Multi-sequence Magnetic Resonance Imaging Research Articles

Development and Validation of a Deep Learning Predictive Model Combining Clinical and Radiomic Features for Short-Term Postoperative Facial Nerve Function in Acoustic Neuroma Patients.

This study aims to construct and validate a predictable deep learning model associated with clinical data and multi-sequence magnetic resonance imaging (MRI) for short-term postoperative facial nerve function in patients with acoustic neuroma. A total of 110 patients with acoustic neuroma who underwent surgery through the retrosigmoid sinus approach were included. Clinical data and raw features from four MRI sequences (T1-weighted, T2-weighted, T1-weighted contrast enhancement, and T2-weighted-Flair images) were analyzed. Spearman correlation analysis along with least absolute shrinkage and selection operator regression were used to screen combined clinical and radiomic features. Nomogram, machine learning, and convolutional neural network (CNN) models were constructed to predict the prognosis of facial nerve function on the seventh day after surgery. Receiver operating characteristic (ROC) curve and decision curve analysis (DCA) were used to evaluate model performance. A total of 1050 radiomic parameters were extracted, from which 13 radiomic and 3 clinical features were selected. The CNN model performed best among all prediction models in the test set with an area under the curve (AUC) of 0.89 (95% CI, 0.84-0.91). CNN modeling that combines clinical and multi-sequence MRI radiomic features provides excellent performance for predicting short-term facial nerve function after surgery in patients with acoustic neuroma. As such, CNN modeling may serve as a potential decision-making tool for neurosurgery.

Applying multisequence MRI radiomics of the primary tumor and lymph node to predict HPV-related p16 status in patients with oropharyngeal squamous cell carcinoma.

Human papillomavirus (HPV)-related oropharyngeal squamous cell carcinoma (OPSCC) is currently rising in incidence; however, a noninvasive approach to predicting HPV status that strongly correlates with p16 expression is lacking. This study aimed to develop a radiomics model based on multisequence magnetic resonance imaging (MRI) of primary tumor (PT) and lymph node (LN)-fused imaging features for the prediction of p16 status in OPSCC. In this retrospective study, 141 patients (comprising 116 patients in the training cohort and 25 patients in the testing cohort) with histopathologically confirmed (OPSCC) were enrolled consecutively from Fudan University Shanghai Cancer Center and Shanghai Ninth People's Hospital between January 2011 and December 2020. HPV status was determined by p16 immunohistochemistry analysis. A total of 2092 radiomics features were initially computed and extracted from the 3D-segmented PT and largest LN based on contrast-enhanced T1-weighted imaging (CE-T1WI) and T2-weighted imaging (T2WI). A support vector machine classifier was employed to build the machine learning-based classification models dependent on p16 status. The models were validated in the testing cohort. The area under the receiver operating characteristic curve (AUC) was computed to assess the performance of each model. This diagnostic study was not registered on the clinical trial platform. In the testing cohort, fusion models yielded better performance (AUC) compared with models based on a sole PT/LN [CE-T1WI: 0.80 (95% CI: 0.55-0.94) vs. 0.71 (95% CI: 0.47-0.88)/0.73 (95% CI: 0.48-0.90); T2WI: 0.74 (95% CI: 0.51-0.95) vs. 0.64 (95% CI: 0.38-0.85)/0.71 (95% CI: 0.48-0.88)]. Models based on multisequence imaging outperformed single CE-T1WI/T2WI models [PT: 0.74 (95% CI: 0.46-0.91) vs. 0.71 (95% CI: 0.47-0.88)/0.64 (95% CI: 0.38-0.85); LN: 0.78 (95% CI: 0.55-0.75) vs. 0.73 (95% CI: 0.48-0.90)/0.71 (95% CI: 0.48-0.88)]. Finally, the PT-LN fusion model based on multisequencing yielded the best classification performance with the highest AUC value of 0.91 (95% CI: 0.72-0.98) for the prediction of p16 expression. The differences between the performance of the final model and the other 8 models were significant (all P values <0.05). The results demonstrated that (I) the PT-LN fusion radiomics models improved the classification performance of the sole use of PT or LN for the prediction of p16 status, (II) the radiomics models based on multisequences outperformed the single-sequence models in the prediction of p16 status, and (III) the PT-LN fusion model based on multisequence MRI radiomics features could serve as a noninvasive method for acquiring the molecular information of patients with OPSCC, potentially assisting oncologists with their clinical decision-making.

Automatic tumor segmentation and metachronous single-organ metastasis prediction of nasopharyngeal carcinoma patients based on multi-sequence magnetic resonance imaging.

