When applying deep learning algorithms to magnetic resonance (MR) image segmentation, a large number of annotated images are required as data support. However, the specificity of MR images makes it difficult and costly to acquire large amounts of annotated image data. To reduce the dependence of MR image segmentation on a large amount of annotated data, this paper proposes a meta-learning U-shaped network (Meta-UNet) for few-shot MR image segmentation. Meta-UNet can use a small amount of annotated image data to complete the task of MR image segmentation and obtain good segmentation results. Meta-UNet improves U-Net by introducing dilated convolution, which can increase the receptive field of the model to improve the sensitivity to targets of different scales. We introduce the attention mechanism to improve the adaptability of the model to different scales. We introduce the meta-learning mechanism, and employ a composite loss function for well-supervised and effective bootstrapping of model training. We use the proposed Meta-UNet model to train on different segmentation tasks, and then use the trained model to evaluate on a new segmentation task, where the Meta-UNet model achieves high-precision segmentation of target images. Meta-UNet has a certain improvement in mean Dice similarity coefficient (DSC) compared with voxel morph network (VoxelMorph), data augmentation using learned transformations (DataAug) and label transfer network (LT-Net). Experiments show that the proposed method can effectively perform MR image segmentation using a small number of samples. It provides a reliable aid for clinical diagnosis and treatment.
OBJECTIVE:To investigate the diagnostic meaning of MRI in intraocular tumors. METHODS:Forty-six cases of confirmed intraocular tumors,including choroidal melanoma(20 cases),retinoblastoma(18 cases),Coats disease(6 cases)and choroidal hemangioma(2 cases),were studied with MRI and compared with ultrasonography and CT. RESULTS:In making discoveries about intraocular tumors,there were no sighificant difference between MRI and B-ultrasonography or CT (P>0.03,chi;2=1.0716)while there were highly statistic sighificance in dediding characters and position (P<0.01,deceding character chi;2=29.8314,positionchi;2=13.659)of them. CONCLUSION:Among the examinations to find out about the position,character and secondary pathological insults of in traocular tumors MRI might be more available than CT and ultrasonography. (Chin J Ocul Fundus Dis,1997,13:93-95 )
Objective To evaluate the clinical value of cardiac MRI for the diagnosis of viral myocarditis (VMC). Methods Such databases as PubMed (1950 to 2009), EMbase (1974 to 2009), and The Cochrane Library (December 2009) were searched to include clinical research reports of diagnosing viral myocarditis with MRI. QUADAS items were used to evaluate the quality of the included studies. The Meta-disc software was used to conduct merger analyses on sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio. The Heterogeneity test was performed and summary receiver operating characteristic curve (SROC) was completed. Results Five trials were included. The value of merger sensitivity, specificity, and diagnostic odds ratio (DOR) were 0.94, 0.69, 2.76, and 28.11, respectively. The area under of SROC curve (AUC) was 0.871 9. Conclusion The current evidence shows that cardiac MRI has high sensitivity (94%) and moderate specificity (69%) in the diagnosis of viral myocarditis. The positive rate in the viral myocarditis group is 28.11 times as high as that in the non-viral myocarditis group, so Cardiac MRI has good diagnostic values for viral myocarditis.
ObjectiveTo make the model of Wistar suckling rats Focal cortical dysplasia (FCD) by liquid nitrogen freezing brain cortex and verify it. Analysed the electroencephalogram (EEG) and magnetic resonance imaging (MRI) features of the FCD model, in order to provide theoretical and experimental basis for human FCD diagnosis and treatment. MethodsTake the first day of Wistar suckling rats as experimental object, liquid nitrogen freezing Wistar suckling rats brain cortex.Make examination of EEG and MRI for Wistar suckling rats. The Brain tissue slice of Wistar suckling rats model dyed by HE and check with light microscope examination. ResultsIn experiment group, the sample epileptic discharge rate of EEG was about 41.6% on average, and showed visible spike wave, spine slow wave frequency distribution. Experimental Wistar suckling rats MRI showed positive performance for long T1 and long T2 signal, brain tissue slices HE staining showed brain cortex layer structure and columnar structure disorder, exist abnormal neurons and the balloon sample cells. ConclusionThe method of liquid nitrogen freezing Wistar suckling rats cortex can established FCDⅢd animal models successfully, and showed specific EEG and MRI, which has important value for diagnosis and treatment of human FCD.
Focus on the inconsistency of the shape, location and size of brain glioma, a dual-channel 3-dimensional (3D) densely connected network is proposed to automatically segment brain glioma tumor on magnetic resonance images. Our method is based on a 3D convolutional neural network frame, and two convolution kernel sizes are adopted in each channel to extract multi-scale features in different scales of receptive fields. Then we construct two densely connected blocks in each pathway for feature learning and transmission. Finally, the concatenation of two pathway features was sent to classification layer to classify central region voxels to segment brain tumor automatically. We train and test our model on open brain tumor segmentation challenge dataset, and we also compared our results with other models. Experimental results show that our algorithm can segment different tumor lesions more accurately. It has important application value in the clinical diagnosis and treatment of brain tumor diseases.
ObjectiveTo investigate the CT and MR imaging manifestation of solid-pseudopapillary neoplasm of pancreas (SPNP), deepen the understanding of imaging and clinical pathological characteristics of SPNP and improve the level of diagnosis. MethodsBetween Jan 2010 and Dec 2015, the CT and MR imaging data of seven patients with SPTP proved by surgery and histopathologically were analyzed retrospectively. The following imaging features were reviewed: tumor size, location, shape, margin, encapsulation, calcification, hemorrhage, solid-cystic ratio, pancreatic and bile duct dilatation, the manifestation of plain scan and dynamic pattern of enhancement. ResultsThe population comprised 7 women, the average age was 28.3 years oldwith a median tumor size of 5.7 cm. Tumors were located at body tail of pancreas in 5 cases, at the head in 1 case, and at the tail in 1 case. The tumor were exogenous in 5 cases, endogenous in 2 cases. Five tumors showed the regular margin, inregular in 2 cases. Four cases of plain and enhanced CT scan showed cystic-solid tumors, the solid and encapsulation part ofSPNP presented as hipo-, iso-density, and gradually enhancement after injecting contrast medium. Three cases were examined by MRI, 2 cases appeared hemorrhage, tumor located in the head of pancreas leaded to the secondary ducts dilatations in 1 case. Conciusions There are some characteristics in CT and MRI manifestation of SPNP. Accurate diagnosis meybe created by the imaging study combined with the clinical feature.