MR-based Radiomics Research Articles

MR-Based Radiomics for Differential Diagnosis between Cystic Pituitary Adenoma and Rathke Cleft Cyst.

Background It is often tricky to differentiate cystic pituitary adenoma from Rathke cleft cyst with visual inspection because of similar MRI presentations between them. We aimed to design an MR-based radiomics model for improving differential diagnosis between them. Methods Conventional diagnostic MRI data (T1-,T2-, and postcontrast T1-weighted MR images) were obtained from 215 pathologically confirmed patients (105 cases with cystic pituitary adenoma and the other 110 cases with Rathke cleft cyst) and were divided into training (n = 172) and test sets (n = 43). MRI radiomics features were extracted from the imaging data, and semantic imaging features (n = 15) were visually estimated by two radiologists. Four classifiers were used to construct radiomics models through 5-fold crossvalidation after feature selection with least absolute shrinkage and selection operator. An integrated model by combining radiomics and semantic features was further constructed. The diagnostic performance was validated in the test set. Receiver operating characteristic curve was used to evaluate and compare the performance of the models at the background of diagnostic performance by radiologist. Results In test set, the combined radiomics and semantic model using ANN classifier obtained the best classification performance with an AUC of 0.848 (95% CI: 0.750-0.946), accuracy of 76.7% (95% CI: 64.1-89.4%), sensitivity of 73.9% (95% CI: 56.0-91.9%), and specificity of 80.0% (95% CI: 62.5-97.5%) and performed better than multiparametric model (AUC = 0.792, 95% CI: 0.674-0.910) or semantic model (AUC = 0.823, 95% CI: 0.705-0.941). The two radiologists had an accuracy of 69.8% and 74.4%, respectively, sensitivity of 69.6% and 73.9%, and specificity of 70.0% and 75.0%. Conclusions The MR-based radiomics model had technical feasibility and good diagnostic performance in the differential diagnosis between cystic pituitary adenoma and Rathke cleft cyst.

Comparable Performance of Deep Learning-Based to Manual-Based Tumor Segmentation in KRAS/NRAS/BRAF Mutation Prediction With MR-Based Radiomics in Rectal Cancer.

Radiomic features extracted from segmented tumor regions have shown great power in gene mutation prediction, while deep learning–based (DL-based) segmentation helps to address the inherent limitations of manual segmentation. We therefore investigated whether deep learning–based segmentation is feasible in predicting KRAS/NRAS/BRAF mutations of rectal cancer using MR-based radiomics. In this study, we proposed DL-based segmentation models with 3D V-net architecture. One hundred and eight patients’ images (T2WI and DWI) were collected for training, and another 94 patients’ images were collected for validation. We evaluated the DL-based segmentation manner and compared it with the manual-based segmentation manner through comparing the gene prediction performance of six radiomics-based models on the test set. The performance of the DL-based segmentation was evaluated by Dice coefficients, which are 0.878 ± 0.214 and 0.955 ± 0.055 for T2WI and DWI, respectively. The performance of the radiomics-based model in gene prediction based on DL-segmented VOI was evaluated by AUCs (0.714 for T2WI, 0.816 for DWI, and 0.887 for T2WI+DWI), which were comparable to that of corresponding manual-based VOI (0.637 for T2WI, P=0.188; 0.872 for DWI, P=0.181; and 0.906 for T2WI+DWI, P=0.676). The results showed that 3D V-Net architecture could conduct reliable rectal cancer segmentation on T2WI and DWI images. All-relevant radiomics-based models presented similar performances in KRAS/NRAS/BRAF prediction between the two segmentation manners.

Early risk-assessment of patients with nasopharyngeal carcinoma: the added prognostic value of MR-based radiomics