Distant metastases is the main failure mode of nasopharyngeal carcinoma. However, early prediction of distant metastases in NPC is extremely challenging. Deep learning has made great progress in recent years. Relying on the rich data features of radiomics and the advantages of deep learning in image representation and intelligent learning, this study intends to explore and construct the metachronous single-organ metastases (MSOM) based on multimodal magnetic resonance imaging. The magnetic resonance imaging data of 186 patients with nasopharyngeal carcinoma before treatment were collected, and the gross tumor volume (GTV) and metastatic lymph nodes (GTVln) prior to treatment were defined on T1WI, T2WI, and CE-T1WI. After image normalization, the deep learning platform Python (version 3.9.12) was used in Ubuntu 20.04.1 LTS to construct automatic tumor detection and the MSOM prediction model. There were 85 of 186 patients who had MSOM (including 32 liver metastases, 25 lung metastases, and 28 bone metastases). The median time to MSOM was 13 months after treatment (7-36 months). The patients were randomly assigned to the training set (N = 140) and validation set (N = 46). By comparison, we found that the overall performance of the automatic tumor detection model based on CE-T1WI was the best (6). The performance of automatic detection for primary tumor (GTV) and lymph node gross tumor volume (GTVln) based on the CE-T1WI model was better than that of models based on T1WI and T2WI (AP@0.5 is 59.6 and 55.6). The prediction model based on CE-T1WI for MSOM prediction achieved the best overall performance, and it obtained the largest AUC value (AUC = 0.733) in the validation set. The precision, recall, precision, and AUC of the prediction model based on CE-T1WI are 0.727, 0.533, 0.730, and 0.733 (95% CI 0.557-0.909), respectively. When clinical data were added to the deep learning prediction model, a better performance of the model could be obtained; the AUC of the integrated model based on T2WI, T1WI, and CE-T1WI were 0.719, 0.738, and 0.775, respectively. By comparing the 3-year survival of high-risk and low-risk patients based on the fusion model, we found that the 3-year DMFS of low and high MSOM risk patients were 95% and 11.4%, respectively (p < 0.001). The intelligent prediction model based on magnetic resonance imaging alone or combined with clinical data achieves excellent performance in automatic tumor detection and MSOM prediction for NPC patients and is worthy of clinical application.

CT synthesis from multi-sequence MRI using adaptive fusion network

ObjectiveTo investigate a method using multi-sequence magnetic resonance imaging (MRI) to synthesize computed tomography (CT) for MRI-only radiation therapy. ApproachWe proposed an adaptive multi-sequence fusion network (AMSF-Net) to exploit both voxel- and context-wise cross-sequence correlations from multiple MRI sequences to synthesize CT using element- and patch-wise fusions, respectively. The element- and patch-wise fusion feature spaces were combined, and the most representative features were selected for modeling. Finally, a densely connected convolutional decoder was applied to utilize the selected features to produce synthetic CT images. Main resultsThis study includes a total number of 90 patients’ T1-weighted MRI, T2-weighted MRI and CT data. The AMSF-Net reduced the average mean absolute error (MAE) from 52.88-57.23 to 49.15 HU, increased the peak signal-to-noise ratio (PSNR) from 24.82-25.32 to 25.63 dB, increased the structural similarity index measure (SSIM) from 0.857-0.869 to 0.878, and increased the dice coefficient of bone from 0.886-0.896 to 0.903 compared to the other three existing multi-sequence learning models. The improvements were statistically significant according to two-tailed paired t-test. In addition, AMSF-Net reduced the intensity difference with real CT in five organs at risk, four types of normal tissue and tumor compared with the baseline models. The MAE decreases in parotid and spinal cord were over 8% and 16% with reference to the mean intensity value of the corresponding organ, respectively. Further, the qualitative evaluations confirmed that AMSF-Net exhibited superior structural image quality of synthesized bone and small organs such as the eye lens. SignificanceThe proposed method can improve the intensity and structural image quality of synthetic CT and has potential for use in clinical applications.

Deep Learning Provides a New Magnetic Resonance Imaging-Based Prognostic Biomarker for Recurrence Prediction in High-Grade Serous Ovarian Cancer

