Search

The objective of this study was to predict hematoma expansion (HE) by radiomic models based on different machine learning methods and determine the best radiomic model through the comparison. A total of 108 patients with intracerebral hemorrhage were retrospectively evaluated. Images of baseline non-contrast computed tomography (NCCT) and follow-up NCCT scan within 24 h were retrospectively reviewed. An HE was defined as a volume increase of more than 33% or an increase greater than 12.5 mL from the volume of the baseline NCCT. Texture parameters of the baseline NCCT images were selected by the least absolute shrinkage and selection operator (LASSO) regression. We used support vector machine (SVM), decision tree (DT), conditional inference trees (CIT), random forest (RF), k‑nearest neighbors (KNN), back-propagation neural network (BPNet) and Bayes to build models. Receiver operating characteristic (ROC) analysis and decision curve analysis (DCA) was performed and compared among models. Every model had a relatively high AUC (all > 0.75), SVM and KNN had the highest AUC of 0.91. There were significant differences between SVM and CIT (Z > 2.266, p = 0.02345), KNN and CIT (Z = 2.4834, p = 0.01301), RF and CIT (Z = 2.6956, p = 0.007027), KNN and BPNet (Z = 2.0122, p = 0.0442), RF and BPNet (Z = 1.9793, p = 0.04778). There was no significant difference among SVM, DT, RF, KNN and Bayes (p > 0.05). The SVM obtained the largest net benefit when the threshold probability was less than 0.33, while KNN obtained the largest net benefit when the threshold probability was greater than 0.33. Combined with ROC and DCA, SVM and KNN performed better in all the models for predicting HE. Radiomic models based on different machine learning methods can be used to predict HE and the models generated by SVM and KNN performed best.

The objective of this study was to investigate whether computed tomography texture analysis can be used to differentiate papillary renal cell carcinoma (PRCC) subtypes. Sixty-two PRCC tumors were retrospectively evaluated, with 30 type 1 tumors and 32 type 2 tumors. Texture parameters quantified from three-phase contrast-enhanced CT images were compared with least absolute shrinkage and selection operator (LASSO) regression. Receiver operating characteristic (ROC) analysis was performed, and the area under the ROC curve (AUC) was calculated for each parameter. The selected texture parameters of each phase were used to generate support vector machine (SVM) classifiers. Decision curve analysis (DCA) of the classification was performed. The two texture parameters with the top two AUC values were – 333–7 Correlation (AUC = 0.772) and 45–7 Entropy (AUC = 0.753) in the corticomedullary phase, 333–4 Correlation (AUC = 0.832) and 45–7 Entropy (AUC = 0.841) in the nephrographic phase, and 135–7 Entropy (AUC = 0.858) and – 333–1 InformationMeasureCorr2 (AUC = 0.849) in the excretory phase. Entropy and Correlation have a high correlation with the two types of PRCC and are increased in type 2 PRCC. A model incorporating the texture parameters with the top two AUC values in each phase produced an AUC of 0.922 with an accuracy of 84% (sensitivity = 89% and specificity = 80%). The nephrographic-phase model and the model combining the texture parameters of the three phases can differentiate the two types with the largest net benefit. Computed tomography texture analysis can be used to distinguish type 2 PRCC from type 1 with high accuracy, which may be clinically important.

Background Meningioma subtypes are one of the most common key points to the treatment and prognosis of patients. The purpose of this study was to investigate the differential diagnostic value of radiomics features on meningioma. Methods A total of 241 patients with meningioma who had undergone tumor resection were randomly selected including 80 with meningothelial meningioma, 80 with fibrous meningioma, and 81 with transitional meningioma. These meningiomas were divided into 4 groups including: meningothelial versus fibrous (group 1), fibrous versus transitional (group 2), meningothelial versus transitional (group 3), and meningothelial versus fibrous versus transitional (group 4). All patients were examined using the same magnetic resonance scanner (GE 3.0 T) and the preoperative contrast-enhanced T1-weighted images were available. Radiomics features from the contrast-enhanced T1-weighted images of 241 patients were evaluated by 2 experienced radiology specialists. Results A total of 385 radiomics features were extracted from the images of each patient. Several preprocessing methods were applied on the radiomics dataset to reduce the redundancy and highlight differences between different meningioma before the Fisher discrimination analysis was adopted and leave one out cross validation methods were used for the model validation. The differentiation accuracies of the Fisher discriminant analysis model for groups 1, 2, 3, and 4 were 99.4%, 98.8%, 100% and 100%, respectively; leave one out cross validation method was achieved for group 1, 2, 3, and 4 with the accuracies of 91.3%, 95.0%, 100%, and 94.2%, respectively. Conclusions Radiomics features and the combined Fisher discriminant analysis could provide satisfactory performance in the preoperative differential diagnosis of meningioma subtypes and enable the potential ability for clinical application.

