TIF1γ and SMAD4 regulation in colorectal cancer: impact on cell proliferation and liver metastasis.

We investigated the effects of transcriptional intermediary factor 1γ (TIF1γ) and SMAD4 on the proliferation and liver metastasis of colorectal cancer (CRC) cells through knockdown of TIF1γ and/or SMAD4 and knockdown of TIF1γ and/or restoration of SMAD4 expression. Furthermore, we examined TIF1γ and SMAD4 expression in human primary CRC and corresponding liver metastatic CRC specimens. TIF1γ promoted but SMAD4 inhibited the proliferation of CRC cells by competitively binding to activated SMAD2/SMAD3 complexes and then reversely regulating c-Myc, p21, p27, and cyclinA2 levels. Surprisingly, both TIF1γ and SMAD4 reduced the liver metastasis of all studied CRC cell lines via inhibition of MEK/ERK pathway-mediated COX-2, Nm23, uPA, and MMP9 expression. In patients with advanced CRC, reduced TIF1γ or SMAD4 expression was correlated with increased invasion and liver metastasis and was a significant, independent risk factor for recurrence and survival after radical resection. Patients with advanced CRC with reduced TIF1γ or SAMD4 expression had higher recurrence rates and shorter overall survival. TIF1γ and SMAD4 competitively exert contrasting effects on cell proliferation but act complementarily to suppress the liver metastasis of CRC via MEK/ERK pathway inhibition. Thus, reduced TIF1γ or SMAD4 expression in advanced CRC predicts earlier liver metastasis and poor prognosis.

Local and Global Context-Enhanced Lightweight CenterNet for PCB Surface Defect Detection.

Printed circuit board (PCB) surface defect detection is an essential part of the PCB manufacturing process. Currently, advanced CCD or CMOS sensors can capture high-resolution PCB images. However, the existing computer vision approaches for PCB surface defect detection require high computing effort, leading to insufficient efficiency. To this end, this article proposes a local and global context-enhanced lightweight CenterNet (LGCL-CenterNet) to detect PCB surface defects in real time. Specifically, we propose a two-branch lightweight vision transformer module with local and global attention, named LGT, as a complement to extract high-dimension features and leverage context-aware local enhancement after the backbone network. In the local branch, we utilize coordinate attention to aggregate more powerful features of PCB defects with different shapes. In the global branch, Bi-Level Routing Attention with pooling is used to capture long-distance pixel interactions with limited computational cost. Furthermore, a Path Aggregation Network (PANet) feature fusion structure is incorporated to mitigate the loss of shallow features caused by the increase in model depth. Then, we design a lightweight prediction head by using depthwise separable convolutions, which further compresses the computational complexity and parameters while maintaining the detection capability of the model. In the experiment, the LGCL-CenterNet increased the mAP@0.5 by 2% and 1.4%, respectively, in comparison to CenterNet-ResNet18 and YOLOv8s. Meanwhile, our approach requires fewer model parameters (0.542M) than existing techniques. The results show that the proposed method improves both detection accuracy and inference speed and indicate that the LGCL-CenterNet has better real-time performance and robustness.

MicroRNA-188-5p inhibits hepatocellular carcinoma proliferation and migration by targeting forkhead box N2.