This study aims to use a deep learning method to develop a signature extract from preoperative magnetic resonance imaging (MRI) and to evaluate its ability as a non-invasive recurrence risk prognostic marker in patients with advanced high-grade serous ovarian cancer (HGSOC). Our study comprises a total of 185 patients with pathologically confirmed HGSOC. A total of 185 patients were randomly assigned in a 5:3:2 ratio to a training cohort (n = 92), validation cohort 1 (n = 56), and validation cohort 2 (n = 37). We built a new deep learning network from 3839 preoperative MRI images (T2-weighted images and diffusion-weighted images) to extract HGSOC prognostic indicators. Following that, a fusion model including clinical and deep learning features is developed to predict patients' individual recurrence risk and 3-year recurrence likelihood. In the two validation cohorts, the consistency index of the fusion model was higher than both the deep learning model and the clinical feature model (0.752, 0.813 vs. 0.625, 0.600 vs. 0.505, 0.501). Among the three models, the fusion model had a higher AUC than either the deep learning model or the clinical model in validation cohorts 1 or 2 (AUC = was 0.986, 0.961 vs. 0.706, 0.676/0.506, 0.506). Using the DeLong method, the difference between them was statistically significant (p < 0.05). The Kaplan-Meier analysis distinguished two patient groups with high and low recurrence risk (p = 0.0008 and 0.0035, respectively). Deep learning may be a low-cost, non-invasive method for predicting risk for advanced HGSOC recurrence. Deep learning based on multi-sequence MRI serves as a prognostic biomarker for advanced HGSOC, which provides a preoperative model for predicting recurrence in HGSOC. Additionally, using the fusion model as a new prognostic analysis means that can use MRI data can be used without the need to follow-up the prognostic biomarker.

Study of high-altitude cerebral edema using multimodal imaging.

To analyze the brain imaging features of high-altitude cerebral edema (HACE) using computed tomography (CT) and multi-sequence magnetic resonance imaging (MRI) and to explore its injury characteristics. We selected 30 patients with HACE diagnosed between January 2012 to August 2022 as the experimental group and 60 patients with dizziness on traveling from the plain to the plateau or from lower altitude to higher altitude in a short period of time as the control group. We collected general clinical data from the experimental group and classified it according to clinical symptoms. In both groups, we then performed a head CT and multi-sequence MRI (T1WI, T2WI, FLAIR, and DWI). Among them, nine patients with HACE were also scanned using susceptibility-weighted imaging (SWI). Finally, we analyzed the images. According to clinical symptoms, we divided the 30 cases of HACE into 12 mild cases and 18 severe cases. There was no significant difference in sex, age, leukocyte, neutrophil, or glucose content between mild and severe HACE. The sensitivity and specificity of the MRI diagnosis were 100 and 100%, respectively, while the sensitivity and specificity of the CT diagnosis were 23.3 and 100%, respectively. The distribution range of deep and juxtacortical white matter edema was significantly larger in severe HACE than in mild HACE (p < 0.001). The corpus callosum edema distribution range in severe HACE was significantly larger than that in mild HACE (p = 0.001). The ADC value of the splenium of the corpus callosum was significantly lower in severe HACE than in mild HACE (p = 0.049). In mild and severe HACE, the signal intensity of the DWI sequence was significantly higher than that of conventional MRI sequences (T1WI, T2WI, FLAIR) (p = 0.008, p = 0.025, respectively). In severe HACE, seven cases showed bilateral corticospinal tract edema at the thalamic level, and SWI showed cerebral microbleeds (CMBs) in five cases, especially in the corpus callosum. MRI has more advantages than CT in the evaluation of HACE, especially in the DWI sequence. The white matter injury of severe HACE is more severe and extensive, especially in the corpus callosum, and some CMBs and corticospinal tract edema may also appear.

Multisequence magnetic resonance imaging-based radiomics models for the prediction of microsatellite instability in endometrial cancer.

To evaluate the performance of multisequence magnetic resonance imaging (MRI)-based radiomics models in the assessment of microsatellite instability (MSI) statusin endometrial cancer (EC). This retrospective multicentre study included 338 EC patients with available MSI status and preoperative MRI scans, divided into training (37 MSI, 123 microsatellite stability [MSS]), internalvalidation (15 MSI, 52 MSS), and external validation cohorts (30 MSI, 81 MSS). Radiomics features were extracted from T2-weighted images, diffusion-weighted images, and contrast-enhanced T1-weighted images. The ComBat harmonisation method was applied to remove intrascanner variability. The Boruta wrapper algorithm was used for key feature selection. Three classification algorithms, logistic regression (LR), random forest (RF), and support vector machine (SVM), wereapplied to build the radiomics models. The area under the receiver operating characteristic curve (AUC) was calculated to compare the diagnostic performance of the models. Decision curve analysis (DCA) was conducted to determine the clinical usefulness of the models. Among the 1980 features, Boruta finally selected nine radiomics features. A higher MSI prediction performance was achieved after running the ComBat harmonisation method. The SVM algorithm had the best performance, with AUCs of 0.921, 0.903, and 0.937 in the training, internalvalidation, and external validation cohorts, respectively. The DCA results showed that the SVM algorithm achieved higher net benefits than the other classifiers over a threshold range of 0.581-0.783. The multisequence MRI-based radiomics models showed promise in preoperatively predicting the MSI status in EC in this multicentre setting.

