 4 groups by one-way analysis of variance. Receiver operating characteristic curve was used to determine the efficacy for differentiating the low-risk thymoma from other thymic tumor by defined parameters. RESULTS: According to quantitative analysis, dual-phase IHU, IC, and MHU values in patients with low-risk thymoma were significantly increased compared with patients with high-risk thymoma, thymic carcinoma, and thymic lymphoma (P < 0.05/4).The venous phase IHU value yielded the highest performance with area under the curve of 0.893, 75.0% sensitivity, and 89.7% specificity for differentiating the low-risk thymomas from high-risk thymomas or thymic carcinoma at the cutoff value of 34.3 HU. When differentiating low-risk thymomas from thymic lymphoma, the venous phase IC value obtained the highest diagnostic efficacy with the area under the curve of 0.969, and sensitivity, specificity, and cutoff value were 87.5%, 100.0%, and 1.25 mg/mL, respectively. CONCLUSIONS: Iodine quantification with DECT may be useful for differentiating the low-risk thymomas from other thymic tumors.