BACKGROUND: This study aimed to identify the biological functions, expression modes, and possible mechanisms underlying the relationship between metastatic human hepatocellular carcinoma (HCC) and MicroRNA-188-5p (miR-188) dysregulation using cell lines. METHODS: A decrease in miR-188 was detected in low and high metastatic HCC cells compared to that in normal hepatic cells and non-invasive cell lines. Gain- and loss-of-function experiments were performed in vitro to investigate the role of miR-188 in cancer cell (Hep3B, HepG2, HLF, and LM3) proliferation and migration. RESULTS: miR-188 mimic transfection inhibited the proliferation of metastatic HLF and LM3 cells but not non-invasive HepG2 and Hep3B cells; nonetheless, miR-188 suppression promoted the growth of HLF and LM3 cells. miR-188 upregulation inhibited the migratory rate and invasive capacity of HLF and LM3, rather than HepG2 and Hep3B cells, whereas transfection of a miR-188 inhibitor in HLF and LM3 cells had the opposite effects. Dual-luciferase reporter assays and bioinformatics prediction confirmed that miR-188 could directly target forkhead box N2 (FOXN2) in HLF and LM3 cells. Transfection of miR-188 mimics reduced FOXN2 levels, whereas miR-188 inhibition resulted in the opposite result, in HLF and LM3 cells. Overexpression of FOXN2 in HLF and LM3 cells abrogated miR-188 mimic-induced downregulation of proliferation, migration, and invasion. In addition, we found that miR-188 upregulation impaired tumor growth in vivo. CONCLUSIONS: In summary, this study showed thatmiR-188 inhibits the proliferation and migration of metastatic HCC cells by targeting FOXN2.

Integrative analysis of DNA methylation and gene expression identify a six epigenetic driver signature for predicting prognosis in hepatocellular carcinoma.

DNA methylation is a crucial regulator of gene transcription in the etiology and pathogenesis of hepatocellular carcinoma (HCC). Thus, it is reasonable to identify DNA methylation-related prognostic markers. Currently, we aimed to make an integrative epigenetic analysis of HCC to identify the effectiveness of epigenetic drivers in predicting prognosis for HCC patients. By the software pipeline TCGA-Assembler 2, RNA-seq, and methylation data were downloaded and processed from The Cancer Genome Atlas. A bioconductor package MethylMix was utilized to incorporate gene expression and methylation data on all 363 samples and identify 589 epigenetic drivers with transcriptionally predictive. By univariate survival analysis, 72 epigenetic drivers correlated with overall survival (OS) were selected for further analysis in our training cohort. By the robust likelihood-based survival model, six epi-drivers (doublecortin domain containing 2, flavin containing monooxygenase 3, G protein-coupled receptor 171, Lck interacting transmembrane adaptor 1, S100 calcium binding protein P, small nucleolar RNA host gene 6) serving as prognostic markers was identified and then a DNA methylation signature for HCC (MSH) predicting OS was identified to stratify patients into low-risk and high-risk groups in the training cohort (p < 0.001). The capability of MSH was also assessed in the validation cohort (p = 0.002). Furthermore, a receiver operating characteristic curve confirmed MSH as an effective prognostic model for predicting OS in HCC patients in training area under curve (AUC = 0.802) and validation (AUC = 0.691) cohorts. Finally, a nomogram comprising MSH and pathologic stage was generated to predict OS in the training cohort, and it also operated effectively in the validation cohort (concordance index: 0.674). In conclusion, MSH, a six epi-drivers based signature, is a potential model to predict prognosis for HCC patients.

Genome-wide identification of a competing endogenous RNA network in cholangiocarcinoma.

Cholangiocarcinoma (CCA) is the second widespread liver tumor with relatively poor survival. Increasing evidence in recent studies showed long noncoding RNAs (lncRNAs) exert a crucial impact on the development and progression of CCA based on the mechanism of competing endogenous RNAs (ceRNAs). However, functional roles and regulatory mechanisms of lncRNA-regulated ceRNA in CCA, are only partially understood. The expression profile of messenger RNAs (mRNAs), lncRNAs, and microRNAs (miRNAs) downloaded from The Cancer Genome Atlas were comprehensively investigated. Differential expression of these three types of RNA between CCA and corresponding precancerous tissues were screened out for further analysis. On the basis of interactive information generated from miRDB, miRTarBase, TargetScan, and miRcode public databases, we then constructed an mRNA-miRNA-lncRNA regulatory network. Kyoto Encyclopedia of Genes and Genomes and Gene Ontology analyses were conducted to identify the biological function of the ceRNA network involved in CCA. As a result, 2883 mRNAs, 136 miRNAs, and 993 lncRNAs were screened out as differentially expressed RNAs in CCA. In addition, a ceRNA network in CCA was constructed, composing of 50 up and 27 downregulated lncRNAs, 14 up and 7 downregulated miRNAs, 29 up and 25 downregulated mRNAs. Finally, gene set enrichment and pathway analysis indicated our CCA-specific ceRNA network was related with cancer-related pathway and molecular function. In conclusion, our research identified a novel lncRNA-related ceRNA network in CCA, which might act as a potential therapeutic target for patients with CCA.

