Decreased brain iron deposition based on quantitative susceptibility mapping correlates with reduced neurodevelopmental status in children with autism spectrum disorder.

To investigate potential correlations between the susceptibility values of certain brain regions and the severity of disease or neurodevelopmental status in children with autism spectrum disorder (ASD), 18 ASD children and 15 healthy controls (HCs) were recruited. The neurodevelopmental status was assessed by the Gesell Developmental Schedules (GDS) and the severity of the disease was evaluated by the Autism Behavior Checklist (ABC). Eleven brain regions were selected as regions of interest and the susceptibility values were measured by quantitative susceptibility mapping. To evaluate the diagnostic capacity of susceptibility values in distinguishing ASD and HC, the receiver operating characteristic (ROC) curve was computed. Pearson and Spearman partial correlation analysis were used to depict the correlations between the susceptibility values, the ABC scores, and the GDS scores in the ASD group. ROC curves showed that the susceptibility values of the left and right frontal white matter had a larger area under the curve in the ASD group. The susceptibility value of the right globus pallidus was positively correlated with the GDS-fine motor scale score. These findings indicated that the susceptibility value of the right globus pallidus might be a viable imaging biomarker for evaluating the neurodevelopmental status of ASD children.

HCA-DAN: hierarchical class-aware domain adaptive network for gastric tumor segmentation in 3D CT images.

BACKGROUND: Accurate segmentation of gastric tumors from CT scans provides useful image information for guiding the diagnosis and treatment of gastric cancer. However, automated gastric tumor segmentation from 3D CT images faces several challenges. The large variation of anisotropic spatial resolution limits the ability of 3D convolutional neural networks (CNNs) to learn features from different views. The background texture of gastric tumor is complex, and its size, shape and intensity distribution are highly variable, which makes it more difficult for deep learning methods to capture the boundary. In particular, while multi-center datasets increase sample size and representation ability, they suffer from inter-center heterogeneity. METHODS: In this study, we propose a new cross-center 3D tumor segmentation method named Hierarchical Class-Aware Domain Adaptive Network (HCA-DAN), which includes a new 3D neural network that efficiently bridges an Anisotropic neural network and a Transformer (AsTr) for extracting multi-scale context features from the CT images with anisotropic resolution, and a hierarchical class-aware domain alignment (HCADA) module for adaptively aligning multi-scale context features across two domains by integrating a class attention map with class-specific information. We evaluate the proposed method on an in-house CT image dataset collected from four medical centers and validate its segmentation performance in both in-center and cross-center test scenarios. RESULTS: Our baseline segmentation network (i.e., AsTr) achieves best results compared to other 3D segmentation models, with a mean dice similarity coefficient (DSC) of 59.26%, 55.97%, 48.83% and 67.28% in four in-center test tasks, and with a DSC of 56.42%, 55.94%, 46.54% and 60.62% in four cross-center test tasks. In addition, the proposed cross-center segmentation network (i.e., HCA-DAN) obtains excellent results compared to other unsupervised domain adaptation methods, with a DSC of 58.36%, 56.72%, 49.25%, and 62.20% in four cross-center test tasks. CONCLUSIONS: Comprehensive experimental results demonstrate that the proposed method outperforms compared methods on this multi-center database and is promising for routine clinical workflows.

White matter structural changes before and after microvascular decompression for hemifacial spasm.

Hemifacial spasm (HFS) is a syndrome characterized by involuntary contractions of the facial muscles innervated by the ipsilateral facial nerve. Currently, microvascular decompression (MVD) is an effective treatment for HFS. Diffusion weighted imaging (DWI) is a non-invasive advanced magnetic resonance technique that allows us to reconstruct white matter (WM) virtually based on water diffusion direction. This enables us to model the human brain as a complex network using graph theory. In our study, we recruited 32 patients with HFS and 32 healthy controls to analyze and compare the topological organization of whole-brain white matter networks between the groups. We also explored the potential relationships between altered topological properties and clinical outcomes. Compared to the HC group, the white matter network was disrupted in both preoperative and postoperative groups of HFS patients, mainly located in the somatomotor network, limbic network, and default network (All P < 0.05, FDR corrected). There was no significant difference between the preoperative and postoperative groups (P > 0.05, FDR corrected). There was a correlation between the altered topological properties and clinical outcomes in the postoperative group of patients (All P < 0.05, FDR corrected). Our findings indicate that in HFS, the white matter structural network was disrupted before and after MVD, and that these alterations in the postoperative group were correlated with the clinical outcomes. White matter alteration here described may subserve as potential biomarkers for HFS and may help us identify patients with HFS who can benefit from MVD and thus can help us make a proper surgical patient selection.

