scholarly journals Enhancing the signal-to-noise ratio and generating contrast for cryo-EM images with convolutional neural networks

IUCrJ ◽  
2020 ◽  
Vol 7 (6) ◽  
pp. 1142-1150
Author(s):  
Eugene Palovcak ◽  
Daniel Asarnow ◽  
Melody G. Campbell ◽  
Zanlin Yu ◽  
Yifan Cheng
Keyword(s):  
Neural Networks ◽  
Image Processing ◽  
Convolutional Neural Networks ◽  
Signal To Noise Ratio ◽  
Image Contrast ◽  
Signal To Noise ◽  
Pass Filter ◽  
Processing Pipeline ◽  
Low Pass ◽  
Noise Ratio

In cryogenic electron microscopy (cryo-EM) of radiation-sensitive biological samples, both the signal-to-noise ratio (SNR) and the contrast of images are critically important in the image-processing pipeline. Classic methods improve low-frequency image contrast experimentally, by imaging with high defocus, or computationally, by applying various types of low-pass filter. These contrast improvements typically come at the expense of the high-frequency SNR, which is suppressed by high-defocus imaging and removed by low-pass filtration. Recently, convolutional neural networks (CNNs) trained to denoise cryo-EM images have produced impressive gains in image contrast, but it is not clear how these algorithms affect the information content of the image. Here, a denoising CNN for cryo-EM images was implemented and a quantitative evaluation of SNR enhancement, induced bias and the effects of denoising on image processing and three-dimensional reconstructions was performed. The study suggests that besides improving the visual contrast of cryo-EM images, the enhanced SNR of denoised images may be used in other parts of the image-processing pipeline, such as classification and 3D alignment. These results lay the groundwork for the use of denoising CNNs in the cryo-EM image-processing pipeline beyond particle picking.

scholarly journals Enhancing SNR and generating contrast for cryo-EM images with convolutional neural networks

2020 ◽  
Author(s):  
Eugene Palovcak ◽  
Daniel Asarnow ◽  
Melody G. Campbell ◽  
Zanlin Yu ◽  
Yifan Cheng
Keyword(s):  
Image Processing ◽  
High Frequency ◽  
Signal To Noise Ratio ◽  
Three Dimensional ◽  
Low Frequency ◽  
Pass Filter ◽  
Low Pass Filter ◽  
Processing Pipeline ◽  
Visual Contrast ◽  
Low Pass

AbstractIn cryogenic electron microscopy (cryo-EM) of radiation-sensitive biological samples, both the signal-to-noise ratio (SNR) and the contrast of images are critically important in the image processing pipeline. Classic methods improve low-frequency image contrast experimentally, by imaging with high defocus, or computationally, by applying various types of low-pass filter. These contrast improvements typically come at the expense of high-frequency SNR, which is suppressed by high-defocus imaging and removed by low pass filtration. Here, we demonstrate that a convolutional neural network (CNN) denoising algorithm can be used to significantly enhance SNR and generate contrast in cryo-EM images. We provide a quantitative evaluation of bias introduced by the denoising procedure and its influences on image processing and three-dimensional reconstructions. Our study suggests that besides enhancing the visual contrast of cryo-EM images, the enhanced SNR of denoised images may facilitate better outcomes in the other parts of the image processing pipeline, such as classification and 3D alignment. Overall, our results provide a ground of using denoising CNNs in the cryo-EM image processing pipeline.

scholarly journals Cooperative Spectrum Sensing Based on Convolutional Neural Networks

2021 ◽  
Vol 11 (10) ◽  
pp. 4440
Author(s):  
Youheng Tan ◽  
Xiaojun Jing
Keyword(s):  
Neural Networks ◽  
Spectrum Sensing ◽  
Convolutional Neural Networks ◽  
Cooperative Spectrum Sensing ◽  
Signal To Noise Ratio ◽  
Majority Voting ◽  
Signal To Noise ◽  
Hidden Terminal ◽  
Simulation Results ◽  
Sensing Performance

Cooperative spectrum sensing (CSS) is an important topic due to its capacity to solve the issue of the hidden terminal. However, the sensing performance of CSS is still poor, especially in low signal-to-noise ratio (SNR) situations. In this paper, convolutional neural networks (CNN) are considered to extract the features of the observed signal and, as a consequence, improve the sensing performance. More specifically, a novel two-dimensional dataset of the received signal is established and three classical CNN (LeNet, AlexNet and VGG-16)-based CSS schemes are trained and analyzed on the proposed dataset. In addition, sensing performance comparisons are made between the proposed CNN-based CSS schemes and the AND, OR, majority voting-based CSS schemes. The simulation results state that the sensing accuracy of the proposed schemes is greatly improved and the network depth helps with this.

Retinal Digital Image Quality Improvement as A Diabetes Retinopatic Disease Detection Effort

2020 ◽  
Vol 4 (2) ◽  
pp. 53-60
Author(s):  
Latifah Listyalina ◽  
Yudianingsih Yudianingsih ◽  
Dhimas Arief Dharmawan
Keyword(s):  
Image Processing ◽  
Diabetic Retinopathy ◽  
Image Quality ◽  
Digital Image ◽  
Signal To Noise Ratio ◽  
Histogram Equalization ◽  
Vital Role ◽  
Retinal Images ◽  
Signal To Noise ◽  
Noise Ratio

