A modified U-Net convolutional neural network for segmenting periprostatic adipose tissue based on contour feature learning.

Objective: This study trains a U-shaped fully convolutional neural network (U-Net) model based on peripheral contour measures to achieve rapid, accurate, automated identification and segmentation of periprostatic adipose tissue (PPAT). Methods: Currently, no studies are using deep learning methods to discriminate and segment periprostatic adipose tissue. This paper proposes a novel and modified, U-shaped convolutional neural network contour control points on a small number of datasets of MRI T2W images of PPAT combined with its gradient images as a feature learning method to reduce feature ambiguity caused by the differences in PPAT contours of different patients. This paper adopts a supervised learning method on the labeled dataset, combining the probability and spatial distribution of control points, and proposes a weighted loss function to optimize the neural network's convergence speed and detection performance. Based on high-precision detection of control points, this paper uses a convex curve fitting to obtain the final PPAT contour. The imaging segmentation results were compared with those of a fully convolutional network (FCN), U-Net, and semantic segmentation convolutional network (SegNet) on three evaluation metrics: Dice similarity coefficient (DSC), Hausdorff distance (HD), and intersection over union ratio (IoU). Results: Cropped images with a 270 × 270-pixel matrix had DSC, HD, and IoU values of 70.1%, 27 mm, and 56.1%, respectively; downscaled images with a 256 × 256-pixel matrix had 68.7%, 26.7 mm, and 54.1%. A U-Net network based on peripheral contour characteristics predicted the complete periprostatic adipose tissue contours on T2W images at different levels. FCN, U-Net, and SegNet could not completely predict them. Conclusion: This U-Net convolutional neural network based on peripheral contour features can identify and segment periprostatic adipose tissue quite well. Cropped images with a 270 × 270-pixel matrix are more appropriate for use with the U-Net convolutional neural network based on contour features; reducing the resolution of the original image will lower the accuracy of the U-Net convolutional neural network. FCN and SegNet are not appropriate for identifying PPAT on T2 sequence MR images. Our method can automatically segment PPAT rapidly and accurately, laying a foundation for PPAT image analysis.

A clinical-radiomics model based on noncontrast computed tomography to predict hemorrhagic transformation after stroke by machine learning: a multicenter study.

OBJECTIVE: To build a clinical-radiomics model based on noncontrast computed tomography images to identify the risk of hemorrhagic transformation (HT) in patients with acute ischemic stroke (AIS) following intravenous thrombolysis (IVT). MATERIALS AND METHODS: A total of 517 consecutive patients with AIS were screened for inclusion. Datasets from six hospitals were randomly divided into a training cohort and an internal cohort with an 8:2 ratio. The dataset of the seventh hospital was used for an independent external verification. The best dimensionality reduction method to choose features and the best machine learning (ML) algorithm to develop a model were selected. Then, the clinical, radiomics and clinical-radiomics models were developed. Finally, the performance of the models was measured using the area under the receiver operating characteristic curve (AUC). RESULTS: Of 517 from seven hospitals, 249 (48%) had HT. The best method for choosing features was recursive feature elimination, and the best ML algorithm to build models was extreme gradient boosting. In distinguishing patients with HT, the AUC of the clinical model was 0.898 (95% CI 0.873-0.921) in the internal validation cohort, and 0.911 (95% CI 0.891-0.928) in the external validation cohort; the AUC of radiomics model was 0.922 (95% CI 0.896-0.941) and 0.883 (95% CI 0.851-0.902), while the AUC of clinical-radiomics model was 0.950 (95% CI 0.925-0.967) and 0.942 (95% CI 0.927-0.958) respectively. CONCLUSION: The proposed clinical-radiomics model is a dependable approach that could provide risk assessment of HT for patients who receive IVT after stroke.

Aquaporin-4 gene silencing protects injured neurons after early cerebral infarction.

Aquaporin-4 regulates water molecule channels and is important in tissue regulation and water transportation in the brain. Upregulation of aquaporin-4 expression is closely related to cellular edema after early cerebral infarction. Cellular edema and aquaporin-4 expression can be determined by measuring cerebral infarct area and apparent diffusion coefficient using diffusion-weighted imaging (DWI). We examined the effects of silencing aquaporin-4 on cerebral infarction. Rat models of cerebral infarction were established by occlusion of the right middle cerebral artery and siRNA-aquaporin-4 was immediately injected via the right basal ganglia. In control animals, the area of high signal intensity and relative apparent diffusion coefficient value on T2-weighted imaging (T2WI) and DWI gradually increased within 0.5-6 hours after cerebral infarction. After aquaporin-4 gene silencing, the area of high signal intensity on T2WI and DWI reduced, relative apparent diffusion coefficient value was increased, and cellular edema was obviously alleviated. At 6 hours after cerebral infarction, the apparent diffusion coefficient value was similar between treatment and model groups, but angioedema was still obvious in the treatment group. These results indicate that aquaporin-4 gene silencing can effectively relieve cellular edema after early cerebral infarction; and when conducted accurately and on time, the diffusion coefficient value and the area of high signal intensity on T2WI and DWI can reflect therapeutic effects of aquaporin-4 gene silencing on cellular edema.