Purpose. To investigate whether the radiomics analysis of MR imaging in the hepatobiliary phase (HBP) can be used to predict microvascular invasion (MVI) in patients with hepatocellular carcinoma (HCC). Method. A total of 130 patients with HCC, including 80 MVI-positive patients and 50 MVI-negative patients, who underwent MR imaging with Gd-EOB-DTPA were enrolled. Least absolute shrinkage and selection operator (LASSO) regression was applied to select radiomics parameters derived from MR images obtained in the HBP 5 min, 10 min, and 15 min images. The selected features at each phase were adopted into support vector machine (SVM) classifiers to establish models. Multiple comparisons of the AUCs at each phase were performed by the Delong test. The decision curve analysis (DCA) was used to analyze the classification of MVI-positive and MVI-negative patients. Results. The most predictive features between MVI-positive and MVI-negative patients included 9, 8, and 14 radiomics parameters on HBP 5 min, 10 min, and 15 min images, respectively. A model incorporating the selected features produced an AUC of 0.685, 0.718, and 0.795 on HBP 5 min, 10 min, and 15 min images, respectively. The predictive model for HBP 5 min, 10 min and 15 min showed no significant difference by the Delong test. DCA indicated that the predictive model for HBP 15 min outperformed the models for HBP 5 min and 10 min. Conclusions. Radiomics parameters in the HBP can be used to predict MVI, with the HBP 15 min model having the best differential diagnosis ability.

Purpose. The MRI features of epithelioid glioblastoma (eGBM) were analyzed. The apparent diffusion coefficient (ADC), MR perfusion-weighted imaging (PWI), and magnetic resonance spectroscopy (MRS) findings were quantitatively analyzed.Methods. The MRI images of 8 cases of eGBM were analyzed retrospectively. The location and edge, signal, peritumoral edema, adjacent meningeal invasion, and enhancement of the lesions were observed. The ADC value, relative cerebral blood volume (rCBV), relative cerebral blood flow (rCBF), and N-acetylaspartate/acetylcholine (NAA/Cho) value were analyzed.Results. Among the 8 patients, the tumors were mainly located in the temporal lobe ( n = 3 ), frontal lobe ( n = 3 ), and parietal lobe ( n = 2 ). The lesion boundary was clear in 6 cases and unclear in 2. The lesions were superficial in 5 cases and in the deep white matter in 3. Internal hemorrhage was observed in 4 cases. There was cystic necrosis in 7 cases, and only 1 case was solid without cystic necrosis. There was no edema around the lesion in 1 case, severe edema in 5, and moderate edema in 2. In 4 cases, the adjacent meninges were involved, and in 1 case, the ependyma was involved. Two patients developed leptomeningeal metastasis within 2 months after the operation. The average ADC value of the tumor parenchyma among all 8 patients was 7.15 × 10 − 4 m m 2 / s ,which was 17.6% lower than that of the contralateral side. The Cho/NAA metabolite ratio was 5.27 and 0.81 in the lesions of 2 patients. The rCBV was 3.51 ml/100 g and 3.32 ml/100 g of lesions in 2 patients; these values were 36% and 29% higher, respectively, than those of the contralateral side. The rCBF was 31.5 ml/100 g/min and 82.1 ml/100 g/min of lesions in two patients; these values were 49% and 203% higher, respectively, than those of the contralateral side.Conclusion. eGBM characteristics include a superficial location, easy cyst degeneration, easy necrosis and hemorrhage, and clear boundaries. It easily invades adjacent meninges and shows cerebrospinal fluid dissemination and metastasis. Combining new MR techniques, such as ADC values, PWI, and MRS, could be helpful for improving diagnostic accuracy.