Activin A-Smad Signaling Mediates Connective Tissue Growth Factor Synthesis in Liver Progenitor Cells.

Liver progenitor cells (LPCs) are activated in chronic liver damage and may contribute to liver fibrosis. Our previous investigation reported that LPCs produced connective tissue growth factor (CTGF/CCN2), an inducer of liver fibrosis, yet the regulatory mechanism of the production of CTGF/CCN2 in LPCs remains elusive. In this study, we report that Activin A is an inducer of CTGF/CCN2 in LPCs. Here we show that expression of both Activin A and CTGF/CCN2 were upregulated in the cirrhotic liver, and the expression of Activin A positively correlates with that of CTGF/CCN2 in liver tissues. We go on to show that Activin A induced de novo synthesis of CTGF/CCN2 in LPC cell lines LE/6 and WB-F344. Furthermore, Activin A contributed to autonomous production of CTGF/CCN2 in liver progenitor cells (LPCs) via activation of the Smad signaling pathway. Smad2, 3 and 4 were all required for this induction. Collectively, these results provide evidence for the fibrotic role of LPCs in the liver and suggest that the Activin A-Smad-CTGF/CCN2 signaling in LPCs may be a therapeutic target of liver fibrosis.

Reduced expression of transcriptional intermediary factor 1 gamma promotes metastasis and indicates poor prognosis of hepatocellular carcinoma.

UNLABELLED: Transcriptional intermediary factor 1 gamma (TIF1γ) may play either a potential tumor-suppressor or -promoter role in cancer. Here we report on a critical role of TIF1γ in the progression of hepatocellular carcinoma (HCC). Reduced expression of TIF1γ was detected in HCC, especially in advanced HCC tissues, compared to adjacent noncancerous tissues. HCC patients with low TIF1γ expression had shorter overall survival times and higher recurrence rates than those with high TIF1γ expression. Reduced TIF1γ expression was an independent and significant risk factor for recurrence and survival after curative resection. In HCC cells, TIF1γ played a dual role: It promoted tumor growth in early-stage HCC, but not in advanced-stage HCC, whereas it inhibited invasion and metastasis in both early- and advanced-stage HCC. Mechanistically, we confirmed that TIF1γ inhibited transforming growth factor-ß/ Drosophila mothers against decapentaplegic protein (TGF-ß/Smad) signaling through monoubiquitination of Smad4 and suppressed the formation of Smad2/3/4 complex in HCC cells. TGF-ß-inducing cytostasis and metastasis were both inhibited by TIF1γ in HCC. We further proved that TIF1γ suppressed cyotstasis-related TGF-ß/Smad downstream c-myc down-regulation, as well as p21/cip1 and p15/ink4b up-regulation in early-stage HCC. Meanwhile, TGF-ß inducible epithelial-mesenchymal transition and TGF-ß/Smad downstream metastatic cascades, including phosphatase and tensin homolog deleted on chromosome ten down-regulation, chemokine (CXC motif) receptor 4 and matrix metalloproteinase 1 induction, and epidermal growth factor receptor- and protein kinase B-signaling transactivation, were inhibited by TIF1γ. In addition, we found that the down-regulation of TIF1γ in HCC was caused by hypermethylation of CpG islands in the TIF1γ promoter, and demonstrated that the combination of TIF1γ and phosphorylated Smad2 was a more powerful predictor of poor prognosis. CONCLUSION: TIF1γ regulates tumor growth and metastasis through inhibition of TGF-ß/Smad signaling and may serve as a novel prognostic biomarker in HCC.