Immune-related lncRNAs signature and radiomics signature predict the prognosis and immune microenvironment of glioblastoma multiforme.

BACKGROUND: Glioblastoma multiforme (GBM) is the most common primary malignant brain tumor in adults. This study aimed to construct immune-related long non-coding RNAs (lncRNAs) signature and radiomics signature to probe the prognosis and immune infiltration of GBM patients. METHODS: We downloaded GBM RNA-seq data and clinical information from The Cancer Genome Atlas (TCGA) project database, and MRI data were obtained from The Cancer Imaging Archive (TCIA). Then, we conducted a cox regression analysis to establish the immune-related lncRNAs signature and radiomics signature. Afterward, we employed a gene set enrichment analysis (GSEA) to explore the biological processes and pathways. Besides, we used CIBERSORT to estimate the abundance of tumor-infiltrating immune cells (TIICs). Furthermore, we investigated the relationship between the immune-related lncRNAs signature, radiomics signature and immune checkpoint genes. Finally, we constructed a multifactors prognostic model and compared it with the clinical prognostic model. RESULTS: We identified four immune-related lncRNAs and two radiomics features, which show the ability to stratify patients into high-risk and low-risk groups with significantly different survival rates. The risk score curves and Kaplan-Meier curves confirmed that the immune-related lncRNAs signature and radiomics signature were a novel independent prognostic factor in GBM patients. The GSEA suggested that the immune-related lncRNAs signature were involved in L1 cell adhesion molecular (L1CAM) interactions and the radiomics signature were involved signaling by Robo receptors. Besides, the two signatures was associated with the infiltration of immune cells. Furthermore, they were linked with the expression of critical immune genes and could predict immunotherapy's clinical response. Finally, the area under the curve (AUC) (0.890,0.887) and C-index (0.737,0.817) of the multifactors prognostic model were greater than those of the clinical prognostic model in both the training and validation sets, indicated significantly improved discrimination. CONCLUSIONS: We identified the immune-related lncRNAs signature and tradiomics signature that can predict the outcomes, immune cell infiltration, and immunotherapy response in patients with GBM.

USBDAN: Unsupervised Scale-aware and Boundary-aware Domain Adaptive Network for Gastric Tumor Segmentation.

Accurate segmentation of gastric tumors from computed tomography (CT) images provides useful image information for guiding the diagnosis and treatment of gastric cancer. Researchers typically collect datasets from multiple medical centers to increase sample size and representation, but this raises the issue of data heterogeneity. To this end, we propose a new cross-center 3D tumor segmentation method named unsupervised scale-aware and boundary-aware domain adaptive network (USBDAN), which includes a new 3D neural network that efficiently bridges an Anisotropic neural network and a Transformer (AsTr) for extracting multi-scale features from the CT images with anisotropic resolution, and a scale-aware and boundary-aware domain alignment (SaBaDA) module for adaptively aligning multi-scale features between two domains and enhancing tumor boundary drawing based on location-related information drawn from each sample across all domains. We evaluate the proposed method on an in-house CT image dataset collected from four medical centers. Our results demonstrate that the proposed method outperforms several state-of-the-art methods.

Comparison of MRI radiomics-based machine learning survival models in predicting prognosis of glioblastoma multiforme.

