Diffusion tensor imaging technology can provide information on the white matter of the brain, which can be used to explore changes in brain tissue structure, but it lacks the specific description of the microstructure information of brain tissue. The neurite orientation dispersion and density imaging make up for its shortcomings. But in order to accurately estimate the brain microstructure, a large number of diffusion gradients are needed, and the calculation is complex and time-consuming through maximum likelihood fitting. Therefore, this paper proposes a kind of microstructure parameters estimation method based on the proximal gradient network, which further avoids the classic fitting paradigm. The method can accurately estimate the parameters while reducing the number of diffusion gradients, and achieve the purpose of imaging quality better than the neurite orientation dispersion and density imaging model and accelerated microstructure imaging via convex optimization model.
A fitting method of calculating local helix parameters of proteins based on dual quaternions registration fitting (DQRFit) is proposed in this paper. First, the C and N atom coordinates of each residue in the protein structure data are extracted. Then the unregistered data and reference data are constructed using the sliding windows. The square sum of the distance of the data points before and after registration is regarded as an optimization goal. We calculate the optimal rotation matrix and the translation vector using the dual quaternion registration algorithm, and get the helix parameters of the secondary structure which contain the number of residues per turn(τ), helix radius(ρ)and helix pitch(p). Furthermore, we can achieve the fitting of three-helix parameters of τ, ρ, p simultaneously with the dual quaternion registration, and can adjust the sliding windows to adapt to different error levels. Compared with the traditional helix fitting method, DQRFit has some advantages such as low computational complexity, strong anti-interference, and high fitting accuracy. It is proven that the precision of proposed DQRFit for α helix detection is comparable to that of the dictionary of secondary structure of proteins (DSSP), and is better than that of other traditional methods. This is of great significance for the protein structure classification and functional prediction, drug design, protein structure visualization and other fields in the future.
The recurrent neural network architecture improves the processing ability of time-series data. However, issues such as exploding gradients and poor feature extraction limit its application in the automatic diagnosis of mild cognitive impairment (MCI). This paper proposed a research approach for building an MCI diagnostic model using a Bayesian-optimized bidirectional long short-term memory network (BO-BiLSTM) to address this problem. The diagnostic model was based on a Bayesian algorithm and combined prior distribution and posterior probability results to optimize the BO-BiLSTM network hyperparameters. It also used multiple feature quantities that fully reflected the cognitive state of the MCI brain, such as power spectral density, fuzzy entropy, and multifractal spectrum, as the input of the diagnostic model to achieve automatic MCI diagnosis. The results showed that the feature-fused Bayesian-optimized BiLSTM network model achieved an MCI diagnostic accuracy of 98.64% and effectively completed the diagnostic assessment of MCI. In conclusion, based on this optimization, the long short-term neural network model has achieved automatic diagnostic assessment of MCI, providing a new diagnostic model for intelligent diagnosis of MCI.
The electroencephalogram (EEG) signal is a general reflection of the neurophysiological activity of the brain, which has the advantages of being safe, efficient, real-time and dynamic. With the development and advancement of machine learning research, automatic diagnosis of Alzheimer’s diseases based on deep learning is becoming a research hotspot. Started from feedforward neural networks, this paper compared and analysed the structural properties of neural network models such as recurrent neural networks, convolutional neural networks and deep belief networks and their performance in the diagnosis of Alzheimer’s disease. It also discussed the possible challenges and research trends of this research in the future, expecting to provide a valuable reference for the clinical application of neural networks in the EEG diagnosis of Alzheimer’s disease.