Relationship between AQP4 expression and structural damage to the blood-brain barrier at early stages of traumatic brain injury in rats.

BACKGROUND: Although some studies have reported that aquaporin-4 (AQP4) plays an important role in the brain edema after traumatic brain injury (TBI), little is known about the AQP4 expression in the early stage of TBI, or about the correlation between the structural damage to the blood-brain barrier (BBB) and angioedema. The aim of this project was to investigate the relationship between AQP4 expression and damage to the BBB at early stages of TBI. METHODS: One hundred and twenty healthy adult Wistar rats were randomly divided into two groups: sham operation group (SO) and TBI group. The TBI group was divided into five sub-groups according to the different time intervals: 1, 3, 6, 12, and 24 hours. The brains of the animals were taken out at different time points after TBI to measure brain water content. The cerebral edema and BBB changes in structure were examined with an optical microscopy (OM) and transmission electron microscopy (TEM), and the IgG content and AQP4 protein expression in traumatic brain tissue were determined by means of immunohistochemistry and Western blotting. The data were analyzed with SPSS 13.0 statistical software. RESULTS: In the SO group, tissue was negative for IgG, and there were no abnormalities in brain water content or AQP4 expression. In the TBI group, brain water content significantly increased at 6 hours and peaked at 24 hours following injury. IgG expression significantly increased from 1 to 6 hours following injury, and remained at a high level at 24 hours. Pathological observation revealed BBB damage at 1 hour following injury. Angioedema appeared at 1 hour, was gradually aggravated, and became obvious at 6 hours. Intracellular edema occurred at 3 hours, with the presence of large glial cell bodies and mitochondrial swelling. These phenomena were aggravated with time and became obvious at 12 hours. In addition, microglial proliferation was visible at 24 hours. AQP4 protein expression were reduced at 1 hour, lowest at 6 hours, and began to increase at 12 hours, showing a V-shaped curve. CONCLUSIONS: The angioedema characterized by BBB damage was the primary type of early traumatic brain edema. It was followed by mixed cerebral edema that consisted of angioedema and cellular edema and was aggravated with time. AQP4 expression was down-regulated during the angioedema attack, but AQP4 expression was upregulated during intracellular edema.

Therapeutic imaging window of cerebral infarction revealed by multisequence magnetic resonance imaging: An animal and clinical study.

In this study, we established a Wistar rat model of right middle cerebral artery occlusion and observed pathological imaging changes (T2-weighted imaging [T2WI], T2FLAIR, and diffusion-weighted imaging [DWI]) following cerebral infarction. The pathological changes were divided into three phases: early cerebral infarction, middle cerebral infarction, and late cerebral infarction. In the early cerebral infarction phase (less than 2 hours post-infarction), there was evidence of intracellular edema, which improved after reperfusion. This improvement was defined as the ischemic penumbra. In this phase, a high DWI signal and a low apparent diffusion coefficient were observed in the right basal ganglia region. By contrast, there were no abnormal T2WI and T2FLAIR signals. For the middle cerebral infarction phase (2-4 hours post-infarction), a mixed edema was observed. After reperfusion, there was a mild improvement in cell edema, while the angioedema became more serious. A high DWI signal and a low apparent diffusion coefficient signal were observed, and some rats showed high T2WI and T2FLAIR signals. For the late cerebral infarction phase (4-6 hours post-infarction), significant angioedema was visible in the infarction site. After reperfusion, there was a significant increase in angioedema, while there was evidence of hemorrhage and necrosis. A mixed signal was observed on DWI, while a high apparent diffusion coefficient signal, a high T2WI signal, and a high T2FLAIR signal were also observed. All 86 cerebral infarction patients were subjected to T2WI, T2FLAIR, and DWI. MRI results of clinic data similar to the early infarction phase of animal experiments were found in 51 patients, for which 10 patients (10/51) had an onset time greater than 6 hours. A total of 35 patients had MRI results similar to the middle and late infarction phase of animal experiments, of which eight patients (8/35) had an onset time less than 6 hours. These data suggest that defining the "therapeutic time window" as the time 6 hours after infarction may not be suitable for all patients. Integrated application of MRI sequences including T2WI, T2FLAIR, DW-MRI, and apparent diffusion coefficient mapping should be used to examine the ischemic penumbra, which may provide valuable information for identifying the "therapeutic time window".