ObjectivesTo assess the additive prognostic value of MR-based radiomics in predicting progression-free survival (PFS) in patients with nasopharyngeal carcinoma (NPC) MethodsPatients newly diagnosed with non-metastatic NPC between June 2006 and October 2019 were retrospectively included and randomly grouped into training and test cohorts (7:3 ratio). Radiomic features (n=213) were extracted from T2-weighted and contrast-enhanced T1-weighted MRI. The patients were staged according to the 8th edition of American Joint Committee on Cancer Staging Manual. The least absolute shrinkage and selection operator was used to select the relevant radiomic features. Univariate and multivariate Cox proportional hazards analyses were conducted for PFS, yielding three different survival models (clinical, stage, and radiomic). The integrated time-dependent area under the curve (iAUC) for PFS was calculated and compared among different combinations of survival models, and the analysis of variance was used to compare the survival models. The prognostic performance of all models was validated using a test set with integrated Brier scores. ResultsThis study included 81 patients (training cohort=57; test cohort=24), and the mean PFS was 57.5 ± 43.6 months. In the training cohort, the prognostic performances of survival models improved significantly with the addition of radiomics to the clinical (iAUC, 0.72–0.80; p=0.04), stage (iAUC, 0.70–0.79; p=0.001), and combined models (iAUC, 0.76–0.81; p<0.001). In the test cohort, the radiomics and combined survival models were robustly validated for their ability to predict PFS. ConclusionIntegration of MR-based radiomic features with clinical and stage variables improved the prediction PFS in patients diagnosed with NPC.

An MR-based radiomics model for differentiation between hepatocellular carcinoma and focal nodular hyperplasia in non-cirrhotic liver

PurposeWe aimed to develop and validate a radiomics model for differentiating hepatocellular carcinoma (HCC) from focal nodular hyperplasia (FNH) in non-cirrhotic livers using Gd-DTPA contrast-enhanced magnetic resonance imaging (MRI).MethodsWe retrospectively enrolled 149 HCC and 75 FNH patients treated between May 2015 and May 2019 at our center. Patients were randomly allocated to a training (n=156) and validation set (n=68). In total, 2260 radiomics features were extracted from the arterial phase and portal venous phase of Gd-DTPA contrast-enhanced MRI. Using Max-Relevance and Min-Redundancy, random forest, least absolute shrinkage, and selection operator algorithm for dimensionality reduction, multivariable logistic regression was used to build the radiomics model. A clinical model and combined model were also established. The diagnostic performance of the models was compared.ResultsEight radiomics features were chosen for the radiomics model, and four clinical factors (age, sex, HbsAg, and enhancement pattern) were chosen for the clinical model. A combined model was built using the factors from the previous models. The classification accuracy of the combined model differentiated HCC from FNH in both the training and validation sets (0.956 and 0.941, respectively). The area under the receiver operating characteristic curve of the combined model was significantly better than that of the clinical model for both the training (0.984 vs. 0.937, p=0.002) and validation (0.972 vs. 0.903, p=0.032) sets.ConclusionsThe combined model provided a non-invasive quantitative method for differentiating HCC from FNH in non-cirrhotic liver with high accuracy. Our model may assist clinicians in the clinical decision-making process.

Comparison of MRI and CT for the Prediction of Microvascular Invasion in Solitary Hepatocellular Carcinoma Based on a Non-Radiomics and Radiomics Method: Which Imaging Modality Is Better?

Computed tomography (CT) and magnetic resonance imaging (MRI) are both capable of predicting microvascular invasion (MVI) in hepatocellular carcinoma (HCC). However, which modality is better is unknown. To intraindividually compare CT and MRI for predicting MVI in solitary HCC and investigate the added value of radiomics analyses. Retrospective. Included were 402 consecutive patients with HCC (training set:validation set=300:102). T2-weighted, diffusion-weighted, and contrast-enhanced T1-weighted imaging MRI at 3.0T and contrast-enhanced CT. CT- and MR-based radiomics signatures (RS) were constructed using the least absolute shrinkage and selection operator regression. CT- and MR-based radiologic (R) and radiologic-radiomics (RR) models were developed by univariate and multivariate logistic regression. The performance of the RS/models was compared between two modalities. To investigate the added value of RS, the performance of the R models was compared with the RR models in HCC of all sizes and 2-5 cm in size. Model performance was quantified by the area under the receiver operating characteristic curve (AUC) and compared using the Delong test. Histopathologic MVI was identified in 161 patients (training set:validation set=130:31). MRI-based RS/models tended to have a marginally higher AUC than CT-based RS/models (AUCs of CT vs. MRI, P: RS, 0.801 vs. 0.804, 0.96; R model, 0.809 vs. 0.832, 0.09; RR model, 0.835 vs. 0.872, 0.54). The improvement of RR models over R models in all sizes was not significant (P=0.21 at CT and 0.09 at MRI), whereas the improvement in 2-5 cm was significant at MRI (P < 0.05) but not at CT (P=0.16). CT and MRI had a comparable predictive performance for MVI in solitary HCC. The RS of MRI only had significant added value for predicting MVI in HCC of 2-5 cm. 3 TECHNICAL EFFICACY: Stage 2.

