Search

Purpose: To introduce a novel deep model‐based architecture (DMBA), SPICER, that uses pairs of noisy and undersampled k‐space measurements of the same object to jointly train a model for MRI reconstruction and automatic coil sensitivity estimation. Methods: SPICER consists of two modules to simultaneously reconstructs accurate MR images and estimates high‐quality coil sensitivity maps (CSMs). The first module, CSM estimation module, uses a convolutional neural network (CNN) to estimate CSMs from the raw measurements. The second module, DMBA‐based MRI reconstruction module, forms reconstructed images from the input measurements and the estimated CSMs using both the physical measurement model and learned CNN prior. With the benefit of our self‐supervised learning strategy, SPICER can be efficiently trained without any fully sampled reference data. Results: We validate SPICER on both open‐access datasets and experimentally collected data, showing that it can achieve state‐of‐the‐art performance in highly accelerated data acquisition settings (up to 10×$$ 10\times $$). Our results also highlight the importance of different modules of SPICER—including the DMBA, the CSM estimation, and the SPICER training loss—on the final performance of the method. Moreover, SPICER can estimate better CSMs than pre‐estimation methods especially when the ACS data is limited. Conclusion: Despite being trained on noisy undersampled data, SPICER can reconstruct high‐quality images and CSMs in highly undersampled settings, which outperforms other self‐supervised learning methods and matches the performance of the well‐known E2E‐VarNet trained on fully sampled ground‐truth data. [ABSTRACT FROM AUTHOR]

Purpose: We aimed to incorporate a deep learning prior with k‐space data fidelity for accelerating hyperpolarized carbon‐13 MRSI, demonstrated on synthetic cancer datasets. Methods: A two‐site exchange model, derived from the Bloch equation of MR signal evolution, was firstly used in simulating training and testing data, that is, synthetic phantom datasets. Five singular maps generated from each simulated dataset were used to train a deep learning prior, which was then employed with the fidelity term to reconstruct the undersampled MRI k‐space data. The proposed method was assessed on synthetic human brain tumor images (N = 33), prostate cancer images (N = 72), and mouse tumor images (N = 58) for three undersampling factors and 2.5% additive Gaussian noise. Furthermore, varied levels of Gaussian noise with SDs of 2.5%, 5%, and 10% were added on synthetic prostate cancer data, and corresponding reconstruction results were evaluated. Results: For quantitative evaluation, peak SNRs were approximately 32 dB, and the accuracy was generally improved for 5 to 8 dB compared with those from compressed sensing with L1‐norm regularization or total variation regularization. Reasonable normalized RMS error were obtained. Our method also worked robustly against noise, even on a data with noise SD of 10%. Conclusion: The proposed singular value decomposition + iterative deep learning model could be considered as a general framework that extended the application of deep learning MRI reconstruction to metabolic imaging. The morphology of tumors and metabolic images could be measured robustly in six times acceleration using our method. [ABSTRACT FROM AUTHOR]

Speaking skills generally receive little attention in traditional English as a Foreign Language (EFL) classrooms, and this is especially the case in secondary education in Indonesia. A vocabulary deficit and poor pronunciation skills hinder learners in their efforts to improve speaking proficiency. In the present study, we investigated the effects of using two language learning websites, I Love Indonesia (ILI) and NovoLearning (NOVO). These websites are equipped with Automatic Speech Recognition (ASR) technology, with each website providing different types of immediate feedback. We measured written receptive and productive vocabulary knowledge of 40 target words before and after the intervention in which 146 students practiced with these two ASR-based websites, and compared it to that of a control group (n = 86). The ASR-based websites successfully helped students enhance their receptive vocabulary. Twenty-four students participated in a spoken pre-and post-test pronouncing the same 40 target words. We successfully developed an approach to measure pronunciation skills which showed that the treatment groups outperformed the control group. Our results indicate that our technology is successful in improving vocabulary and pronunciation skills. [ABSTRACT FROM AUTHOR]