Background: Glioblastoma (GB) is the most common primary malignant brain tumor and occurs predominantly in the white matter of the brain. MRI often shows an irregular mass with uneven internal signals, necrosis, surrounding edema and a space-occupying effect. The malignant degree of GB is high, and it often relapses after surgery. Case presentation: A 66-year-old male patient was admitted to our hospital due to sudden left superior strabismus and convulsions . During the patient’s first hospitalization, MRI showed only an abnormal cortical signal, and the diagnosis was viral encephalitis.After the treatment of antiepileptic, antiviral and relieving brain edema, the patient was discharged. Seven months later, he was admitted to the hospital again because of memory impairment and slow reaction speed. MRI showed that there was an obvious mass in the original lesion area, and this was diagnosed as glioblastoma. Postoperative pathology confirmed glioblastoma (WHO) grade IV. Conclusions: The clinical diagnosis, treatment and imaging manifestations of this case of GB are reported as follows to improve understanding of this type of GB, which only has cortical abnormal signals in the early stage, to reduce misdiagnosis.

Background. The prognosis of IDH1-mutant glioma is significantly better than that of wild-type glioma, and the preoperative identification of IDH mutations in glioma is essential for the formulation of surgical procedures and prognostic assessment. Purpose. To explore the value of a radiomic model based on preoperative-enhanced MR images in the assessment of the IDH1 genotype in high-grade glioma. Materials and Methods. A retrospective analysis was performed on 182 patients with high-grade glioma confirmed by surgical pathology between December 2012 and January 2019 in our hospital with complete preoperative brain-enhanced MR images, including 79 patients with an IDH1 mutation (45 patients with WHO grade III and 34 patients with WHO grade IV) and 103 patients with wild-type IDH1 (33 patients with WHO grade III and 70 patients with WHO grade IV). Patients were divided into a primary dataset and a validation dataset at a ratio of 7 : 3 using a stratified random sampling; radiomic features were extracted using A.K. (Analysis Kit, GE Healthcare) software and were initially reduced using the Kruskal-Wallis and Spearman analyses. Lasso was finally conducted to obtain the optimized subset of the feature to build the radiomic model, and the model was then tested with cross-validation. ROC (receiver operating characteristic curve) analysis was performed to evaluate the performance of the model. Results. The radiomic model showed good discrimination in both the primary dataset ( AUC = 0.87 , 95% CI: 0.754 to 0.855, ACC = 0.798 , sensitivity = 85.5 % , specificity = 75.4 % , positive predictive value = 0.734 , and negative predictive value = 0.867 ) and the validation dataset ( AUC = 0.86 , 95% CI: 0.690 to 0.913, ACC = 0.789 , sensitivity = 91.3 % , specificity = 69.0 % , positive predictive value = 0.700 , and negative predictive value = 0.909 ). Conclusion. The radiomic model, based on the preoperative-enhanced MR, can effectively predict the IDH1 genotype in high-grade glioma.

We used magnetic resonance perfusion weighted imaging and pathological evaluation to examine different stages of radiation-induced brain injury and to investigate the correlation between the relative cerebral blood volume (rCBV) ratio and vascular endothelial growth factor (VEGF). Thirty adult rats were randomly divided into 2 groups: control and radiation group. The control group was not subjected to irradiation. The irradiation group rats were examined by magnetic resonance imaging and magnetic resonance perfusion weighted imaging at 1, 3, 6, 9, and 12 months after radiation treatment. We measured the rCBV, mean transit time, and time to peak. Hematoxylin and eosin staining, immunohistochemical staining, and electron microscopy were performed. VEGF absorbance was evaluated by immunohistochemical staining. Compared with the control group, the differences in rCBV, mean transit time, time to peak, and VEGF absorbance after 3 months were statistically significant (P < 0.05). rCBV was positively correlated with VEGF (r = 0.94, P < 0.05). Magnetic resonance perfusion weighted imaging can reflect pathophysiological changes in brain tissue after irradiation. Decreased expression of VEGF plays a critical role in the pathogenesis of radiation-induced brain injury.