Smad6 suppresses the growth and self-renewal of hepatic progenitor cells.

Activation of hepatic progenitor cells (HPCs) is commonly observed in chronic liver disease and Wnt/ß-catenin signaling plays a crucial role in the expansion of HPCs. However, the molecular mechanisms that regulate the activation of Wnt/ß-catenin signaling in the liver, especially in HPCs, remain largely elusive. Here, we reported that ectopic expression of Smad6 suppressed the proliferation and self-renewal of WB-F344 cells, a HPC cell line. Mechanistically, we found that Smad6 inhibited Wnt/ß-catenin signaling through promoting the interaction of C-terminal binding protein (CtBP) with ß-catenin/T-cell factor (TCF) complex to inhibit ß-catenin mediated transcriptional activation in WB-F344 cells. We used siRNA targeting ß-catenin to demonstrate that Wnt/ß-catenin signaling was required for the proliferation and self-renewal of HPCs. Taken together, these results suggest that Smad6 is a regulatory molecule which regulates the proliferation, self-renewal and Wnt/ß-catenin signaling in HPCs.

MiR-132 prohibits proliferation, invasion, migration, and metastasis in breast cancer by targeting HN1.

Accumulating evidence indicates that miRNAs play critical roles in tumorigenesis and cancer progression. This study aims to investigate the role and the underlying mechanism of miR-132 in breast cancer. Here, we report that miR-132 is significantly down-regulated in breast cancer tissues and cancer cell lines. Additional study identifies HN1 as a novel direct target of miR-132. MiR-132 down-regulates HN1 expression by binding to the 3' UTR of HN1 transcript, thereby, suppressing multiple oncogenic traits such as cancer cell proliferation, invasion, migration and metastasis in vivo and in vitro. Overexpression of HN1 restores miR-132-suppressed malignancy. Importantly, higher HN1 expression is significantly associated with worse overall survival of breast cancer patients. Taken together, our data demonstrate a critical role of miR-132 in prohibiting cell proliferation, invasion, migration and metastasis in breast cancer through direct suppression of HN1, supporting the potential utility of miR-132 as a novel therapeutic strategy against breast cancer.

The State of Systematic Therapies in Clinic for Hepatobiliary Cancers.

Hepatobiliary cancer (HBC) includes hepatocellular carcinoma and biliary tract carcinoma (cholangiocarcinoma and gallbladder carcinoma), and its morbidity and mortality are significantly correlated with disease stage. Surgery is the cornerstone of curative therapy for early stage of HBC. However, a large proportion of patients with HBC are diagnosed with advanced stage and can only receive systemic treatment. According to the results of clinical trials, the first-line and second-line treatment programs are constantly updated with the improvement of therapeutic effectiveness. In order to improve the therapeutic effect, reduce the occurrence of drug resistance, and reduce the adverse reactions of patients, the treatment of HBC has gradually developed from single-agent therapy to combination. The traditional therapeutic philosophy proposed that patients with advanced HBC are only amenable to systematic therapies. With some encouraging clinical trial results, the treatment concept has been revolutionized, and patients with advanced HBC who receive novel systemic combination therapies with multi-modality treatment (including surgery, transplant, TACE, HAIC, RT) have significantly improved survival time. This review summarizes the treatment options and the latest clinical advances of HBC in each stage and discusses future direction, in order to inform the development of more effective treatments for HBC.

DNeuroMAT: A Deep-Learning-Based Neuron Morphology Analysis Toolbox.

Digital reconstruction of neuronal structures from 3D neuron microscopy images is critical for the quantitative investigation of brain circuits and functions. Currently, neuron reconstructions are mainly obtained by manual or semiautomatic methods. However, these ways are labor-intensive, especially when handling the huge volume of whole brain microscopy imaging data. Here, we present a deep-learning-based neuron morphology analysis toolbox (DNeuroMAT) for automated analysis of neuron microscopy images, which consists of three modules: neuron segmentation, neuron reconstruction, and neuron critical points detection.