Radiomics signature of brain metastasis: prediction of EGFR mutation status.

To predict epidermal growth factor receptor (EGFR) mutation status in lung adenocarcinoma using MR-based radiomics signature of brain metastasis and explore the optimal MR sequence for prediction. Data from 52 patients with brain metastasis from lung adenocarcinoma (28 with mutant EGFR, 24 with wild-type EGFR) were retrospectively reviewed. Contrast-enhanced T1-weighted imaging (T1-CE), T2 fluid-attenuated inversion recovery (T2-FLAIR), T2WI, and DWI sequences were selected for radiomics features extraction. A total of 438 radiomics features were extracted from each MR sequence. All sequences were randomly divided into training and validation cohorts. The least absolute shrinkage selection operator was used to select informative features, a radiomics signature was built with the logistic regression model of the training cohort, and the radiomics signature performance was evaluated using the validation cohort and an independent testing data set. The radiomics signature built on 9 selected features showed good discrimination in both the training and validation cohorts for T2-FLAIR. The radiomics signature of T2-FLAIR yielded an AUC of 0.987, a classification accuracy of 0.991, sensitivity of 1.000, and specificity of 0.980 in the validation cohort. The AUC was 0.871 in the independent testing data set. The AUCs of our radiomics signature to differentiate exon 19 and exon 21 mutations were 0.529, 0.580, 0.645, and 0.406 for T1-CE, T2-FLAIR, T2WI, and DWI, respectively. We developed a T2-FLAIR radiomics signature that can be used as a noninvasive auxiliary tool for predicting EGFR mutation status in lung adenocarcinoma, which is helpful to guide therapeutic strategies. • MR-based radiomics signature of brain metastasis may help predict EGFR mutation status in lung adenocarcinoma, especially using T2-FLAIR. • Nine radiomics features extracted from T2-FLAIR sequence strongly correlate with EGFR mutation status. • Radiomics features reflect tumor heterogeneity through potential changes in tissue morphology caused by EGFR mutation.

Radiomics nomogram of MR: a prediction of cervical lymph node metastasis in laryngeal cancer

Objective: To establish and validate a radiomics nomogram based on MR for predicting cervical lymph node metastasis in laryngeal cancer. Methods: One hundred and seventeen patients with laryngeal cancer who underwent MR examinations and received open surgery and neck dissection between January 2016 and December 2019 were included in this study. All patients were randomly divided into a training cohort (n=89) and test cohort (n=28) using computer-generated random numbers. Clinical characteristics and MR were collected. Radiological features were extracted from the MR images. Enhanced T1 and T2WI were selected for radiomics analysis, and the volume of interest was manually segmented from the Huiyihuiying radiomics cloud platform. The variance analysis (ANOVA) and the least absolute shrinkage and selection operator (LASSO) algorithm were used to reduce the dimensionality of the radiomics features in the training cohort. Then, a radiomic signature was established. The clinical risk factors were screened by using ANOVA and multivariate logistic regression. A nomogram was generated using clinical risk factors and the radiomic signature. The calibration curve and receiver operator characteristic (ROC) curve were used to confirm the nomogram's performance in the training and test sets. The clinical usefulness of the nomogram was evaluated by decision curve analysis (DCA). Furthermore, a testing cohort was used to validate the model. Results: The radiomics signature consisted of 21 features, and the nomogram model included the radiomics signature and the MR-reported lymph node status. The model showed good calibration and discrimination. The model yielded areas under the ROC curve (AUC) in the training cohort, specificity, and sensitivity of 0.930, 0.930 and 0.875. In the test cohort, the model yielded AUC, specificity and sensitivity of 0.883, 0.889 and 0.800. DCA indicated that the nomogram model was clinically useful. Conclusion: The MR-based radiomics nomogram model may be used to predict cervical lymph node metastasis of laryngeal cancer preoperatively. MR-based radiomics could serve as a potential tool to help clinicians make an optimal clinical decision.

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

PurposeIn 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. MethodsFifty-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. ResultsBased 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. ConclusionThe 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.

