Signal Information Research Articles

Cosmological correlators with double massive exchanges: bootstrap equation and phenomenology

Using the recently developed cosmological bootstrap method, we compute the exact analytical solution for the seed integral appearing in cosmological correlators with double massive scalar exchanges. The result is explicit, valid in any kinematic configuration, and free from spurious divergences. It is applicable to any number of fields’ species with any masses. With an appropriate choice of variables, the results contain only single-layer summations. We also propose simple approximate formulas valid in different limits, enabling direct and instantaneous evaluation. Supported by exact numerical results using CosmoFlow, we explore the phenomenology of double massive exchange diagrams. Contrary to single-exchange diagrams with ubiquitous Lorentz-covariant interactions, the size of the cubic coupling constant can be large while respecting perturbativity bounds. Because of this property, the primordial bispectrum from double-exchange diagrams can be as large as, coincidentally, current observational constraints. In addition to being sizable on equilateral configurations, we show that the primordial bispectrum exhibits a large cosmological collider signal in the squeezed limit, making the double massive exchanges interesting channels for the detection of massive primordial fields. We propose to decisively disentangle double-exchange channels from single-exchange ones with cosmological observations by exploiting the phase information of the cosmological collider signal, the inflationary flavor oscillations from multiple fields’ species exchanges and the double soft limit in the primordial trispectrum.

Fiber optic magnetic field sensor based on a magnetic-fluid-induced phase-shift FLRD

A magnetic field sensing system based on a phase-shift fiber loop ring-down (FLRD) technique and multi-mode interferometer (MI) coated with magnetic fluid (MF) is proposed and demonstrated. The MI is constructed by splicing a segment of no-core fiber between two sections of single-mode fibers, which is then immersed in MF to serve as a sensing head with the advantages of simple fabrication and specific magnetic sensitivity. Due to the magnetic refractive index tunable properties of the MF, the magnetic-field-dependent loss will be introduced in the fiber loop by the sensing head. Such magnetic-induced loss would be accumulated during the round trip of the optical carrier and reflected on the phase information of the modulated signal. The phase-shift changes with the applied magnetic field strength, enabling magnetic field sensing through phase-shift measurements. The sensing system is experimentally demonstrated and a sensitivity of 0.704×10−3deg/Gs in the linear region is achieved. Moreover, the stability and repeatability of the system are verified, leading to a promising method for magnetic field measurements.

Exploring Nonlinear Dynamics In Brain Functionality Through Phase Portraits And Fuzzy Recurrence Plots.

Much of the complexity and diversity found in nature is driven by nonlinear phenomena, and this holds true for the brain. Nonlinear dynamics theory has been successfully utilized in explaining brain functions from a biophysics standpoint, and the field of statistical physics continues to make substantial progress in understanding brain connectivity and function. This study delves into complex brain functional connectivity using biophysical nonlinear dynamics approaches. We aim to uncover hidden information in high-dimensional and nonlinear neural signals, with the hope of providing a useful tool for analyzing information transitions in functionally complex networks. By utilizing phase portraits and fuzzy recurrence plots, we investigated the latent information in the functional connectivity of complex brain networks. Our numerical experiments, which include synthetic linear dynamics neural time series and a biophysically realistic neural mass model, showed that phase portraits and fuzzy recurrence plots are highly sensitive to changes in neural dynamics and can also be used to predict functional connectivity based on structural connectivity. Furthermore, the results showed that phase trajectories of neuronal activity encode low-dimensional dynamics, and the geometric properties of the limit-cycle attractor formed by the phase portraits can be used to explain the neurodynamics. Additionally, our results showed that the phase portrait and fuzzy recurrence plots can be used as functional connectivity descriptors, and both metrics were able to capture and explain nonlinear dynamics behavior during specific cognitive tasks. In conclusion, our findings suggest that phase portraits and fuzzy recurrence plots could be highly effective as functional connectivity descriptors, providing valuable insights into nonlinear dynamics in the brain.