Deep learning in mesoscale brain image analysis: A review.

Mesoscale microscopy images of the brain contain a wealth of information which can help us understand the working mechanisms of the brain. However, it is a challenging task to process and analyze these data because of the large size of the images, their high noise levels, the complex morphology of the brain from the cellular to the regional and anatomical levels, the inhomogeneous distribution of fluorescent labels in the cells and tissues, and imaging artifacts. Due to their impressive ability to extract relevant information from images, deep learning algorithms are widely applied to microscopy images of the brain to address these challenges and they perform superiorly in a wide range of microscopy image processing and analysis tasks. This article reviews the applications of deep learning algorithms in brain mesoscale microscopy image processing and analysis, including image synthesis, image segmentation, object detection, and neuron reconstruction and analysis. We also discuss the difficulties of each task and possible directions for further research.

Exploring a four-gene risk model based on doxorubicin resistance-associated lncRNAs in hepatocellular carcinoma.

Background: Liver cancer is a lethal cancer type among which hepatocellular carcinoma (HCC) is the most common manifestation globally. Drug resistance is a central problem impeding the efficiency of HCC treatment. Long non-coding RNAs reportedly result in drug resistance. This study aimed to identify key lncRNAs associated with doxorubicin resistance and HCC prognosis. Materials and Methods: HCC samples with gene expression profiles and clinical data were accessed from public databases. We applied differential analysis to identify key lncRNAs that differed between HCC and normal samples and between drug-fast and control samples. We also used univariate Cox regression analysis to screen lncRNAs or genes associated with HCC prognosis. The least absolute shrinkage and selection operator (LASSO) was used to identify the key prognostic genes. Finally, we used receiver operating characteristic analysis to validate the effectiveness of the risk model. Results: The results of this study revealed RNF157-AS1 as a key lncRNA associated with both doxorubicin resistance and HCC prognosis. Metabolic pathways such as fatty acid metabolism and oxidative phosphorylation were enriched in RNF157-AS1-related genes. LASSO identified four protein-coding genes-CENPP, TSGA10, MRPL53, and BFSP1-to construct a risk model. The four-gene risk model effectively classified HCC samples into two risk groups with different overall survival. Finally, we established a nomogram, which showed superior performance in predicting the long-term prognosis of HCC. Conclusion: RNF157-AS1 may be involved in doxorubicin resistance and may serve as a potential therapeutic target. The four-gene risk model showed potential for the prediction of HCC prognosis.

Foreign Object Detection in Railway Images Based on an Efficient Two-Stage Convolutional Neural Network.

Foreign object intrusion is one of the main causes of train accidents that threaten human life and public property. Thus, the real-time detection of foreign objects intruding on the railway is important to prevent the train from colliding with foreign objects. Currently, the detection of railway foreign objects is mainly performed manually, which is prone to negligence and inefficient. In this study, an efficient two-stage framework is proposed for foreign object detection in railway images. In the first stage, a lightweight railway image classification network is established to classify any input railway images into one of two classes: normal or intruded. To enable real-time and accurate classification, we propose an improved inverted residual unit by introducing two improvements to the original inverted residual unit. First, the selective kernel convolution is used to dynamically select kernel size and learn multiscale features from railway images. Second, we employ a lightweight attention mechanism, called the convolutional block attention module, to exploit both spatial and channel-wise relationships between feature maps. In the second stage of our framework, the intruded image is fed to the foreign object detection network to further detect the location and class of the objects in the image. Experimental results confirm that the performance of our classification network is comparable to the widely used baselines, and it obtains outperforming efficiency. Moreover, the performances of the second-stage object detection are satisfying.

DeepRayburst for Automatic Shape Analysis of Tree-Like Structures in Biomedical Images.