Preoperative prediction of miliary changes in the small bowel mesentery in advanced high-grade serous ovarian cancer using MRI radiomics nomogram.

To develop and validate an MRI-based radiomics nomogram for the preoperative prediction of miliary changes in the small bowel mesentery (MCSBM) in advanced high-grade serous ovarian cancer (HGSOC). One hundred and twenty-eight patients with pathologically proved advanced HGSOC (training cohort: n = 91; validation cohort: n = 37) were retrospectively included. All patients were initially evaluated as MCSBM-negative by preoperative imaging modalities but were finally confirmed by surgery and histopathology (MCSBM-positive: n = 53; MCSBM-negative: n = 75). Five radiomics signatures were built based on the features from multisequence magnetic resonance images. Independent clinicoradiological factors and radiomics-fusion signature were further integrated to construct a radiomics nomogram. The performance of the nomogram was assessed using receiver operating characteristic (ROC) curves, calibration curves and clinical utility. Radiomics signatures, ascites, and tumor size were independent predictors of MCSBM. A nomogram integrating radiomics features and clinicoradiological factors demonstrated satisfactory predictive performance with areas under the curves (AUCs) of 0.871 (95% CI 0.801-0.941) and 0.858 (95% CI 0.739-0.976) in the training and validation cohorts, respectively. The net reclassification index (NRI) and integrated discrimination improvement (IDI) revealed that the nomogram had a significantly improved ability compared with the clinical model in the training cohort (NRI = 0.343, p = 0.002; IDI = 0.299, p < 0.001) and validation cohort (NRI = 0.409, p = 0.015; IDI = 0.283, p = 0.001). Our proposed nomogram has the potential to serve as a noninvasive tool for the prediction of MCSBM, which is helpful for the individualized assessment of advanced HGSOC patients.

Predicting methylation class from diffusely infiltrating adult gliomas using multimodality MRI data.

Radiogenomic studies of adult-type diffuse gliomas have used magnetic resonance imaging (MRI) data to infer tumor attributes, including abnormalities such as IDH-mutation status and 1p19q deletion. This approach is effective but does not generalize to tumor types that lack highly recurrent alterations. Tumors have intrinsic DNA methylation patterns and can be grouped into stable methylation classes even when lacking recurrent mutations or copy number changes. The purpose of this study was to prove the principle that a tumor's DNA-methylation class could be used as a predictive feature for radiogenomic modeling. Using a custom DNA methylation-based classification model, molecular classes were assigned to diffuse gliomas in The Cancer Genome Atlas (TCGA) dataset. We then constructed and validated machine learning models to predict a tumor's methylation family or subclass from matched multisequence MRI data using either extracted radiomic features or directly from MRI images. For models using extracted radiomic features, we demonstrated top accuracies above 90% for predicting IDH-glioma and GBM-IDHwt methylation families, IDH-mutant tumor methylation subclasses, or GBM-IDHwt molecular subclasses. Classification models utilizing MRI images directly demonstrated average accuracies of 80.6% for predicting methylation families, compared to 87.2% and 89.0% for differentiating IDH-mutated astrocytomas from oligodendrogliomas and glioblastoma molecular subclasses, respectively. These findings demonstrate that MRI-based machine learning models can effectively predict the methylation class of brain tumors. Given appropriate datasets, this approach could generalize to most brain tumor types, expanding the number and types of tumors that could be used to develop radiomic or radiogenomic models.

Predicting IDH Mutation Status in Low-Grade Gliomas Based on Optimal Radiomic Features Combined with Multi-Sequence Magnetic Resonance Imaging.

The IDH somatic mutation status is an important basis for the diagnosis and classification of gliomas. We proposed a "6-Step" general radiomics model to noninvasively predict the IDH mutation status by simultaneously tuning combined multi-sequence MRI and optimizing the full radiomics processing pipeline. Radiomic features (n = 3776) were extracted from multi-sequence MRI (T1, T2, FLAIR, and T1Gd) in low-grade gliomas (LGGs), and a total of 45,360 radiomics pipeline were investigated according to different settings. The predictive ability of the general radiomics model was evaluated with regards to accuracy, stability, and efficiency. Based on numerous experiments, we finally reached an optimal pipeline for classifying IDH mutation status, namely the T2+FLAIR combined multi-sequence with the wavelet image filter, mean data normalization, PCC dimension reduction, RFE feature selection, and SVM classifier. The mean and standard deviation of AUC, accuracy, sensitivity, and specificity were 0.873 ± 0.05, 0.876 ± 0.09, 0.875 ± 0.11, and 0.877 ± 0.15, respectively. Furthermore, 14 radiomic features that best distinguished the IDH mutation status of the T2+FLAIR multi-sequence were analyzed, and the gray level co-occurrence matrix (GLCM) features were shown to be of high importance. Apart from the promising prediction of the molecular subtypes, this study also provided a general tool for radiomics investigation.