Transformer-empowered receiver design of OFDM communication systems

With deep learning, we perform channel estimation and signal detection in massive Multiple Input Multiple Output (MIMO)-Orthogonal Frequency Division Multiplexing (OFDM) systems in this paper. Specifically, we design and extend the basic framework of receivers for MIMO-OFDM systems in an end-to-end approach. A Transformer-based MIMO-OFDM receiver called TCD-Receiver is proposed, which introduces a multi-attention mechanism to learn the channel characteristics by introducing a generic and flexible Transformer network structure. The network parameters are updated based on the relationship between the received signal and the original signal, where the final signal information is obtained without explicit channel estimation and the predicted transmit bits are directly output. The experimental results show that the TCD-Receiver proposed can effectively solve the channel distortion and detect the transmitted signals compared with the traditional communication receivers, and its performance can be comparable to that of the traditional OFDM receivers, and it also has obvious advantages in combating the complex and difficult-to-model channel environment as well as the nonlinear interference factors.

An autoencoder model based on one-dimensional neural network for epileptic EEG anomaly detection

We propose an autoencoder model based on a one-dimensional convolutional neural network (1DCNN) as the feature extraction network for efficient detection of epileptic EEG anomalies. The local information of normal EEG signals was captured by utilizing the local feature extraction ability of 1DCNN for training of an autoencoder to learn the expression of normal EEG data in low dimensional feature space. With the difference between the input and output as the anomaly score, the threshold was determined by the optimal equilibrium point of the ROC curve, and the EEG signals exceeding the threshold were diagnosed as the seizure data. The performance of the 1DCNN-AE epilepsy detection model was evaluated using the publicly available CHB-MIT scalp EEG dataset and TUH scalp EEG dataset. The AUC of the 1DCNN-AE model reached 0.890 of CHB-MIT and 0.686 of TUH under the average level of patients, and the epilepsy detection rate reached 0.974 and 0.893, and these results were better than the latest epilepsy anomaly detection models LSTM-VAE and GRU-VAE. The 1DCNN model had a parameter quantity of 58.5M, which was at the same level with LSTM-VAE (47.4 M) and GRU-VAE (36.9 M) but with much smaller FLOPs (0.377 G) than LSTM-VAE (21.6 G) and GRU-VAE (16.2 G). The autoencoder model based one-dimensional convolutional neural network can effectively detect abnormal EEG signals in epileptic seizure.

Multiplex graph aggregation and feature refinement for unsupervised incomplete multimodal emotion recognition

Multimodal Emotion Recognition (MER) involves integrating information of various modalities, including audio, visual, text and physiological signals, to comprehensively grasp human sentiments, which has emerged as a vibrant area within human–computer interaction. Researchers have developed many methods for this task, but many of these methods rely on labeled supervised learning and struggle to address the issue of missing some modalities of data. To address these issues, we propose a Multiplex Graph Aggregation and Feature Refinement framework for unsupervised incomplete MER, comprising four modules: Completion, Aggregation, Refinement, and Embedding. Specifically, we first capture the correlation information between samples using the graph structures, which aids in the completion of missing data and the multiplex aggregation of multimodal data. Then, we perform refinement operations on the aggregated features as well as alignment and enhancement operations on the embedding features to obtain the fused feature representations, which are consistent, highly separable and conducive to emotion recognition. Experimental results on multimodal emotion recognition datasets demonstrate that our method achieves state-of-the-art performance among unsupervised methods, validating its effectiveness.