The value of MR-based radiomics in identifying residual disease in patients with carcinoma in situ after cervical conization

Carcinoma in situ (CIS) of the uterine cervix is a precursor to cervical carcinoma. However, hysterectomy can be avoided in patients who can be treated by cone biopsy. Previous studies have shown that imaging-based approaches allow for the noninvasive visualization of cervical cancer, and radiomics has high accuracy in classifying cancer and predicting treatment outcome for different cancer types. To develop a magnetic resonance (MR)-based radiomics model for identifying residual disease in patients with CIS after cervical conization. Patients who had CIS after conization and finally underwent hysterectomy were collected to comprise a database to establish an imaging model for predicting the residual status after conization. Then, patients who opted for uterine preservation were classified as high-risk or low-risk patients according to the model. The disease-free survival was compared between the different risk groups using the Kaplan–Meier curve. The model built with the Boruta features outperformed the random forest model. Further validation with patients with uterine preservation showed that the patients classified as high risk were more likely to have tumor recurrence/residual disease in the follow-up period. In conclusion, radiomics can be used to identify residual disease in patients with CIS after cervical conization and could have the potential to predict recurrence in patients who opt for uterine preservation.

Radiomics may increase the prognostic value for survival in glioblastoma patients when combined with conventional clinical and genetic prognostic models.

To evaluate the additional prognostic value of multiparametric MR-based radiomics in patients with glioblastoma when combined with conventional clinical and genetic prognostic factors. In this single-center study, patients diagnosed with glioblastoma between October 2007 and December 2019 were retrospectively screened and grouped into training and test sets with a 7:3 distribution. Segmentations of glioblastoma using multiparametric MRI were performed automatically via a convolutional-neural network. Prognostic factors in the clinical model included age, sex, type of surgery/post-operative treatment, and tumor location; those in the genetic model included statuses of isocitrate dehydrogenase-1 mutation and O-6-methylguanine-DNA-methyltransferase promoter methylation. Univariate and multivariate Cox proportional hazards analyses were performed for overall survival (OS) and progression-free survival (PFS). Integrated time-dependent area under the curve (iAUC) for survival was calculated and compared between prognostic models via the bootstrapping method (performances were validated with prediction error curves). Overall, 120 patients were included (training set, 85; test set, 35). The mean OS and PFS were 25.5 and 18.6 months, respectively. The prognostic performances of multivariate models improved when radiomics was added to the clinical model (iAUC: OS, 0.62 to 0.73; PFS, 0.58 to 0.66), genetic model (iAUC: OS, 0.59 to 0.67; PFS, 0.59 to 0.65), and combined model (iAUC: OS, 0.65 to 0.73; PFS, 0.62 to 0.67). In the test set, the combined model (clinical, genetic, and radiomics) demonstrated robust validation for risk prediction of OS and PFS. Radiomics increased the prognostic value when combined with conventional clinical and genetic prognostic models for OS and PFS in glioblastoma patients. • CNN-based automatic segmentation of glioblastoma on multiparametric MRI was useful in extracting radiomic features. • Patients with glioblastoma with high-risk radiomics scores had poor overall survival (hazards ratio 8.33, p < 0.001) and progression-free survival (hazards ratio 3.76, p < 0.001). • MR-based radiomics improved the survival prediction when combined with clinical and genetic factors (overall and progression-free survival iAUC from 0.65 to 0.73 and 0.62 to 0.67, respectively; both p < 0.001).

Value of MR-based radiomics in differentiating uveal melanoma from other intraocular masses in adults