Radiomics models based on multi-sequence MRI for preoperative evaluation of MUC4 status in pancreatic ductal adenocarcinoma: a preliminary study.

Mucin 4 (MUC4) overexpression promotes tumorigenesis and increases the aggressiveness of pancreatic ductal adenocarcinoma (PDAC). To date, no study has reported the association between radiomics and MUC4 expression in PDAC. Thus, we aimed to explore the utility of radiomics based on multi-sequence magnetic resonance imaging (MRI) to predict the status of MUC4 expression in PDAC preoperatively. This retrospective study included 52 patients with PDAC who underwent MRI. The patients were divided into two groups based on MUC4 expression status. Two feature sets were extracted from the arterial and portal phases (PPs) of dynamic contrast-enhanced MRI (DCE-MRI). Univariate analysis, minimum redundancy maximum relevance (MRMR), and principal component analysis (PCA) were performed for the feature selection of each dataset, and features with a cumulative variance of 90% were selected to develop radiomics models. Clinical characteristics were gathered to develop a clinical model. The selected radiomics features and clinical characteristics were modeled by multivariable logistic regression. The combined model integrated radiomics features from different selected data sets and clinical characteristics. The classification metrics were applied to assess the discriminatory power of the models. There were 22 PDACs with a high expression of MUC4 and 30 PDACs with a low expression of MUC4. The area under the receiver operating characteristic (ROC) curve (AUC) values of the arterial phase (AP) model, the PP model, and the combined model were 0.732 (0.591-0.872), 0.709 (0.569-0.849), and 0.861 (0.760-0.961), respectively. The AUC of the clinical model was 0.666 (0.600-0.682). The combined model that was constructed outperformed the AP, the PP, and the clinical models (P<0.05, although no statistical significance was observed in the combined model vs. AP model). Radiomics models based on multi-sequence MRI have the potential to predict MUC4 expression levels in PDAC.

Self-Supervised Multi-Modal Hybrid Fusion Network for Brain Tumor Segmentation

Accurate medical image segmentation of brain tumors is necessary for the diagnosing, monitoring, and treating disease. In recent years, with the gradual emergence of multi-sequence magnetic resonance imaging (MRI), multi-modal MRI diagnosis has played an increasingly important role in the early diagnosis of brain tumors by providing complementary information for a given lesion. Different MRI modalities vary significantly in context, as well as in coarse and fine information. As the manual identification of brain tumors is very complicated, it usually requires the lengthy consultation of multiple experts. The automatic segmentation of brain tumors from MRI images can thus greatly reduce the workload of doctors and buy more time for treating patients. In this paper, we propose a multi-modal brain tumor segmentation framework that adopts the hybrid fusion of modality-specific features using a self-supervised learning strategy. The algorithm is based on a fully convolutional neural network. Firstly, we propose a multi-input architecture that learns independent features from multi-modal data, and can be adapted to different numbers of multi-modal inputs. Compared with single-modal multi-channel networks, our model provides a better feature extractor for segmentation tasks, which learns cross-modal information from multi-modal data. Secondly, we propose a new feature fusion scheme, named hybrid attentional fusion. This scheme enables the network to learn the hybrid representation of multiple features and capture the correlation information between them through an attention mechanism. Unlike popular methods, such as feature map concatenation, this scheme focuses on the complementarity between multi-modal data, which can significantly improve the segmentation results of specific regions. Thirdly, we propose a self-supervised learning strategy for brain tumor segmentation tasks. Our experimental results demonstrate the effectiveness of the proposed model against other state-of-the-art multi-modal medical segmentation methods.