Combined failure identification of intermediate bearing in aeroengine based on wavelet transform, second-order difference, and 1D-LBP

Intermediate bearing is one of key parts in double-rotor aeroengine, whose running state usually can only be monitored by casing signals. As the fault characteristic information detected from casing is weak and complex, it is generally difficult to be dug correctly. One-dimensional local binary pattern (1D-LBP) can depict failure information from the perspective of local feature extraction. Currently, the study on the application of 1D-LBP in fault identification is mostly based on original signals; due to the influence of component signals irrelevant to fault and noise, the fault information in original signals is weak and complex; besides, 1D-LBP is rather sensitive to noise, which is extremely possible to cause insufficient extraction of local textural features and the difficulty to identify a fault. To solve this problem and make it possible to precisely identify the fault of intermediate bearing of real aeroengine, the paper has proposed the combined method of wavelet transform (WT), second-order difference with 1D-LBP. WT is highly uncertain to determine the decomposition layer number. To solve the difficulty, correlation coefficient was introduced. Meanwhile, casing vibration signal was subjected to WT. Furthermore, because vibration signal often went with impact components in a bearing failure and second-order difference of signals was sensitive to impact feature, component signals obtained by WT were subjected to second-order difference operation. Additionally, taking advantage of 1D-LBP, second-order differences of signals were locally binarized with average value as criterion. Finally, 1D-LBP signals were re-transformed to decimal 1D local texture signals (1D-LTS). These 1D-LTS can embody local feature information of signals. According to the spectrum of 1D-LTS, combined failure type of intermediate bearing was identified. Through the comparison with WT and classical 1D-LBP method, and failure analysis after disassembly of aeroengine, the effectiveness and engineering applicability of proposed method have been verified.

Music source feature extraction based on improved attention mechanism and phase feature

Music source feature extraction is an important research direction in music information retrieval and music recommendation system. To extract the features of music sources more effectively, the study introduces the jump attention mechanism and combines it with the convolutional attention module. Also, a feature extraction module based on Unet + + and spatial attention module is proposed. In addition, the phase feature information of the mixed music signals is utilized to improve the network performance. Results showed that this model was studied to perform well in music source separation experiments of vocals and accompaniment. For vocal separation on the MIR-1K dataset, the model achieves 11.25 dB, 17.34 dB, and 13.83 dB for each metric, respectively. Meanwhile, for drum separation on the DSD100 dataset, the model achieves a median signal-to-source distortion ratio of 4.36 dB, which is 2.91 dB better than that of the Spectral Hierarchical Network model. For the separation of the bass sound and the human voice, the model's in the separation of bass and human voice, the median distortion ratio of the model is as high as 4.87 dB and 6.09 dB, which is better than that of the Spectral Hierarchical Network model. This indicates the significant performance advantages in feature extraction and separation of music sources, and it has important application values in music production and speech recognition.

Fault Intelligent Diagnosis for Distribution Box in Hot Rolling Based on Depthwise Separable Convolution and Bi-LSTM

The finishing mill is a critical link in the hot rolling process, influencing the final product’s quality, and even economic efficiency. The distribution box of the finishing mill plays a vital role in power transmission and distribution. However, harsh operating conditions can frequently lead to distribution box damage and even failure. To diagnose faults in the distribution box promptly, a fault diagnosis network model is constructed in this paper. This model combines depthwise separable convolution and Bi-LSTM. Depthwise separable convolution and Bi-LSTM can extract both spatial and temporal features from signals. This structure enables comprehensive feature extraction and fully utilizes signal information. To verify the diagnostic capability of the model, five types of data are collected and used: the pitting of tooth flank, flat-headed sleeve tooth crack, gear surface crack, gear tooth surface spalling, and normal conditions. The model achieves an accuracy of 97.46% and incorporates a lightweight design, which enhances computational efficiency. Furthermore, the model maintains approximately 90% accuracy under three noise conditions. Based on these results, the proposed model can effectively diagnose faults in the distribution box, and reduce downtime in engineering.

A tunable spatial coherence imaging with ultrasonic array for defects in composites

ABSTRACT Ultrasonic phased array detection technology is a promising non-destructive testing (NDT) method that has been widely used for defect detection. Due to the high attenuation characteristics of ultrasonic wave propagation in carbon fibre reinforced polymer (CFRP) and the directivity of the array elements, the possibility of false or missed detection for defects that are far from the array elements or located very close together will increase. In addition, the total focusing method (TFM) employs only the defect information contained in the full matrix capture (FMC) dataset while ignoring the spatial information of the array signals, leading to image mixed with artefact and noise. An improved method called p-weighted-TFM (p-WTFM) is proposed. First, this method scales the magnitude of time-delayed channel signal by p-th root while maintaining the phase unchanged. Second, channel summation with energy compensation and directional correction will be realised. Finally, the signal dimensionality is restored by p-th power. Experimental results show that the p-WTFM can use any rational p-value to flexibly adjust image quality. Specially, this method can accurately distinguish adjacent defects spaced 1 mm apart with p value of 2.5, and it can also detect two defects located 10 mm deep with p value of 3.0.