PurposeTo assess the performance of machine learning (ML)-based magnetic resonance imaging (MRI) radiomics analysis for discriminating between uveal melanoma (UM) and other intraocular masses. MethodsThis retrospective study analyzed 245 patients with intraocular masses (165 UMs and 80 other intraocular masses). Radiomics features were extracted from T1WI, T2WI, and contrast enhanced T1-weighted images (CET1WI), respectively. The intraclass correlation coefficient (ICC) was calculated to quantify the reproducibility of features. The training and test sets consisted of 195 and 50 cases. Least absolute shrinkage and selection operator (LASSO) regression method was employed for feature selection. The ML classifiers were logistic regression (LR), multilayer perceptron (MLP), and support vector machine (SVM). The performance of classifiers was evaluated by ROC analysis, and was compared to the performance of visual assessment by DeLong test. ResultsThe optimal radiomics feature set was 10, 15, 15, and 24 for T1W, T2W, CET1W, and joint T2W and CET1W images, respectively. The accuracy of T1WI, T2WI, CET1WI, and the joint T2WI and CET1WI models ranged from 72.0 %–78.0 %, from 79.6 %–81.6 %, from 74.0 %–82.0 %, and from 76.0 %–86.0 % in the test set. In the test set, the AUC for T1WI, T2WI, CET1WI, joint T2WI, and CET1WI models ranged from 0.775 to 0.829, 0.816 to 0.826, 0.836 to 0.861, and 0.870 to 0.877, respectively. In the combined model, the performance of ML classifiers was better than the performance of visual assessment in the training set and in all patients (p＜0.05). ConclusionsRadiomics analysis represents a promising tool in separating UM from other intraocular masses.

Pretreatment MR-based radiomics nomogram as potential imaging biomarker for individualized assessment of perineural invasion status in rectal cancer.

To investigate whether pretreatment magnetic resonance (MR)-based radiomics nomogram can individualize prediction of perineural invasion (PNI) status in rectal cancer (RC). A total of 122 RC patients with pathologically confirmed were classified as training cohort (n = 87) and test cohort (n = 35). 180 radiomics features were extracted from all lesions based on oblique axial T2WI TSE images. The dimensionality reduction and feature selection in training cohort were realized by the maximum relevance minimum redundancy (mRMR) algorithm and the least absolute shrinkage and selection operator (LASSO) regression model. A predictive model combining radiomics features and clinical risk factors (pathological N stage, pathological LVI status) was established by multivariate logistic regression analysis. The performance of the model was assessed based on its receiver operating characteristic (ROC) curve, nomogram, and calibration. The developed radiomics nomogram that integrated the radiomics signature and clinical risk factors could provide discrimination in the training and test cohorts. The accuracy and the area under the curve (AUC) for assessing PNI status were 0.82, 0.86, respectively, in the training cohort, while they were 0.71 and 0.85 in the test cohort. The goodness-of-fit of the nomogram was evaluated using the Hosmer-Lemeshow test (p = 0.52 in training cohort and p = 0.24 in test cohort). Decision curve analysis (DCA) showed that the radiomics nomogram was clinically useful. The developed radiomics nomogram might be helpful in the individualized assessment PNI status in patients with RC. This stratification of RC patients according to their PNI status may provide the basis for individualized adjuvant therapy, especially for stage II patients.

Utilization of radiomics to predict long-term outcome of magnetic resonance-guided focused ultrasound ablation therapy in adenomyosis.

To develop and evaluate a T2 MR-based radiomics prediction model incorporating radiomics features and clinical parameters to predict the response to magnetic resonance-guided focused ultrasound surgery (MRgFUS) in patients with adenomyosis. Sixty-nine patients (mean age, 38.6years; age range, 26-50years) with adenomyosis treated by MRgFUS were reviewed and allocated to training (n = 48) and testing cohorts (n = 21). One thousand one hundred eighteen radiomics features were extracted from T2-weighted imaging before MRgFUS. The radiomics features' dimension was reduced by Pearson correlation coefficient after normalization. Analysis of variance and logistical regression were used for feature selection by fivefold cross-validation in the training cohort, and the machine learning model was constructed for comparing the clinical model, radiomics model, and radiomics-clinical model which combined survived radiomics features and clinical parameters. The discrimination result of the model was obtained by bootstrap; receiver operating characteristic curve, area under the curve (AUC), and decision curve analyses were performed to illustrate the model performance in both the training and testing cohorts. Good response was achieved in 47 patients (68.1%) and failed in 22 patients (38.9%). The radiomics model comprised four selected features and demonstrated a degree of prediction capability of patients' poor response to MRgFUS treatment. The radiomics-clinical model showed good discrimination, with an AUC of 0.81 (95% confidence interval, 0.592-0.975) in the testing cohort. The decision curve analysis also showed favorable performance of the radiomics-clinical model. A prediction model composed of T2WI-based radiomics features and clinical parameters could be applied to guide the radiologist to evaluate MRgFUS for patients with adenomyosis who will achieve good response. • Magnetic resonance imaging-guided focused ultrasound surgery represents an alternative treatment for adenomyosis, but nearly one third of patients remain symptomatic 6months after MRgFUS. • Combining four radiomics features of T2-weighted MRI with eight clinical features further improves prediction of poor responders to MR-guided focused ultrasound treatment of uterine adenomyosis (AUC = 0.81 in the testing cohort). • The radiomics model based on T2-weighted imaging combined with clinical parameters can help predict which patients are likely to have a good response to MRgFUS for adenomyosis.