Computer-aided hepatocellular carcinoma detection on the hepatobiliary phase of gadoxetic acid-enhanced magnetic resonance imaging using a convolutional neural network: Feasibility evaluation with multi-sequence data

Diagnosis of hepatocellular carcinoma (HCC) on liver MRI needs analysis of multi-sequence images. However, developing computer-aided detection (CAD) for every single sequence requires considerable time and labor for image segmentation. Therefore, we developed CAD for HCC on the hepatobiliary phase (HBP) of gadoxetic acid-enhanced magnetic resonance imaging (MRI) using a convolutional neural network (CNN) and evaluated its feasibility on multi-sequence, multi-unit, and multi-center data. Patients who underwent gadoxetic acid-enhanced MRI and surgery for HCC in Korea University Anam Hospital (KUAH) and Korea University Guro Hospital (KUGH) were reviewed. Finally, 170 nodules from 155 consecutive patients from KUAH and 28 nodules from 28 patients randomly selected from KUGH were included. Regions of interests were drawn on the whole HCC volume on HBP, T1-weighted (T1WI), T2-weighted (T2WI), and portal venous phase (PVP) images. The CAD was developed from the HBP images of KUAH using customized-nnUNet and post-processed for false-positive reduction. Internal and external validation of the CAD was performed with HBP, T1WI, T2WI, and PVP of KUAH and KUGH. The figure of merit and recall of the jackknife alternative free-response receiver operating characteristic of the CAD for HBP, T1WI, T2WI, and PVP at false-positive rate 0.5 were (0.87 and 87.0), (0.73 and 73.3), (0.13 and 13.3), and (0.67 and 66.7) in KUAH and (0.86 and 86.0), (0.61 and 53.6), (0.07 and 0.07), and (0.57 and 53.6) in KUGH, respectively. The CAD for HCC on gadoxetic acid-enhanced MRI developed by CNN from HBP detected HCCs feasibly on HBP, T1WI, and PVP of gadoxetic acid-enhanced MRI obtained from multiple units and centers. This result imply that the CAD developed using single MRI sequence may be applied to other similar sequences and this will reduce labor and time for CAD development in multi-sequence MRI.

Radiomic Signatures for Predicting EGFR Mutation Status in Lung Cancer Brain Metastases.

BackgroundLung cancer is the most common primary tumor metastasizing to the brain. A significant proportion of lung cancer patients show epidermal growth factor receptor (EGFR) mutation status discordance between the primary cancer and the corresponding brain metastases, which can affect prognosis and therapeutic decision-making. However, it is not always feasible to obtain brain metastases samples. The aim of this study was to establish a radiomic model to predict the EGFR mutation status of lung cancer brain metastases.MethodsData from 162 patients with resected brain metastases originating from lung cancer (70 with mutant EGFR, 92 with wild-type EGFR) were retrospectively analyzed. Radiomic features were extracted using preoperative brain magnetic resonance (MR) images (contrast-enhanced T1-weighted imaging, T1CE; T2-weighted imaging, T2WI; T2 fluid-attenuated inversion recovery, T2 FLAIR; and combinations of these sequences), to establish machine learning-based models for predicting the EGFR status of excised brain metastases (108 metastases for training and 54 metastases for testing). The least absolute shrinkage selection operator was used to select informative features; radiomics models were built with logistic regression of the training cohort, and model performance was evaluated using an independent test set.ResultsThe best-performing model was a combination of 10 features selected from multiple sequences (two from T1CE, five from T2WI, and three from T2 FLAIR) in both the training and test sets, resulting in classification area under the curve, accuracy, sensitivity, and specificity values of 0.85 and 0.81, 77.8% and 75.9%, 83.7% and 73.1%, and 73.8% and 78.6%, respectively.ConclusionsRadiomic signatures integrating multi-sequence MR images have the potential to noninvasively predict the EGFR mutation status of lung cancer brain metastases.

Three-Dimensional Arterial Spin Labeling-Guided Sub-Volume Segmentation of Radiotherapy in Adult Non-Enhancing Low-Grade Gliomas.

ObjectiveThe present study aimed to evaluate the feasibility of sub-volume segmentation for radiotherapy planning of adult non-enhancing low-grade gliomas (NE-LGGs) guided by three-dimensional arterial spin labeling (3D-ASL). The differences in high- and low-perfusion areas of NE-LGGs were analyzed using multi-sequence magnetic resonance imaging (MRI) radiomics.MethodsFifteen adult patients with NE-LGGs were included in the study. MR images, including T1-weighted imaging (T1WI), T2 Propeller, T2 fluid-attenuated inversion recovery (T2 Flair), 3D-ASL, and contrast-enhanced T1WI (CE-T1WI), were obtained. The gross tumor volume (GTV) was delineated according to the hyperintensity on T2 Flair. The GTV was divided into high- and low-perfusion areas, namely GTV-ASL and GTV-SUB, respectively, based on the differences in cerebral blood flow (CBF) value. The volumes and CBF values of high- and low-perfusion areas were measured and compared. The least absolute shrinkage and selection operator (LASSO) regression was used to select the optimal features of all MR maps. Receiver operating characteristic (ROC) curve analysis was used to evaluate the diagnostic accuracy of the absolute CBFmean (aCBFmean), relative CBFmean (rCBFmean, normalized by the CBF value of the normal gray matter), and screened features in differentiating high- and low-perfusion areas.ResultsAmong the enrolled patients, three (20%) patients with NE-LGGs showed focal intra- and post-radiotherapy contrast enhancement within a prior high-perfusion area of 3D-ASL. The volume ratio of the GTV-ASL to the GTV was (37.08% ± 17.88)% (46.26 ± 44.51 vs. 167.46 ± 209.64 cm3, P = 0.000). The CBFmean in the high-perfusion area was approximately two times of that in the edema area or normal gray matter (66.98 ± 18.03 vs. 35.19 ± 7.75 or 33.92 ± 8.48 ml/100g/min, P = 0.000). Thirteen features were screened, seven of which were extracted from 3D-ASL. The area undercurve (AUC) values of aCBFmean, rCBFmean, and firstorder_10Percentile from 3D-ASL were more than 0.9, of which firstorder_10Percentile was the highest. Their cut-off values were 44.16 ml/100 g/min, 1.49 and 31, respectively.ConclusionThe difference in blood perfusion in the GTV can be quantified and analyzed based on 3D-ASL images for NE-LGGs, which could guide the sub-volume segmentation of the GTV. 3D-ASL should become a routine method for NE-LGGs during simulation and radiotherapy.