Efficient Second-Harmonic Generation in Adapted-Width Waveguides Based on Periodically Poled Thin-Film Lithium Niobate.

Frequency conversion process based on periodically poled thin-film lithium niobate (PPTFLN) has been widely recognized as an important component for quantum information and photonic signal processing. Benefiting from the tight confinement of optical modes, the normalized conversion efficiency (NCE) of nanophotonic waveguides is improved by orders of magnitude compared to their bulk counterparts. However, the power conversion efficiency of these devices is limited by inherent nanoscale inhomogeneity of thin-film lithium niobate (TFLN), leading to undesirable phase errors. In this paper, we theoretically present a novel approach to solve this problem. Based on dispersion engineering, we aim at adjusting the waveguide structure, making local waveguide width adjustment at positions of different thicknesses, thus eliminating the phase errors. The adapted waveguide width design is applied for etched and loaded waveguides based on PPTFLN, achieving the ultrahigh power conversion efficiency of second harmonic generation (SHG) up to 2.1 × 104%W-1 and 6936%W-1, respectively, which surpasses the power conversion efficiency of other related works. Our approach just needs standard periodic poling with a single period, significantly reducing the complexity of electrode fabrication and the difficulty of poling, and allows for the placing of multiple waveguides, without individual poling designs for each waveguide. With the advantages of simplicity, high production, and meeting current micro-nano fabrication technology, our work may open a new way for achieving highly efficient second-order nonlinear optical processes based on PPTFLN.

Investigating the simulation test and acoustic emission characteristics of structural-control type rockbursts in deep underground environments

Deep underground projects are in a complex in-situ stress environment, and rock burst disasters induced by the hard and brittle failure of the rock mass are prone to occur around the tunnel openings. In this study, we aim to investigate the impacts of geological weak surfaces in rock masses on rockbursts and the spatiotemporal evolution of microcracks during the development of rockbursts. To achieve this objective, two sets of cubic specimens containing a circular through-hole were designed. One set of specimens had prefabricated unfilled cracks around the openings. Subsequently, the rockburst development process within the deep tunnel was replicated by subjecting both groups of specimens to identical gradient loading using a true triaxial test system. During the test, a micro camera was used to record the failure of the borehole wall in real-time, and an acoustic emission monitoring system was utilized to capture the stress waves released during structural damage. Finally, a multi-level synergistic analysis of the rockburst damage mechanism was carried out based on the macro-imaging data and micro-acoustic emission data. The results show that failure in high-stress environments within tunnels mainly includes microcrack initiation, particle ejection, crack propagation, local cracking, slabbing spalling, damage penetration, and rockburst damage. The time–frequency domain information of acoustic emission signals is highly perceptive and representative of the structural damage state of the specimens. The mutation characteristics observed in time-domain parameters, such as acoustic emission amplitude, event rate, ringing, and cumulative energy, correspond to drastic alterations in the internal structure of the rock. The distribution characteristics of frequency-domain parameters, such as acoustic emission peak frequency, dominant frequency energy, and frequency centroid, reflect different crack scales and damage modes. The diffusion of frequency centroid indicates that the damage and failure patterns within the rock are evolving towards a more complex direction. The energy magnitude of acoustic emissions represents the damage intensity during the development of rockburst. The root causes of induced structural-control type rockbursts in deep hard brittle rock masses are the combined effects of localized high-stress concentrations and large-scale discontinuous geological structures, such as natural joints, fissures, and structural planes. The naturally occurring large-scale through-type geological weak surfaces within the rock mass reduce the strength of the surrounding rock, alter the location of stress concentration, and change the initial damage characteristics. Moreover, they promote the evolution of rockbursts, thereby increasing their destructive intensity.