Precise quantification of tree-like structures from biomedical images, such as neuronal shape reconstruction and retinal blood vessel caliber estimation, is increasingly important in understanding normal function and pathologic processes in biology. Some handcrafted methods have been proposed for this purpose in recent years. However, they are designed only for a specific application. In this paper, we propose a shape analysis algorithm, DeepRayburst, that can be applied to many different applications based on a Multi-Feature Rayburst Sampling (MFRS) and a Dual Channel Temporal Convolutional Network (DC-TCN). Specifically, we first generate a Rayburst Sampling (RS) core containing a set of multidirectional rays. Then the MFRS is designed by extending each ray of the RS to multiple parallel rays which extract a set of feature sequences. A Gaussian kernel is then used to fuse these feature sequences and outputs one feature sequence. Furthermore, we design a DC-TCN to make the rays terminate on the surface of tree-like structures according to the fused feature sequence. Finally, by analyzing the distribution patterns of the terminated rays, the algorithm can serve multiple shape analysis applications of tree-like structures. Experiments on three different applications, including soma shape reconstruction, neuronal shape reconstruction, and vessel caliber estimation, confirm that the proposed method outperforms other state-of-the-art shape analysis methods, which demonstrate its flexibility and robustness.

The state of therapy modalities in clinic for biliary tract cancer.

Biliary tract cancers (BTCs) include intrahepatic cholangiocarcinoma (iCCA), perihilar and distal cholangiocarcinoma (pCCA and dCCA), and gallbladder carcinoma based on the epithelial site of origin. BTCs are highly aggressive tumors associated with poor prognosis due to widespread metastasis and high recurrence. Surgery is the typical curative-intent treatment, yet the cornerstone of cure depends on the anatomical site of the primary tumor, and only a minority of patients (approximately 30%) has an indication necessitating surgery. Similarly, only a small subset of carefully selected patients with early iCCA who are not candidates for liver resection can opt for liver transplantation. Chemotherapy, target therapy, and immunotherapy are the main treatment options for patients who have advanced stage or unresectable disease. The genetic background of each cholangiocarcinoma subtype has been accurately described based on whole gene exome and transcriptome sequencing. Accordingly, precision medicine in targeted therapies has been identified to be aimed at distinct patient subgroups harboring unique molecular alterations. Immunotherapy such as immune checkpoint inhibitors (ICIs) was identified as antitumor responses in a minority of select patients. Current studies indicate that immunotherapy of adoptive cell therapy represents a promising approach in hematological and solid tumor malignancies, yet clinical trials are needed to validate its effectiveness in BTC. Herein, we review the progress of BTC treatment, stratified patients according to the anatomic subtypes of cholangiocarcinoma and the gene drivers of cholangiocarcinoma progression, and compare the efficacy and safety of chemotherapy, targeted therapy, and immunotherapy, which will be conducive to the design of individualized therapies.

Primary spindle cell sarcoma of gallbladder: An unusual case report and a literature review.

INTRODUCTION: Primary spindle cell sarcoma of the gallbladder is a rare condition. PATIENT CONCERNS: A 67-year-old woman was admitted to a local hospital with a chief complaint of abdominal pain in the right upper quadrant for the past 2 months. DIAGNOSIS AND INTERVENTION: Surgical resection was performed following the diagnosis of primary gallbladder sarcoma with local hepatic metastasis. Histological examination confirmed a diagnosis of primary spindle cell sarcoma and hepatic metastasis with simultaneous cholecystolithiasis. OUTCOMES: Adjuvant chemoradiation therapy was not performed because the patient refused treatment. Three months after the surgery, a relapsed lesion was diagnosed. The patient underwent transcatheter arterial chemoembolization. CONCLUSIONS: The disease should be differentially diagnosed from gallbladder carcinoma or carcinosarcoma with hepatic metastasis. An aggressive surgical approach should be based on a balance between the risk of surgery and the outcome.

Deep-Learning-Based Automated Neuron Reconstruction From 3D Microscopy Images Using Synthetic Training Images.