Standardization of brain MR images across machines and protocols: bridging the gap for MRI-based radiomics

Radiomics relies on the extraction of a wide variety of quantitative image-based features to provide decision support. Magnetic resonance imaging (MRI) contributes to the personalization of patient care but suffers from being highly dependent on acquisition and reconstruction parameters. Today, there are no guidelines regarding the optimal pre-processing of MR images in the context of radiomics, which is crucial for the generalization of published image-based signatures. This study aims to assess the impact of three different intensity normalization methods (Nyul, WhiteStripe, Z-Score) typically used in MRI together with two methods for intensity discretization (fixed bin size and fixed bin number). The impact of these methods was evaluated on first- and second-order radiomics features extracted from brain MRI, establishing a unified methodology for future radiomics studies. Two independent MRI datasets were used. The first one (DATASET1) included 20 institutional patients with WHO grade II and III gliomas who underwent post-contrast 3D axial T1-weighted (T1w-gd) and axial T2-weighted fluid attenuation inversion recovery (T2w-flair) sequences on two different MR devices (1.5 T and 3.0 T) with a 1-month delay. Jensen–Shannon divergence was used to compare pairs of intensity histograms before and after normalization. The stability of first-order and second-order features across the two acquisitions was analysed using the concordance correlation coefficient and the intra-class correlation coefficient. The second dataset (DATASET2) was extracted from the public TCIA database and included 108 patients with WHO grade II and III gliomas and 135 patients with WHO grade IV glioblastomas. The impact of normalization and discretization methods was evaluated based on a tumour grade classification task (balanced accuracy measurement) using five well-established machine learning algorithms. Intensity normalization highly improved the robustness of first-order features and the performances of subsequent classification models. For the T1w-gd sequence, the mean balanced accuracy for tumour grade classification was increased from 0.67 (95% CI 0.61–0.73) to 0.82 (95% CI 0.79–0.84, P = .006), 0.79 (95% CI 0.76–0.82, P = .021) and 0.82 (95% CI 0.80–0.85, P = .005), respectively, using the Nyul, WhiteStripe and Z-Score normalization methods compared to no normalization. The relative discretization makes unnecessary the use of intensity normalization for the second-order radiomics features. Even if the bin number for the discretization had a small impact on classification performances, a good compromise was obtained using the 32 bins considering both T1w-gd and T2w-flair sequences. No significant improvements in classification performances were observed using feature selection. A standardized pre-processing pipeline is proposed for the use of radiomics in MRI of brain tumours. For models based on first- and second-order features, we recommend normalizing images with the Z-Score method and adopting an absolute discretization approach. For second-order feature-based signatures, relative discretization can be used without prior normalization. In both cases, 32 bins for discretization are recommended. This study may pave the way for the multicentric development and validation of MR-based radiomics biomarkers.

MRI-Based Radiomics Signature: A Potential Biomarker for Identifying Glypican 3-Positive Hepatocellular Carcinoma.

Glypican 3 (GPC3) expression has proved to be a critical risk factor related to prognosis in hepatocellular carcinoma (HCC) patients. To investigate the performance of MRI-based radiomics signature in identifying GPC3-positive HCC. Retrospective. An initial cohort of 293 patients with pathologically confirmed HCC was involved in this study, and patients were randomly divided into training (195) and validation (98) cohorts. Contrast-enhanced T1 -weight MRI was performed with a 1.5T scanner. A total of 853 radiomic features were extracted from the volume imaging. Univariate analysis and Fisher scoring were utilized for feature reduction. Subsequently, forward stepwise feature selection and radiomics signature building were performed based on a support vector machine (SVM). Incorporating independent risk factors, a combined nomogram was developed by multivariable logistic regression modeling. The predictive performance of the nomogram was calculated using the area under the receive operating characteristic curve (AUC). Decision curve analysis (DCA) was applied to estimate the clinical usefulness. The radiomics signature consisting of 10 selected features achieved good prediction efficacy (training cohort: AUC = 0.879, validation cohort: AUC = 0.871). Additionally, the combined nomogram integrating independent clinical risk factor α-fetoprotein (AFP) and radiomics signature showed improved calibration and prominent predictive performance with AUCs of 0.926 and 0.914 in the training and validation cohorts, respectively. The proposed MR-based radiomics signature is strongly related to GPC3-positive. The combined nomogram incorporating AFP and radiomics signature may provide an effective tool for noninvasive and individualized prediction of GPC3-positive in patients with HCC. J. MAGN. RESON. IMAGING 2020;52:1679-1687.