A Dual-Stream Deep Learning-Based Acoustic Denoising Model to Enhance Underwater Information Perception

Estimating the line spectra of ship-radiated noise is a crucial remote sensing technique for detecting and recognizing underwater acoustic targets. Improving the signal-to-noise ratio (SNR) makes the low-frequency components of the target signal more prominent. This enhancement aids in the detection of underwater acoustic signals using sonar. Based on the characteristics of low-frequency narrow-band line spectra signals in underwater target radiated noise, we propose a dual-stream deep learning network with frequency characteristics transformation (DS_FCTNet) for line spectra estimation. The dual streams predict amplitude and phase masks separately and use an information exchange module to swap learn features between the amplitude and phase spectra, aiding in better phase information reconstruction and signal denoising. Additionally, a frequency characteristics transformation module is employed to extract convolutional features between channels, obtaining global correlations of the amplitude spectrum and enhancing the ability to learn target signal features. Through experimental analysis on ShipsEar, a dataset of underwater acoustic signals by hydrophones deployed in shallow water, the effectiveness and rationality of different modules within DS_FCTNet are verified.Under low SNR conditions and with unknown ship types, the proposed DS_FCTNet model exhibits the best line spectrum enhancement compared to methods such as SEGAN and DPT_FSNet. Specifically, SDR and SSNR are improved by 14.77 dB and 13.58 dB, respectively, enabling the detection of weaker target signals and laying the foundation for target localization and recognition applications.

Variability of morphology in photoplethysmographic waveform quantified with unsupervised wave-shape manifold learning for clinical assessment

Objective.We investigated fluctuations of the photoplethysmography (PPG) waveform in patients undergoing surgery. There is an association between the morphologic variation extracted from arterial blood pressure (ABP) signals and short-term surgical outcomes. The underlying physiology could be the numerous regulatory mechanisms on the cardiovascular system. We hypothesized that similar information might exist in PPG waveform. However, due to the principles of light absorption, the noninvasive PPG signals are more susceptible to artifacts and necessitate meticulous signal processing.Approach.Employing the unsupervised manifold learning algorithm, dynamic diffusion map, we quantified multivariate waveform morphological variations from the PPG continuous waveform signal. Additionally, we developed several data analysis techniques to mitigate PPG signal artifacts to enhance performance and subsequently validated them using real-life clinical database.Main results.Our findings show similar associations between PPG waveform during surgery and short-term surgical outcomes, consistent with the observations from ABP waveform analysis.Significance.The variation of morphology information in the PPG waveform signal in major surgery provides clinical meanings, which may offer new opportunity of PPG waveform in a wider range of biomedical applications, due to its non-invasive nature.

Motor fault diagnosis based on multisensor-driven visual information fusion

To need of accurate motor fault diagnosis in industrial system, we propose a fault diagnosis framework that utilizes motor current and electromagnetic signals, combining them with a self-attention-enhanced capsule network for enhanced signal analysis and accuracy. Firstly, the original signal extracted by multiple sensors is constructed into a symmetric point mode (SDP) image, and the visual fault information of different sensors and fusion signals of different motion health states are obtained by the proposed multi-channel image fusion method. Then, the capsule network, combined with self-attention, extracts spatial features from the high-dimensional tensor of the multi-channel fused image for adaptive recognition and extraction. Subsequently, advanced feature vector information is obtained through softmax for diagnosis. Diagnosis results of several datasets indicate that the developed diagnosis framework with compressed image information can availably identify 8 kinds of motor fault states under various loads, and the fault diagnosis rate is as high as 99.95 %, it is helpful for low cost and high-speed diagnosis of motors. In addition, by learning multiple sensor signals in the same state, it obtains stronger robustness and effectiveness than a single signal model.