Objective: To compare the performance of radiomics-based machine learning survival models in predicting the prognosis of glioblastoma multiforme (GBM) patients. Methods: 131 GBM patients were included in our study. The traditional Cox proportional-hazards (CoxPH) model and four machine learning models (SurvivalTree, Random survival forest (RSF), DeepSurv, DeepHit) were constructed, and the performance of the five models was evaluated using the C-index. Results: After the screening, 1792 radiomics features were obtained. Seven radiomics features with the strongest relationship with prognosis were obtained following the application of the least absolute shrinkage and selection operator (LASSO) regression. The CoxPH model demonstrated that age (HR = 1.576, p = 0.037), Karnofsky performance status (KPS) score (HR = 1.890, p = 0.006), radiomics risk score (HR = 3.497, p = 0.001), and radiomics risk level (HR = 1.572, p = 0.043) were associated with poorer prognosis. The DeepSurv model performed the best among the five models, obtaining C-index of 0.882 and 0.732 for the training and test set, respectively. The performances of the other four models were lower: CoxPH (0.663 training set / 0.635 test set), SurvivalTree (0.702/0.655), RSF (0.735/0.667), DeepHit (0.608/0.560). Conclusion: This study confirmed the superior performance of deep learning algorithms based on radiomics relative to the traditional method in predicting the overall survival of GBM patients; specifically, the DeepSurv model showed the best predictive ability.

Investigation of functional connectivity in Bell's palsy using functional magnetic resonance imaging: prospective cross-sectional study.

Background: The most common cause of lower motor neuron facial palsy is Bell's palsy (BP). BP results in partial or complete inability to automatically move the facial muscles on the affected side and, in some cases, to close the eyelids, which can cause permanent eye damage. This study investigated changes in brain function and connectivity abnormalities in patients with BP. Methods: This study included 46 patients with unilateral BP and 34 healthy controls (HCs). Resting-state brain functional magnetic resonance imaging (fMRI) images were acquired, and Toronto Facial Grading System (TFGS) scores were obtained for all participants. The fractional amplitude of low-frequency fluctuation (fALFF) was estimated, and the relationship between the TFGS and fALFF was determined using correlation analysis for brain regions with changes in fALFF in those with BP versus HCs. Brain regions associated with TFGS were used as seeds for further functional connectivity (FC) analysis; relationships between FC values of abnormal areas and TFGS scores were also analyzed. Results: Activation of the right precuneus, right angular gyrus, left supramarginal gyrus, and left middle occipital gyrus was significantly decreased in the BP group. fALFF was significantly higher in the right thalamus, vermis, and cerebellum of the BP group compared with that in the HC group (P<0.05). The FC between the left middle occipital gyrus and right angular gyrus, left precuneus, and right middle frontal gyrus increased sharply, but decreased in the left angular gyrus, left posterior cingulate gyrus, left middle frontal gyrus, inferior cerebellum, and left middle temporal gyrus. Furthermore, the fALFF in the left middle occipital gyrus was negatively correlated with TFGS score (R=0.144; P=0.008). Conclusions: The pathogenesis of BP is closely related to functional reorganization of the cerebral cortex. Patients with BP have altered fALFF activity in cortical regions associated with facial motion feedback monitoring.

MED27 plays a tumor-promoting role in breast cancer progression by targeting KLF4.

The mediator complex usually cooperates with transcription factors to be involved in RNA polymerase II-mediated gene transcription. As one component of this complex, MED27 has been reported in our previous studies to promote thyroid cancer and melanoma progression. However, the precise function of MED27 in breast cancer development remains poorly understood. Here, we found that MED27 was more highly expressed in breast cancer samples than in normal tissues, especially in triple-negative breast cancer, and its expression level was elevated with the increase in pathological stage. MED27 knockdown in triple-negative breast cancer cells inhibited cancer cell metastasis and stemness maintenance, which was accompanied by downregulation of the expression of EMT- and stem traits-associated proteins, and vice versa in non-triple-negative breast cancer. Furthermore, MED27 knockdown sensitized breast cancer cells to epirubicin treatment by inducing cellular apoptosis and reducing tumorsphere-forming ability. Based on RNA-seq, we identified KLF4 as the possible downstream target of MED27. KLF4 overexpression reversed the MED27 silencing-mediated arrest of cellular metastasis and stemness maintenance capacity in breast cancer in vitro and in vivo. Mechanistically, MED27 transcriptionally regulated KLF4 by binding to its promoter region at positions -156 to +177. Collectively, our study not only demonstrated the tumor-promoting role of MED27 in breast cancer progression by transcriptionally targeting KLF4, but also suggested the possibility of developing the MED27/KLF4 signaling axis as a potential therapeutic target in breast cancer.