Development and external validation of an MRI-based radiomics nomogram for pretreatment prediction for early relapse in osteosarcoma: A retrospective multicenter study

PurposeTo develop and externally validate an MR-based radiomics nomogram from retrospective multicenter datasets for pretreatment prediction of early relapse (≤ 1 year) in osteosarcoma after surgical resection. MethodsThis multicenter study retrospectively enrolled 93 patients (training cohort: 62 patients from four hospitals; validation cohort: 31 patients from two hospitals) with clinicopathologically confirmed osteosarcoma who received neoadjuvant chemotherapy and surgical resection at six hospitals between January 2009 and October 2017. Radiomics features were extracted from contrast-enhanced fat-suppressed T1-weighted (CE FS T1-w) images. Least absolute shrinkage and selection operator (LASSO) regression was applied for feature selection and radiomics signature construction. The radiomics nomogram that incorporated the radiomics signature and subjective MRI-assessed candidate predictors was developed to predict early relapse with a multivariate logistic regression model in the training cohort and validated in the external validation cohort. The performance of the nomogram was assessed by its discrimination, calibration, and clinical usefulness. ResultsThe radiomics signature comprised six selected features and achieved favorable prediction efficacy. The radiomics nomogram incorporating the radiomics signature and subjective MRI-assessed candidate predictors (joint invasion and perivascular involvement) from the multicenter datasets achieved better discrimination in the training cohort (C-index：0.907, 95 % CI: 0.838−0.977) and external validation cohort (C-index: 0.811, 95 % CI: 0.653−0.970), and good calibration. Decision curve analysis suggested that the combined nomogram was clinically useful. ConclusionThe proposed MRI-based radiomics nomogram could provide a non-invasive tool to predict early relapse of osteosarcoma, which has the potential to improve personalized pretreatment management of osteosarcoma.

IDH1 mutation prediction using MR-based radiomics in glioblastoma: comparison between manual and fully automated deep learning-based approach of tumor segmentation

PurposeThis study aimed to determine whether MR-based radiomics of glioblastoma can predict the isocitrate dehydrogenase-1 (IDH1) mutation status and compare predictive performances between manual and fully automatic deep-learning segmentations. MethodForty-five glioblastoma patients with pretreatment T2-weighted MRIs were retrospectively evaluated. Manual segmentations of glioblastoma and peri-tumoral edema were trained via a deep neural network (V-Net). An independent external cohort of 137 glioblastoma patients from the Cancer Imaging Archive was also included (test set 1, n = 46; test set 2, n = 91). Test set 1—without known IDH1 status—was used to calculate dice similarity coefficients (DSC) between the two segmentation methods (manual & V-Net). From test set 2, all-relevant radiomic features were selected via a random forest-based wrapper algorithm for IDH1 prediction. Receiver operating characteristics (ROC) curves with areas under the curve (AUC) were plotted as performance metrics for both methods. ResultsAmong 136 patients (45 and 91 patients from our institution and test set 2, respectively), 17 patients (11.2 %) had IDH1 mutations. The mean DSC of test set 1 was 0.78 ± 0.14 (range, 0.34−0.94). A subset of 9 all-relevant features (8.4 %, 9/107) was selected. V-Net segmentation of the test set 2 yielded similar performance in predicting IDH1 mutation as compared to manual segmentation (V-Net AUC = 0.86 vs. manual AUC = 0.90). The optimal cut-point threshold of AUC yielded 86.8 % accuracy for manual segmentation and 75.8 % for V-Net segmentation. ConclusionsV-Net showed robust segmentation capability of glioblastoma on T2-weighted MRI. All-relevant radiomics features from both segmentation methods yielded a similar performance in IDH1 prediction.