A Method of Ship Target Modulation Feature Enhancement Based on Phase Information

Abstract Ship modulation features are commonly used in underwater acoustic target recognition, which contains propeller parameter information of the target. The basis processing of modulation spectrum is demodulation. Traditional envelope demodulation methods only consider amplitude information and make little use of phase information, which can fully characterize the dynamic characteristics of radiated noise. In this paper, a modulation feature enhancement method based on phase information is proposed, which takes full account of the phase information of the target signal. It uses the modified group delay function to demodulate the target signal, and utilizes the phase stability characteristics of the modulated line spectrum to enhance the modulation features of the ship target by weighted phase features. Simulation and sea trial data verification show that the proposed method can effectively enhance the energy of the line spectrum, suppress the background noise, and improve the signal-to-noise ratio of the modulation line spectrum, which lays a good foundation for further using the modulation line spectrum to carry out the estimation of modulation parameters, and has a good application prospect in the field of underwater acoustic target recognition.

Lightweight Network Bearing Intelligent Fault Diagnosis Based on VMD-FK-ShuffleNetV2

With the increasing complexity of mechanical equipment and diversification of deep learning models, vibration signals collected from such equipment are susceptible to noise interference. Moreover, traditional neural network models struggle to be effectively deployed in production environments with limited computational resources, severely impacting the accurate extraction and effective diagnosis of FK fault characteristics. In response to this challenge, this study proposes a fault diagnosis method for rolling bearings, integrating a lightweight ShuffleNetV2 network with variational mode decomposition (VMD) and the fast kurtogram (FK) algorithm. Initially, this paper introduces an enhanced FK method where the VMD algorithm is employed for data denoising, extracting FK post-denoising. These feature maps not only preserve critical signal information but also simplify data complexity. Subsequently, these feature maps are utilized to train and test the ShuffleNetV2 model, facilitating effective fault identification and classification. Ultimately, by conducting experimental comparisons with several mainstream lightweight network models, such as MobileNet and SqueezeNet, as well as traditional convolutional neural network models, this study validates the effectiveness of the proposed method in extracting fault characteristics from vibration signals, demonstrating superior diagnostic accuracy and computational efficiency. This provides a novel technical approach for health monitoring and fault diagnosis of industrial bearings and offers theoretical and experimental support for the deployment of lightweight networks in industrial applications.

A Novel TCN-LSTM Hybrid Model for sEMG-Based Continuous Estimation of Wrist Joint Angles.

Surface electromyography (sEMG) offers a novel method in human-machine interactions (HMIs) since it is a distinct physiological electrical signal that conceals human movement intention and muscle information. Unfortunately, the nonlinear and non-smooth features of sEMG signals often make joint angle estimation difficult. This paper proposes a joint angle prediction model for the continuous estimation of wrist motion angle changes based on sEMG signals. The proposed model combines a temporal convolutional network (TCN) with a long short-term memory (LSTM) network, where the TCN can sense local information and mine the deeper information of the sEMG signals, while LSTM, with its excellent temporal memory capability, can make up for the lack of the ability of the TCN to capture the long-term dependence of the sEMG signals, resulting in a better prediction. We validated the proposed method in the publicly available Ninapro DB1 dataset by selecting the first eight subjects and picking three types of wrist-dependent movements: wrist flexion (WF), wrist ulnar deviation (WUD), and wrist extension and closed hand (WECH). Finally, the proposed TCN-LSTM model was compared with the TCN and LSTM models. The proposed TCN-LSTM outperformed the TCN and LSTM models in terms of the root mean square error (RMSE) and average coefficient of determination (R2). The TCN-LSTM model achieved an average RMSE of 0.064, representing a 41% reduction compared to the TCN model and a 52% reduction compared to the LSTM model. The TCN-LSTM also achieved an average R2 of 0.93, indicating an 11% improvement over the TCN model and an 18% improvement over the LSTM model.