Image processing is a technical term useful for modifying images in various ways. In medicine, image processing has a vital role. One example of images in the medical world, namely retinal images, can be obtained from a fundus camera. The retina image is useful in the detection of diabetic retinopathy. In general, direct observation of diabetic retinopathy is conducted by a doctor on the retinal image. The weakness of this method is the slow handling of the disease. For this reason, a computer system is required to help doctors detect diabetes retinopathy quickly and accurately. This system involves a series of digital image processing techniques that can process retinal images into good quality images. In this research, a method to improve the quality of retinal images was designed by comparing the methods for adjusting histogram equalization, contrast stretching, and increasing brightness. The performance of the three methods was evaluated using Mean Square Error (MSE), Peak Signal to Noise Ratio (PSNR), and Signal to Noise Ratio (SNR). Low MSE values and high PSNR and SNR values indicated that the image had good quality. The results of the study revealed that the image was the best to use, as evidenced by the lowest MSE values and the highest SNR and PSNR values compared to other techniques. It indicated that adaptive histogram equalization techniques could improve image quality while maintaining its information.

Analysis of neural networks efficiency when accounting for weight coefficients jitter, leading to reduction in output signal/noise ratio

2021 ◽  
Author(s):  
S.V. Zimina
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Artificial Neural Network ◽  
Output Signal ◽  
Signal To Noise Ratio ◽  
Signal To Noise ◽  
Useful Signal ◽  
Artificial Neural ◽  
Noise Ratio ◽  
Weight Coefficients

Setting up artificial neural networks using iterative algorithms is accompanied by fluctuations in weight coefficients. When an artificial neural network solves the problem of allocating a useful signal against the background of interference, fluctuations in the weight vector lead to a deterioration of the useful signal allocated by the network and, in particular, losses in the output signal-to-noise ratio. The goal of the research is to perform a statistical analysis of an artificial neural network, that includes analysis of losses in the output signal-to-noise ratio associated with fluctuations in the weight coefficients of an artificial neural network. We considered artificial neural networks that are configured using discrete gradient, fast recurrent algorithms with restrictions, and the Hebb algorithm. It is shown that fluctuations lead to losses in the output signal/noise ratio, the level of which depends on the type of algorithm under consideration and the speed of setting up an artificial neural network. Taking into account the fluctuations of the weight vector in the analysis of the output signal-to-noise ratio allows us to correlate the permissible level of loss in the output signal-to-noise ratio and the speed of network configuration corresponding to this level when working with an artificial neural network.

scholarly journals NOISE STABILITY OF SIGNAL RECEPTION ALGORITHMS WITH MULTIPULSE PULSE-POSITION MODULATION

2018 ◽  
Vol 42 (1) ◽  
pp. 167-174 ◽  
Author(s):  
V. I. Parfenov ◽  
D. Y. Golovanov
Keyword(s):  
Communication Systems ◽  
Pulse Sequence ◽  
Signal To Noise Ratio ◽  
Point Of View ◽  
Estimation Accuracy ◽  
Pulse Position Modulation ◽  
Signal To Noise ◽  
Pass Filter ◽  
Low Pass Filter ◽  
Noise Ratio

An algorithm for estimating time positions and amplitudes of a periodic pulse sequence from a small number of samples was proposed. The number of these samples was determined only by the number of pulses. The performance of this algorithm was considered on the assumption that the spectrum of the original signal is limited with an ideal low-pass filter or the Nyquist filter, and conditions for the conversion from one filter to the other were determined. The efficiency of the proposed algorithm was investigated through analyzing in which way the dispersion of estimates of time positions and amplitudes depends on the signal-to-noise ratio and on the number of pulses in the sequence. It was shown that, from this point of view, the efficiency of the algorithm decreases with increasing number of sequence pulses. Besides, the efficiency of the proposed algorithm decreases with decreasing signal-to-noise ratio.It was found that, unlike the classical maximum likelihood algorithm, the proposed algorithm does not require a search for the maximum of a multivariable function, meanwhile characteristics of the estimates are practically the same for both these methods. Also, it was shown that the estimation accuracy of the proposed algorithm can be increased by an insignificant increase in the number of signal samples.The results obtained may be used in the practical design of laser communication systems, in which the multipulse pulse-position modulation is used for message transmission. 

Auditory Frequency Selectivity in Normal and Presbycusic Subjects

1980 ◽  
Vol 23 (3) ◽  
pp. 603-613 ◽  
Author(s):  
Robert H. Margolis ◽  
Seth M. Goldberg
Keyword(s):  
Signal To Noise Ratio ◽  
Cutoff Frequency ◽  
Frequency Selectivity ◽  
Hearing Impaired ◽  
Critical Ratio ◽  
Signal To Noise ◽  
Noise Masker ◽  
Low Pass ◽  
Noise Ratio ◽  
Stable Estimate

Auditory frequency selectivity was inferred from measurements of the detectability of tonal signals as a function of the cutoff frequency of a low-pass computer-generated noise masker. In Experiment I the effect of small changes in signal-to-noise ratio on inferred auditory frequency selectivity was studied. In Experiment II, frequency selectivity was determined for five normal-hearing subjects and four subjects with sensorineural hearing loss due to presbycusis. Critical ratios (signal-to-noise ratio at masked threshold) also were determined in Experiment II. The results of Experiment I indicate that the low-pass masking experiment provides a stable estimate of the width, but not the position, of the critical masking band. Experiment II revealed elevated critical ratios for three of the four presbycusic subjects. Some hearing-impaired subjects appeared to have normal frequency selectivity despite elevated critical ratios. Other presbycusic subjects demonstrated impaired auditory frequency selectivity. The results suggest that critical ratio and critical masking band data are free to vary independently in hearing-impaired subjects.

Sign in / Sign up

Export Citation Format

Share Document