Associations of cortical iron accumulation with cognition and cerebral atrophy in Alzheimer's disease.

Background: In Alzheimer's disease (AD), cerebral iron accumulation colocalizes with the pathological proteins amyloid-ß (Aß) and tau. Furthermore, tau-induced cortical thinning is associated with cognitive decline. In this study, quantitative susceptibility mapping (QSM) was used to investigate the whole-brain distribution pattern of cortical iron deposition and its relationships with cognition and cortical thickness in AD. Methods: This cross-sectional study prospectively recruited 30 participants with AD and 26 age- and sex-matched healthy controls (HCs). All participants underwent QSM and T1-weighted examinations on a 3.0T MRI scanner. Global cognition was assessed using the Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA). Whole-brain cross-sectional QSM analysis and whole-brain QSM regression analyses against the MMSE and MoCA scores were performed. Surface-based morphometry analysis was also performed. Subsequently, in regions with significant atrophy, magnetic susceptibility was compared between the AD and HC groups, and the association between magnetic susceptibility and cortical thickness was assessed. Results: Whole-brain QSM cross-sectional analysis in the AD group demonstrated widespread increased susceptibility across the cortical ribbon, asymmetrically covering the left hemisphere cerebral cortex, caudate nucleus, putamen, and partial cerebellar cortex. Whole-brain QSM regression analyses in the AD group showed that increased susceptibility covaried with lower MMSE and MoCA scores, and was predominantly located in the right parietal cortex and lateral occipital cortex. In the AD group, cortical thickness was reduced in the left superior temporal gyrus, right frontal pole, fusiform gyus, and pars opercularis, and there were increases in susceptibility in the right frontal pole (AD: mean ± SD 0.034±0.007 ppm, 95% CI: 0.032-0.037 ppm; HC: 0.030±0.005 ppm, 95% CI: 0.028-0.032 ppm; P=0.016) and pars opercularis (AD: 0.020±0.003 ppm, 95% CI: 0.018-0.021 ppm; HC: 0.017±0.002 ppm, 95% CI: 0.017-0.018 ppm; P=0.002). Susceptibility was negatively correlated with cortical thickness in the right pars opercularis in the entire cohort (r=-0.521, P<0.001) and AD group (r=-0.510, P=0.005). Conclusions: Widespread cortical iron, as measured by QSM, accumulated in AD and iron deposition was associated with poor cognitive performance. Increased iron content was also associated with brain atrophy. Our study suggests QSM may be a useful imaging biomarker for monitoring the neurodegenerative progression of AD.

BPTF inhibition antagonizes colorectal cancer progression by transcriptionally inactivating Cdc25A.

As the largest subunit of the nuclear remodeling factor complex, Bromodomain PHD Finger Transcription Factor (BPTF) has been reported to be involved in tumorigenesis and development in several cancers. However, to date, its functions and related molecular mechanisms in colorectal cancer (CRC) are still poorly defined and deserve to be revealed. In this study, we uncovered that, under the expression regulation of c-Myc, BPTF promoted CRC progression by targeting Cdc25A. BPTF was found to be highly expressed in CRC and promoted the proliferation and metastasis of CRC cells through BPTF specific siRNAs, shRNAs or inhibitors. Based on RNA-seq, combined with DNA-pulldown, ChIP and luciferase reporter assay, we proved that, by binding to -178/+107 region within Cdc25A promoter, BPTF transcriptionally activated Cdc25A, thus accelerating the cell cycle process of CRC cells. Meanwhile, BPTF itself was found to be transcriptionally regulated by c-Myc. Moreover, BPTF knockdown or inactivation was verified to sensitize CRC cells to chemotherapeutics, 5-Fluorouracil (5FU) and Oxaliplatin (Oxa), c-Myc inhibitor and cell cycle inhibitor not just at the cellular level in vitro, but in subcutaneous xenografts or AOM/DSS-induced in situ models of CRC in mice, while Cdc25A overexpression partially reversed BPTF silencing-caused tumor growth inhibition. Clinically, BPTF, c-Myc and Cdc25A were highly expressed in CRC tissues simultaneously, the expression of any two of the three was positively correlated, and their expressions were highly relevant to tumor differentiation, TNM staging and poor prognosis of CRC patients. Thus, our study indicated that the targeted inhibition of BPTF alone, or together with chemotherapy and/or cell cycle-targeted therapy, might act as a promising new strategy for CRC treatment, while c-Myc/BPTF/Cdc25A signaling axis is expected to be developed as an associated set of candidate biomarkers for CRC diagnosis and prognosis prediction.