Radiomics model based on multi-sequence MR images for predicting preoperative immunoscore in rectal cancer.

To establish and validate a radiomics model based on multi-sequence magnetic resonance (MR) images for preoperative prediction of immunoscore in rectal cancer. This retrospective study included 133 patients with pathologically confirmed rectal cancer after surgical resection who underwent MR examination before treatment within two weeks. All patients were randomly divided into training cohort (n = 92) and validation (n = 41) cohort according to a ratio of 7:3. The volumes of interest were manually delineated in the T2-weighted images (T2WI) and apparent diffusion coefficient (ADC) images, from which a total of 804 radiomics features were extracted. Thereafter, we used Spearman correlation analysis and gradient boosting decision tree (GBDT) algorithm to select the strongest features, and the radiomics scores were established using multivariate logistic regression algorithm, including two single-mode models and two dual-mode models. The predictive performance and the clinical usefulness of the model were assessed by the receiver operating characteristic (ROC) curve, calibration curve and decision curve analysis (DCA). Integrated model A based on T2WI and ADC images showed a better predictive performance, which yielded an AUC of 0.770 (95% CI 0.673-0.867) in the training cohort and 0.768 (95% CI 0.619-0.917) in the validation cohort. Calibration curve showed good agreement between predicted results of the model and actual events, and DCA indicated good clinical usefulness. Moreover, stratification analysis proved that the integrated model A had strong robustness. Integrated model A based on T2WI and ADC images has the potential to be used as a non-invasive tool for preoperative prediction of immunoscore in rectal cancer. It may be useful in evaluating prognosis and guiding individualized immunotherapy of patients.

Cross-Modal Distillation to Improve MRI-Based Brain Tumor Segmentation With Missing MRI Sequences.

Convolutional neural networks (CNNs) for brain tumor segmentation are generally developed using complete sets of magnetic resonance imaging (MRI) sequences for both training and inference. As such, these algorithms are not trained for realistic, clinical scenarios where parts of the MRI sequences which were used for training, are missing during inference. To increase clinical applicability, we proposed a cross-modal distillation approach to leverage the availability of multi-sequence MRI data for training and generate an enriched CNN model which uses only single-sequence MRI data for inference but outperforms a single-sequence CNN model. We assessed the performance of the proposed method for whole tumor and tumor core segmentation with multi-sequence MRI data available for training but only T1-weighted ([Formula: see text]) sequence data available for inference, using BraTS 2018, and in-house datasets. Results showed that cross-modal distillation significantly improved the Dice score for both whole tumor and tumor core segmentation when only [Formula: see text] sequence data were available for inference. For the evaluation using the in-house dataset, cross-modal distillation achieved an average Dice score of 79.04% and 69.39% for whole tumor and tumor core segmentation, respectively, while a single-sequence U-Net model using [Formula: see text] sequence data for both training and inference achieved an average Dice score of 73.60% and 62.62%, respectively. These findings confirmed cross-modal distillation as an effective method to increase the potential of single-sequence CNN models such that segmentation performance is less compromised by missing MRI sequences or having only one MRI sequence available for segmentation.

Transformer-Based Deep-Learning Algorithm for Discriminating Demyelinating Diseases of the Central Nervous System With Neuroimaging.