Gene interference regulates aquaporin-4 expression in swollen tissue of rats with cerebral ischemic edema: Correlation with variation in apparent diffusion coefficient.

To investigate the effects of mRNA interference on aquaporin-4 expression in swollen tissue of rats with ischemic cerebral edema, and diagnose the significance of diffusion-weighted MRI, we injected 5 µL shRNA- aquaporin-4 (control group) or siRNA- aquaporin-4 solution (1:800) (RNA interference group) into the rat right basal ganglia immediately before occlusion of the middle cerebral artery. At 0.25 hours after occlusion of the middle cerebral artery, diffusion-weighted MRI displayed a high signal; within 2 hours, the relative apparent diffusion coefficient decreased markedly, aquaporin-4 expression increased rapidly, and intracellular edema was obviously aggravated; at 4 and 6 hours, the relative apparent diffusion coefficient slowly returned to control levels, aquaporin-4 expression slightly increased, and angioedema was observed. In the RNA interference group, during 0.25-6 hours after injection of siRNA- aquaporin-4 solution, the relative apparent diffusion coefficient slightly fluctuated and aquaporin-4 expression was upregulated; during 0.5-4 hours, the relative apparent diffusion coefficient was significantly higher, while aquaporin-4 expression was significantly lower when compared with the control group, and intracellular edema was markedly reduced; at 0.25 and 6 hours, the relative apparent diffusion coefficient and aquaporin-4 expression were similar when compared with the control group; obvious angioedema remained at 6 hours. Pearson's correlation test results showed that aquaporin-4 expression was negatively correlated with the apparent diffusion coefficient (r = -0.806, P < 0.01). These findings suggest that upregulated aquaporin-4 expression is likely to be the main molecular mechanism of intracellular edema and may be the molecular basis for decreased relative apparent diffusion coefficient. Aquaporin-4 gene interference can effectively inhibit the upregulation of aquaporin-4 expression during the stage of intracellular edema with time-effectiveness. Moreover, diffusion-weighted MRI can accurately detect intracellular edema.

Reperfusion of the rat brain tissues following acute ischemia: the correlation among diffusion-weighted imaging, histopathology, and aquaporin-4 expression.

BACKGROUND: Although some studies have reported that aquaporin-4 (AQP4) plays a role in the post-ischemic edema formation and diffusion-weighted imaging (DWI), little is known about the AQP4 expression in stage of the reperfusion following acute cerebral ischemia, as well as the correlation between histopathology and DWI. The aim of the study was to investigate the correlation among DWI, histopathology and the AQP4 expression in the reperfused rat brain tissues following acute ischemia. METHODS: Seventy Wistar rats were randomly divided into a control group (group A), and several occluded and reperfusion groups. They had their middle cerebral artery unilaterally occluded (MCAO) for 30 minutes (group B) followed by 30 minutes (group D) or 60 minutes (group E) of reperfusion, or 60 minutes of MCAO (group C) followed by 30 minutes (group F), or 60 minutes (group G) of reperfusion (n = 10 for each group). All rats underwent DWI scanning. The relative apparent diffusion coefficient (rADC) value of each rat was calculated. All the rats were sacrificed and the cerebral ischemic tissues were examined for histopathology. Real-time fluro-quantitative polymerase chain reaction (RT-PCR) and Western-blotting were performed. The amount of AQP4 mRNA (Ex(-ΔΔCt)) and AQP4 protein (Q) was statistically analyzed. The correlation between rADC values and AQP4 mRNA expression was analyzed with the Pearson correlation test. RESULTS: In all the reperfusion groups, the areas of hyper-intensity signal in DWI were decreased, and the rADC value increased and the AQP4 expression decreased significantly compared with the occluded group (t = 26.89, t = 18.26, P < 0.01). There was a negative correlation between AQP4 mRNA expression and rADC values (r = -0.72, P < 0.01). A mixed edema, composed of cerebral intracelluar edema and vasogenic brain edema, was observed in all the reperfusion groups. It was more prevalent in groups D and F than in the groups E and G. With the reperfusion time postponed, the cerebral intracelluar edema of the rat was significantly mitigated, but the vasogenic brain edema was not significantly changed. CONCLUSIONS: There is a close correlation between AQP4 expression and the cerebral intracellular edema. The change of ADC values may indirectly reflect the level of the AQP4 expression. DWI may become a promising, noninvasive imaging modality to predict early stroke and reperfusion injury.