Unsupervised domain selective graph convolutional network for preoperative prediction of lymph node metastasis in gastric cancer.

Preoperative prediction of lymph node (LN) metastasis based on computed tomography (CT) scans is an important task in gastric cancer, but few machine learning-based techniques have been proposed. While multi-center datasets increase sample size and representation ability, they suffer from inter-center heterogeneity. To tackle the above issue, we propose a novel multi-source domain adaptation framework for this diagnosis task, which not only considers domain-invariant and domain-specific features, but also achieves the imbalanced knowledge transfer and class-aware feature alignment across domains. First, we develop a 3D improved feature pyramidal network (i.e., 3D IFPN) to extract common multi-level features from the high-resolution 3D CT images, where a feature dynamic transfer (FDT) module can promote the network's ability to recognize the small target (i.e., LN). Then, we design an unsupervised domain selective graph convolutional network (i.e., UDS-GCN), which mainly includes three types of components: domain-specific feature extractor, domain selector and class-aware GCN classifier. Specifically, multiple domain-specific feature extractors are employed for learning domain-specific features from the common multi-level features generated by the 3D IFPN. A domain selector via the optimal transport (OT) theory is designed for controlling the amount of knowledge transferred from source domains to the target domain. A class-aware GCN classifier is developed to explicitly enhance/weaken the intra-class/inter-class similarity of all sample pairs across domains. To optimize UDS-GCN, the domain selector and the class-aware GCN classifier provide reliable target pseudo-labels to each other in the iterative process by collaborative learning. The extensive experiments are conducted on an in-house CT image dataset collected from four medical centers to demonstrate the efficacy of our proposed method. Experimental results verify that the proposed method boosts LN metastasis diagnosis performance and outperforms state-of-the-art methods. Our code is publically available at https://github.com/infinite-tao/LN_MSDA.

Role of 18F-FDG PET/CT in detecting pelvic lymph-node metastases in patients with early-stage uterine cervical cancer: comparison with MRI findings.

OBJECTIVE: This study aimed to investigate the applied value of F-fluoro-2-dexoxyglucose (F-FDG) PET/computed tomography (CT) and MRI in detecting lymph-node metastasis in early-stage cervical cancer. MATERIALS AND METHODS: A retrospective study was performed on 87 early-stage cervical cancer patients evaluated with PET/CT and pelvic MRI before surgery. Histopathological evaluation of lymph nodes served as the diagnostic standard. F-FDG PET/CT and MRI images were analyzed and correlated with histopathological findings. RESULTS: The overall node-based sensitivity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of PET/CT were 91% (61/67), 78.2% (61/78), 99.4% (1079/1085), and 98% (1140/1163), respectively, which were higher than the corresponding values of MRI, at 37.3% (25/67), 61% (25/41), 96.3% (1080/1122), and 95% (1105/1163) (P<0.034). The difference in diagnostic efficacy for identifying node-based metastases between PET/CT and MRI was significant (PET/CT vs. MRI, 0.719 vs. 0.587, P=0.017). Meanwhile, the overall patient-based sensitivity, PPV, NPV, and accuracy of PET/CT were 100% (34/34), 87.2% (34/39), 100% (48/48), and 94.3% (82/87), respectively, whereas the corresponding MRI values were 44% (15/34), 65% (15/23),74% (45/61), and 69% (60/87) (P<0.04). The difference in diagnostic efficacy for identifying patient-based metastases between PET/CT and MRI was significant (PET/CT vs. MRI, 0.974 vs. 0.705, P<0.001). CONCLUSION: PET/CT has been proven to be valuable in detecting lymph-node metastases. Compared with MRI, PET/CT has higher sensitivity, PPV, NPV, and accuracy in patients with early-stage cervical cancer for detecting lymphatic metastases.