BackgroundDifferential diagnosis of demyelinating diseases of the central nervous system is a challenging task that is prone to errors and inconsistent reading, requiring expertise and additional examination approaches. Advancements in deep-learning-based image interpretations allow for prompt and automated analyses of conventional magnetic resonance imaging (MRI), which can be utilized in classifying multi-sequence MRI, and thus may help in subsequent treatment referral.MethodsImaging and clinical data from 290 patients diagnosed with demyelinating diseases from August 2013 to October 2021 were included for analysis, including 67 patients with multiple sclerosis (MS), 162 patients with aquaporin 4 antibody-positive (AQP4+) neuromyelitis optica spectrum disorder (NMOSD), and 61 patients with myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD). Considering the heterogeneous nature of lesion size and distribution in demyelinating diseases, multi-modal MRI of brain and/or spinal cord were utilized to build the deep-learning model. This novel transformer-based deep-learning model architecture was designed to be versatile in handling with multiple image sequences (coronal T2-weighted and sagittal T2-fluid attenuation inversion recovery) and scanning locations (brain and spinal cord) for differentiating among MS, NMOSD, and MOGAD. Model performances were evaluated using the area under the receiver operating curve (AUC) and the confusion matrices measurements. The classification accuracy between the fusion model and the neuroradiological raters was also compared.ResultsThe fusion model that was trained with combined brain and spinal cord MRI achieved an overall improved performance, with the AUC of 0.933 (95%CI: 0.848, 0.991), 0.942 (95%CI: 0.879, 0.987) and 0.803 (95%CI: 0.629, 0.949) for MS, AQP4+ NMOSD, and MOGAD, respectively. This exceeded the performance using the brain or spinal cord MRI alone for the identification of the AQP4+ NMOSD (AUC of 0.940, brain only and 0.689, spinal cord only) and MOGAD (0.782, brain only and 0.714, spinal cord only). In the multi-category classification, the fusion model had an accuracy of 81.4%, which was significantly higher compared to rater 1 (64.4%, p=0.04<0.05) and comparable to rater 2 (74.6%, p=0.388).ConclusionThe proposed novel transformer-based model showed desirable performance in the differentiation of MS, AQP4+ NMOSD, and MOGAD on brain and spinal cord MRI, which is comparable to that of neuroradiologists. Our model is thus applicable for interpretating conventional MRI in the differential diagnosis of demyelinating diseases with overlapping lesions.

Early brainstem hemorrhage progression: multi-sequence magnetic resonance imaging and histopathology.

According to clinical statistics, the mortality of patients with early brainstem hemorrhage is high. In this study, we established rat models of brainstem hemorrhage by injecting type VII collagenase into the right basotegmental pontine and investigated the pathological changes of early brainstem hemorrhage using multi-sequence magnetic resonance imaging and histopathological methods. We found that brainstem hematoma gradually formed in the injured rats over the first 3 days and then reduced after 7 days. The edema that occurred was mainly of the vasogenic type. No complete myelin sheath structure was found around the focus of the brainstem hemorrhage. The integrity and continuity of nerve fibers gradually deteriorated over the first 7 days. Neuronal degeneration was mild in the first 3 days and then obviously aggravated on the 7th day. Inflammatory cytokines, interleukin-1β, and tumor necrosis factor α appeared on the 1st day after intracerebral hemorrhage, reached peak levels on the 3rd day, and decreased from the 7th day. Our findings show the characteristics of the progression of early brainstem hemorrhage.

Brain structural alterations and clinical features of cognitive frailty in Japanese community-dwelling older adults: the Arao study (JPSC-AD)

Cognitive frailty (CF) is a clinical condition defined by the presence of both mild cognitive impairment (MCI) and physical frailty (PF). Elderly with CF are at greater risk of dementia than those with MCI or PF alone, but there are few known clinical or neuroimaging features to reliably distinguish CF from PF or MCI. We therefore conducted a population-based cross-sectional study of community elderly combining physical, cognitive, neuropsychiatric, and multisequence magnetic resonance imaging (MRI) evaluations. The MRI evaluation parameters included white matter (WM) lesion volumes, perivascular and deep subcortical WM lesion grades, lacunar infarct prevalence, microbleed number, and regional medial temporal lobe (MTL) volumes. Participants were divided into 4 groups according to the presence or absence of MCI and PF—(1) no MCI, PF (n = 27); (2) no PF, MCI (n = 119); (3) CF (MCI + PF) (n = 21), (4) normal controls (n = 716). Unique features of CF included shorter one-leg standing time; severe depressive symptoms; and MRI signs of significantly more WM lesions, lacunar infarcts, small-vessel disease lesions, microbleeds, and reduced MTL volumes. These unique deficits suggest that interventions for CF prevention and treatment should focus on motor skills, depressive symptoms, and vascular disease risk factor control.