End-to-end (E2E) automatic speech recognition (ASR) models, which consist of deep learning models, are able to perform ASR tasks using a single neural network. These models should be trained using a large amount of data; however, collecting speech data which matches the targeted speech domain can be difficult, so speech data is often used that is not an exact match to the target domain, resulting in lower performance. In comparison to speech data, in-domain text data is much easier to obtain. Thus, traditional ASR systems use separately trained language models and HMM-based acoustic models. However, it is difficult to separate language information from an E2E ASR model because the model learns both acoustic and language information in an integrated manner, making it very difficult to create E2E ASR models for specialized target domain which are able to achieve sufficient recognition performance at a reasonable cost. In this paper, we propose a method of replacing the language information within pre-trained E2E ASR models in order to achieve adaptation to a target domain. This is achieved by deleting the "implicit" language information contained within the ASR model by subtracting the source-domain language model trained with a transcription of the ASR's training data in a logarithmic domain. We then integrate a target domain language model through addition in the logarithmic domain. This subtraction and addition to replace of the language model is based on Bayes' theorem. In our experiment, we first used two datasets of the Corpus of Spontaneous Japanese (CSJ) to evaluate the effectiveness of our method. We then we evaluated our method using the Japanese Newspaper Article Speech (JNAS) and CSJ corpora, which contain audio data from the read speech and spontaneous speech domain, respectively, to test the effectiveness of our proposed method at bridging the gap between these two language domains. Our results show that our proposed language model replacement method achieved better ASR performance than both non-adapted (baseline) ASR models and ASR models adapted using the conventional Shallow Fusion method. [ABSTRACT FROM AUTHOR]

Detecting double compressed (DC) speech signals is an important audio forensics task since it is highly related to the integrity and the authenticity of the recording. Adaptive multi-rate (AMR) speech codec is a popular audio compression technique specifically optimized for speech signals and it is a standard audio recording format in the vast majority of the smart phones. All of the previous studies addressing the detection of DC AMR signals report their findings for the speech signals compressed using the narrowband AMR codec (AMR-NB). Meanwhile, wideband AMR codec (AMR-WB) has been used by several mobile phone manufacturers, but DC AMR-WB speech signal detection performance remains unknown. To the best of our knowledge, this is the first study focusing on detecting the DC signals compressed using the AMR-WB speech codec. To this end, we propose three different deep neural network-based DC AMR-WB signal detection systems where the spectrogram representations of the speech signals are used as the input features. Experimental results conducted on TIMIT database provide several important findings regarding the DC AMR-WB speech detection. Firstly, DC AMR-WB detection is found to be a more challenging task than detecting the AMR-NB signals. For example, convolutional neural network (CNN)-based system yields 74.83% and 99.93% detection rates on AMR-WB and AMR-NB coded signals, respectively. Secondly, capturing the temporal information using long short-term memory (LSTM) network with the DC AMR-WB signal detection accuracy of 86.25% is found to be superior to the CNN system. Thirdly, combining the deep feature representations learned by CNN and LSTM networks further improves the performance. Fourthly, the detection rates are found to deteriorate when the signals are first encoded using different audio codecs prior to AMR-WB compression. Finally, applying score level or decision level fusion to the proposed three systems improves the detection rates, in general. [ABSTRACT FROM AUTHOR]

In speech coding, denoising of the speech signal is essential as well as crucial. The filters for minimizing errors through denoising employ the autoregressive moving average (ARMA) approach, introducing higher computational complexity in speech coder design. This research work presents the design and implementation of an effective perceptual weighting filter (PWF) for speech coding. The high-level synthesis of the fixed-point PWF filter is optimized by multiple optimization techniques along with detailed design space exploration using the weighted sum (WS) method. To enhance the performance, an FPGA-based hardware accelerator is proposed using hardware/software (HW/SW) co-design in an embedded environment. Simulative analysis in Vivado HLS and final accelerator design in the Vitis IDE tool validate the proposed architecture by using real-time speech samples, demonstrating a 50% reduction in area and a 99% execution improvement. This makes it well-suited for use in modern speech codecs, enhancing the efficiency. [ABSTRACT FROM AUTHOR]

