A lightweight hierarchical convolution network for brain tumor segmentation.

BACKGROUND: Brain tumor segmentation plays a significant role in clinical treatment and surgical planning. Recently, several deep convolutional networks have been proposed for brain tumor segmentation and have achieved impressive performance. However, most state-of-the-art models use 3D convolution networks, which require high computational costs. This makes it difficult to apply these models to medical equipment in the future. Additionally, due to the large diversity of the brain tumor and uncertain boundaries between sub-regions, some models cannot well-segment multiple tumors in the brain at the same time. RESULTS: In this paper, we proposed a lightweight hierarchical convolution network, called LHC-Net. Our network uses a multi-scale strategy which the common 3D convolution is replaced by the hierarchical convolution with residual-like connections. It improves the ability of multi-scale feature extraction and greatly reduces parameters and computation resources. On the BraTS2020 dataset, LHC-Net achieves the Dice scores of 76.38%, 90.01% and 83.32% for ET, WT and TC, respectively, which is better than that of 3D U-Net with 73.50%, 89.42% and 81.92%. Especially on the multi-tumor set, our model shows significant performance improvement. In addition, LHC-Net has 1.65M parameters and 35.58G FLOPs, which is two times fewer parameters and three times less computation compared with 3D U-Net. CONCLUSION: Our proposed method achieves automatic segmentation of tumor sub-regions from four-modal brain MRI images. LHC-Net achieves competitive segmentation performance with fewer parameters and less computation than the state-of-the-art models. It means that our model can be applied under limited medical computing resources. By using the multi-scale strategy on channels, LHC-Net can well-segment multiple tumors in the patient's brain. It has great potential for application to other multi-scale segmentation tasks.

Method for improving the accuracy of fluorescence molecular tomography based on multi-wavelength concurrent reconstruction.

A Concurrent-wavelength Reconstruction Algorithm (CRA) based on multi-wavelength information fusion is proposed in this paper that aims to further improve the accuracy of Fluorescence Molecular Tomography (FMT) reconstruction. Combining multi-spectral data with FMT technology, the information of 650 and 750 nm wavelengths near-infrared was used to increase the feature information of the dominant 850 nm wavelength near-infrared effectively. Principal component analysis, which can remove redundant information and achieve data dimensionality reduction, was then utilized to extract the feature information. Finally, tomographic reconstruction of the anomalies was performed based on the stacked auto-encoder neural network model. The comparison results of numerical experiments showed that the reconstruction effect of CRA was superior to the performance of the single wavelength model. The correlation coefficient between CRA reconstructed anomalies' fluorescence yield values and the real fluorescence yield values remained at 0.95 or more under the noise of different levels of signal-to-noise ratios. Therefore, the CRA proposed in this paper could effectively improve on the ill-posedness of the inverse problem, which could further enhance the accuracy of FMT reconstruction.

Goat Model for Direct Visualizing the Effectiveness of Detaching Sinus Mucosa in Real Time During Crestal Maxillary Sinus Floor Elevation.

The procedure of crestal maxillary sinus floor elevation presents a great challenge to the field of implant dentistry. Due to the limited visualization in this procedure, the effectiveness of detaching sinus mucosa could not be assessed in real time. We recently developed an ex vivo goat sinus model by cutting the goat residual skulls along four lines determined from computerized tomography (CT) scans, extracting the maxillary premolar or molar teeth, and preparing implant socket in the maxilla. The generated ex vivo goat sinus models exposed the maxilla and the whole maxillary sinus mucosa, thus enabling real-time observation of detaching maxillary sinus mucosa via directly visualizing the working situation of sinus lift tool in the models and directly measuring the length of detached mucosa and space volume generated under the elevated sinus mucosa. One commercially available umbrella-shaped sinus lift curette was used to detach the maxillary sinus mucosa to evaluate the effectiveness of the ex vivo goat sinus models. The results showed that this curette could detach the sinus mucosa 3.75 mm in length in the mesiodistal direction and 2.81 mm in the buccal-palatal direction. Moreover, a space volume of 52.7 µl could be created under the elevated sinus mucosa in the goat ex vivo models. All the experimental results suggested that this ex vivo goat sinus model might be useful in the evaluation of improved or newly designed sinus lift tools for elevating the maxillary sinus mucosa via the crestal approach.

Ex vivo comparative study on three sinus lift tools for transcrestal detaching maxillary sinus mucosa.