Digital reconstruction of neuronal structures from 3D microscopy images is critical for the quantitative investigation of brain circuits and functions. It is a challenging task that would greatly benefit from automatic neuron reconstruction methods. In this paper, we propose a novel method called SPE-DNR that combines spherical-patches extraction (SPE) and deep-learning for neuron reconstruction (DNR). Based on 2D Convolutional Neural Networks (CNNs) and the intensity distribution features extracted by SPE, it determines the tracing directions and classifies voxels into foreground or background. This way, starting from a set of seed points, it automatically traces the neurite centerlines and determines when to stop tracing. To avoid errors caused by imperfect manual reconstructions, we develop an image synthesizing scheme to generate synthetic training images with exact reconstructions. This scheme simulates 3D microscopy imaging conditions as well as structural defects, such as gaps and abrupt radii changes, to improve the visual realism of the synthetic images. To demonstrate the applicability and generalizability of SPE-DNR, we test it on 67 real 3D neuron microscopy images from three datasets. The experimental results show that the proposed SPE-DNR method is robust and competitive compared with other state-of-the-art neuron reconstruction methods.

3D Neuron Microscopy Image Segmentation via the Ray-Shooting Model and a DC-BLSTM Network.

The morphology reconstruction (tracing) of neurons in 3D microscopy images is important to neuroscience research. However, this task remains very challenging because of the low signal-to-noise ratio (SNR) and the discontinued segments of neurite patterns in the images. In this paper, we present a neuronal structure segmentation method based on the ray-shooting model and the Long Short-Term Memory (LSTM)-based network to enhance the weak-signal neuronal structures and remove background noise in 3D neuron microscopy images. Specifically, the ray-shooting model is used to extract the intensity distribution features within a local region of the image. And we design a neural network based on the dual channel bidirectional LSTM (DC-BLSTM) to detect the foreground voxels according to the voxel-intensity features and boundary-response features extracted by multiple ray-shooting models that are generated in the whole image. This way, we transform the 3D image segmentation task into multiple 1D ray/sequence segmentation tasks, which makes it much easier to label the training samples than many existing Convolutional Neural Network (CNN) based 3D neuron image segmentation methods. In the experiments, we evaluate the performance of our method on the challenging 3D neuron images from two datasets, the BigNeuron dataset and the Whole Mouse Brain Sub-image (WMBS) dataset. Compared with the neuron tracing results on the segmented images produced by other state-of-the-art neuron segmentation methods, our method improves the distance scores by about 32% and 27% in the BigNeuron dataset, and about 38% and 27% in the WMBS dataset.

Neuron Image Segmentation via Learning Deep Features and Enhancing Weak Neuronal Structures.

Neuron morphology reconstruction (tracing) in 3D volumetric images is critical for neuronal research. However, most existing neuron tracing methods are not applicable in challenging datasets where the neuron images are contaminated by noises or containing weak filament signals. In this paper, we present a two-stage 3D neuron segmentation approach via learning deep features and enhancing weak neuronal structures, to reduce the impact of image noise in the data and enhance the weak-signal neuronal structures. In the first stage, we train a voxel-wise multi-level fully convolutional network (FCN), which specializes in learning deep features, to obtain the 3D neuron image segmentation maps in an end-to-end manner. In the second stage, a ray-shooting model is employed to detect the discontinued segments in segmentation results of the first-stage, and the local neuron diameter of the broken point is estimated and direction of the filamentary fragment is detected by rayburst sampling algorithm. Then, a Hessian-repair model is built to repair the broken structures, by enhancing weak neuronal structures in a fibrous structure determined by the estimated local neuron diameter and the filamentary fragment direction. Experimental results demonstrate that our proposed segmentation approach achieves better segmentation performance than other state-of-the-art methods for 3D neuron segmentation. Compared with the neuron reconstruction results on the segmented images produced by other segmentation methods, the proposed approach gains 47.83% and 34.83% improvement in the average distance scores. The average Precision and Recall rates of the branch point detection with our proposed method are 38.74% and 22.53% higher than the detection results without segmentation.