The Usefulness of Pretreatment MR-Based Radiomics on Early Response of Neoadjuvant Chemotherapy in Patients With Locally Advanced Nasopharyngeal Carcinoma.

The aim of this study was to explore the predictive role of pretreatment MRI-based radiomics on early response of neoadjuvant chemotherapy (NAC) in locoregionally advanced nasopharyngeal carcinoma (NPC) patients. Between January 2016 and December 2016, a total of 108 newly diagnosed NPC patients who were hospitalized in the Cancer Hospital of the University of Chinese Academy of Sciences were reviewed. All patients had complete data of enhanced MR of nasopharynx before treatment, and then received two to three cycles of TP-based NAC. After 2 cycles of NAC, enhanced MR of nasopharynx was conducted again. Compared with the enhanced MR images before treatment, the response after NAC was evaluated. According to the evaluation criteria of RECIST1.1, 108 cases were divided into two groups: 52 cases for the NAC-sensitive group and 56 cases for the NAC-resistance group. ITK-SNAP software was used to manually sketch and segment the region of interest (ROI) of nasopharyngeal tumor on the MR enhanced T1WI sequence image. The parameters were analyzed and extracted by using AI Kit software. ANOVA/MW test, correlation analysis, and LASSO were used to select texture features. We used multivariate logistic regressions to select texture features and establish a predictive model. The ROC curve was used to evaluate the efficiency of the predictive model. A total of 396 texture features were obtained by using feature calculation. After all features were screened, we selected two features including ClusterShade_angle135_offset4 and Correlation_AllDirection_offshe1_SD. Based on these two features, we established a predictive model by using multivariate logistic regression. The AUC of the two features used alone (0.804, 95% CI = 0.602–0.932; 0.762, 95% CI = 0.556–0.905) was smaller than the combination of these two features (0.905, 95% CI = 0.724–0.984, p = 0.0005). Moreover, the sensitivity values of the two features used alone and the combined use were 92.9%, 51.7%, and 85.7%, respectively, while the specificity values were 66.7%, 91.7%, and 83.3%, respectively, in the early response of NAC for NPC. The predictive model based on MRI-enhanced sequence imaging could distinguish the sensitivity and resistance to NAC and provide new biomarkers for the early prediction of the curative effect in NPC patients.

Quantitative MRI Biomarkers of Stereotactic Radiotherapy Outcome in Brain Metastasis

About 20–40% of cancer patients develop brain metastases, causing significant morbidity and mortality. Stereotactic radiation treatment is an established option that delivers high dose radiation to the target while sparing the surrounding normal tissue. However, up to 20% of metastatic brain tumours progress despite stereotactic treatment, and it can take months before it is evident on follow-up imaging. An early predictor of radiation therapy outcome in terms of tumour local failure (LF) is crucial, and can facilitate treatment adjustments or allow for early salvage treatment. In this study, an MR-based radiomics framework was proposed to derive and investigate quantitative MRI (qMRI) biomarkers for the outcome of LF in brain metastasis patients treated with hypo-fractionated stereotactic radiation therapy (SRT). The qMRI biomarkers were constructed through a multi-step feature extraction/reduction/selection framework using the conventional MR imaging data acquired from 100 patients (133 lesions), and were applied in conjunction with machine learning techniques for outcome prediction and risk assessment. The results indicated that the majority of the features in the optimal qMRI biomarkers characterize the heterogeneity in the surrounding regions of tumour including edema and tumour/lesion margins. The optimal qMRI biomarker consisted of five features that predict the outcome of LF with an area under the curve (AUC) of 0.79, and a cross-validated sensitivity and specificity of 81% and 79%, respectively. The Kaplan-Meier analyses showed a statistically significant difference in local control (p-value < 0.0001) and overall survival (p = 0.01). Findings from this study are a step towards using qMRI for early prediction of local failure in brain metastasis patients treated with SRT. This may facilitate early adjustments in treatment, such as surgical resection or salvage radiation, that can potentially improve treatment outcomes. Investigations on larger cohorts of patients are, however, required for further validation of the technique.

PO-162: Predicting response to neoadjuvant chemoradiotherapy in cervical cancer: a MR-based radiomics model

PO-162: Predicting response to neoadjuvant chemoradiotherapy in cervical cancer: a MR-based radiomics model