Hybrid quantum key distribution: A novel approach for secure end-to-end communication in network infrastructure

The Quantum Key Distribution (QKD) network approach has emerged as a promising solution and has garnered significant interest across various domains. QKD networks can be broadly categorized into two main types: those based on measurement-device-dependent (MDD) protocols and those based on measurement-device-independent (MDI) protocols. A novel network scheme is proposed combining these two protocols to guarantee end-to-end secure communication. It comprises two interconnected networks — the first uses a measurement-device-dependent protocol to connect routers with nearby nodes, while the second employs a measurement-device-independent protocol to interconnect the routers. In the first network, the process begins with Alice (Router) generating two types of weak quantum signals — the first has intensity, frequency and polarization or phase randomly varied, while the second with different delay times. Alice randomly transmits either signal states carrying information or decoy states with varying properties to the receivers (Bobs) over the quantum channels. She also sends a synchronization signal along with the quantum signals. At the receiver end, the synchronization pulse timestamps the incoming quantum signals. Each Bob then randomly shifts the frequencies of the received signals and measures their phase/polarization of the first signal and measures the delay times of the second signal. The secret key is derived by combining the frequency, the delay time and the phase/polarization information of the quantum signals, resulting in longer key lengths. Encoding information across multiple degrees of freedom enables extended key generation for secure communication. However, between the routers, MDI-QKD protocols are used to securely share the keys. This hybrid approach aims to ensure secure end-to-end communication across the network.

Ultrasonic Assessment of Liver Fibrosis Using One-Dimensional Convolutional Neural Networks Based on Frequency Spectra of Radiofrequency Signals with Deep Learning Segmentation of Liver Regions in B-Mode Images: A Feasibility Study.

The early detection of liver fibrosis is of significant importance. Deep learning analysis of ultrasound backscattered radiofrequency (RF) signals is emerging for tissue characterization as the RF signals carry abundant information related to tissue microstructures. However, the existing methods only used the time-domain information of the RF signals for liver fibrosis assessment, and the liver region of interest (ROI) is outlined manually. In this study, we proposed an approach for liver fibrosis assessment using deep learning models on ultrasound RF signals. The proposed method consisted of two-dimensional (2D) convolutional neural networks (CNNs) for automatic liver ROI segmentation from reconstructed B-mode ultrasound images and one-dimensional (1D) CNNs for liver fibrosis stage classification based on the frequency spectra (amplitude, phase, and power) of the segmented ROI signals. The Fourier transform was used to obtain the three kinds of frequency spectra. Two classical 2D CNNs were employed for liver ROI segmentation: U-Net and Attention U-Net. ROI spectrum signals were normalized and augmented using a sliding window technique. Ultrasound RF signals collected (with a 3-MHz transducer) from 613 participants (Group A) were included for liver ROI segmentation and those from 237 participants (Group B) for liver fibrosis stage classification, with a liver biopsy as the reference standard (Fibrosis stage: F0 = 27, F1 = 49, F2 = 51, F3 = 49, F4 = 61). In the test set of Group A, U-Net and Attention U-Net yielded Dice similarity coefficients of 95.05% and 94.68%, respectively. In the test set of Group B, the 1D CNN performed the best when using ROI phase spectrum signals to evaluate liver fibrosis stages ≥F1 (area under the receive operating characteristic curve, AUC: 0.957; accuracy: 89.19%; sensitivity: 85.17%; specificity: 93.75%), ≥F2 (AUC: 0.808; accuracy: 83.34%; sensitivity: 87.50%; specificity: 78.57%), and ≥F4 (AUC: 0.876; accuracy: 85.71%; sensitivity: 77.78%; specificity: 94.12%), and when using the power spectrum signals to evaluate ≥F3 (AUC: 0.729; accuracy: 77.14%; sensitivity: 77.27%; specificity: 76.92%). The experimental results demonstrated the feasibility of both the 2D and 1D CNNs in liver parenchyma detection and liver fibrosis characterization. The proposed methods have provided a new strategy for liver fibrosis assessment based on ultrasound RF signals, especially for early fibrosis detection. The findings of this study shed light on deep learning analysis of ultrasound RF signals in the frequency domain with automatic ROI segmentation.