The objective of this study was to comparatively evaluate 3 different sinus lift tools, namely umbrella-shaped sinus lift curette YSL-04, our recently designed probe-improved sinus lift curettes, and our newly invented elevator 014, using our previous developed goat ex vivo models for direct visualizing the effectiveness of detaching sinus mucosa in real time. Goat ex vivo models for direct visualizing the effectiveness of detaching sinus mucosa in real time were generated according to our previously developed protocol. The effectiveness for each tool was evaluated through the length of sinus mucosa detached in mesial and distal directions or buccal and palatal directions, and the space volume created by detaching maxillary sinus mucosa in mesial, distal, buccal and palatal directions. The results showed that all 3 sinus lift tools could transcrestally detach the maxillary sinus mucosa and create extra space under the elevated sinus floor on the goat ex vivo sinus models. Moreover, our newly invented elevator 014 had advantages over the other 2 in term of the capability to detach the sinus mucosa. Our newly invented elevator 014 might be a promising tool for detaching maxillary sinus mucosa in transcrestal maxillary sinus floor elevation.

[Non-linear research of alertness levels under sleep deprivation].

We applied Lempel-Ziv complexity (LZC) combined with brain electrical activity mapping (BEAM) to study the change of alertness under sleep deprivation in our research. Ten subjects were involved in 36 hours sleep deprivation (SD), during which spontaneous electroencephalogram (EEG) experiments and auditory evoked EEG experiments-Oddball were recorded once every 6 hours. Spontaneous and evoked EEG data were calculated and BEAMs were structured. Results showed that during the 36 hours of SD, alertness could be divided into three stages, i. e. the first 12 hours as the high stage, the middle 12 hours as the rapid decline stage and the last 12 hours as the low stage. During the period SD, LZC of Spontaneous EEG decreased over the whole brain to some extent, but remained consistent with the subjective scales. By BEAMs of event related potential, LZC on frontal cortex decreased, but kept consistent with the behavioral responses. Therefore, LZC can be effective to reflect the change of brain alertness. At the same time LZC could be used as a practical index to monitor real-time alertness because of its simple computation and fast calculation.

[Research on individual sleep staging based on principal component analysis and support vector machine].

The research of sleep staging is an important basis of evaluating sleep quality and diagnosing diseases. In order to achieve automatic sleep staging, we proposed a new method which combines with principal component analysis (PCA) and support vector machine (SVM) for automatic sleep staging. Firstly, we used PCA to reduce dimension of time-frequency-space domains and nonlinear dynamical characteristics of sleep EEG from 5 subjects to reduce data redundancy. Secondly, we used 1-a-1 SVM to classify sleep stages. The results showed that the correct rate can reach 89.9%, which was better than those of many other similar studies.

4,6-Dimethyl-2-(naphthalen-1-yl)pyrimidine.

The asymmetric unit of the title compound, C(16)H(14)N(2), contains two independent mol-ecules in which the dihedral angles between the pyrimidine and naphthaline rings are 38.20 (5) and 39.35 (5)°. Inter-molecular C-H⋯π contacts and π-π stacking inter-actions [centroid-centroid distances = 3.766 (1) and 3.792 (1) Å] are present in the crystal structure.

[Gamma knife for hypersecreting pituitary adenoma: analysis of 120 cases].

OBJECTIVE: To estimate the efficacy of Gamma knife radiosurgery (GKR) especially as a primary surgical treatment for hypersecreting pituitary adenoma. METHODS: One hundred and twenty cases with hypersecreting pituitary adenoma had been treated by Gamma knife radiosurgery. The clinical date had been analysed retrospectively. The tumor margin was covered by an isodose ranging from 45% to 70%. The margin dose was 15 to 32 Gy (mean 28.5 Gy) and the maximum dose varied from 35 to 70 Gy (mean 45.5 Gy). The total number of isocenter was 165 (mean 1-3). RESULTS: One hundred and eleven cases had been followed-up by hormone level, and 104 cases had been followed-up by image of MRI. The mean follow-up duration was 12-72 months (mean 36 months). The control rate of hormone level was 48.6%, the control rate of tumor growth was 96.2%, the incidence of hypopituitarism was in 2.7% and the incidence of tumor apoplexy was in 0.9% in followed-up cases. CONCLUSIONS: As a primary surgical treatment for hypersecreting pituitary adenoma, GKR can be effective and safe in controlling tumor growth and inducing hormonal normalization.