Objectives: To quantify regional manifestations related to COPD as anomalies from a modeled distribution of normal-appearing lung on chest CT using a deep learning (DL) approach, and to assess its potential to predict disease severity. Materials and methods: Paired inspiratory/expiratory CT and clinical data from COPDGene and COSYCONET cohort studies were included. COPDGene data served as training/validation/test data sets (N = 3144/786/1310) and COSYCONET as external test set (N = 446). To differentiate low-risk (healthy/minimal disease, [GOLD 0]) from COPD patients (GOLD 1–4), the self-supervised DL model learned semantic information from 50 × 50 × 50 voxel samples from segmented intact lungs. An anomaly detection approach was trained to quantify lung abnormalities related to COPD, as regional deviations. Four supervised DL models were run for comparison. The clinical and radiological predictive power of the proposed anomaly score was assessed using linear mixed effects models (LMM). Results: The proposed approach achieved an area under the curve of 84.3 ± 0.3 (p < 0.001) for COPDGene and 76.3 ± 0.6 (p < 0.001) for COSYCONET, outperforming supervised models even when including only inspiratory CT. Anomaly scores significantly improved fitting of LMM for predicting lung function, health status, and quantitative CT features (emphysema/air trapping; p < 0.001). Higher anomaly scores were significantly associated with exacerbations for both cohorts (p < 0.001) and greater dyspnea scores for COPDGene (p < 0.001). Conclusion: Quantifying heterogeneous COPD manifestations as anomaly offers advantages over supervised methods and was found to be predictive for lung function impairment and morphology deterioration. Clinical relevance statement: Using deep learning, lung manifestations of COPD can be identified as deviations from normal-appearing chest CT and attributed an anomaly score which is consistent with decreased pulmonary function, emphysema, and air trapping. Key Points: • A self-supervised DL anomaly detection method discriminated low-risk individuals and COPD subjects, outperforming classic DL methods on two datasets (COPDGene AUC = 84.3%, COSYCONET AUC = 76.3%). • Our contrastive task exhibits robust performance even without the inclusion of expiratory images, while voxel-based methods demonstrate significant performance enhancement when incorporating expiratory images, in the COPDGene dataset. • Anomaly scores improved the fitting of linear mixed effects models in predicting clinical parameters and imaging alterations (p < 0.001) and were directly associated with clinical outcomes (p < 0.001). [ABSTRACT FROM AUTHOR]

Purpose: To examine whether there is a significant difference in image quality between the deep learning reconstruction (DLR [AiCE, Advanced Intelligent Clear-IQ Engine]) and hybrid iterative reconstruction (HIR [AIDR 3D, adaptive iterative dose reduction three dimensional]) algorithms on the conventional enhanced and CE-boost (contrast-enhancement-boost) images of indirect computed tomography venography (CTV) of lower extremities. Materials and methods: In this retrospective study, seventy patients who underwent CTV from June 2021 to October 2022 to assess deep vein thrombosis and varicose veins were included. Unenhanced and enhanced images were reconstructed for AIDR 3D and AiCE, AIDR 3D-boost and AiCE-boost images were obtained using subtraction software. Objective and subjective image qualities were assessed, and radiation doses were recorded. Results: The CT values of the inferior vena cava (IVC), femoral vein (FV), and popliteal vein (PV) in the CE-boost images were approximately 1.3 (1.31–1.36) times higher than in those of the enhanced images. There were no significant differences in mean CT values of IVC, FV, and PV between AIDR 3D and AiCE, AIDR 3D-boost and AiCE-boost images. Noise in AiCE, AiCE-boost images was significantly lower than in AIDR 3D and AIDR 3D-boost images (P < 0.05). The SNR (signal-to-noise ratio), CNR (contrast-to-noise ratio), and subjective scores of AiCE-boost images were the highest among 4 groups, surpassing AiCE, AIDR 3D, and AIDR 3D-boost images (all P < 0.05). Conclusion: In indirect CTV of the lower extremities images, DLR with the CE-boost technique could decrease the image noise and improve the CT values, SNR, CNR, and subjective image scores. AiCE-boost images received the highest subjective image quality score and were more readily accepted by radiologists. [ABSTRACT FROM AUTHOR]

Background: The deep learning (DL)-based reconstruction algorithm reduces noise in magnetic resonance imaging (MRI), thereby enabling faster MRI acquisition. Purpose: To compare the image quality and diagnostic performance of conventional turbo spin-echo (TSE) T2-weighted (T2W) imaging with DL-accelerated sagittal T2W imaging in the female pelvic cavity. Methods: This study evaluated 149 consecutive female pelvic MRI examinations, including conventional T2W imaging with TSE (acquisition time = 2:59) and DL-accelerated T2W imaging with breath hold (DL-BH) (1:05 [0:14 × 3 breath-holds]) in the sagittal plane. In 294 randomly ordered sagittal T2W images, two radiologists independently assessed image quality (sharpness, subjective noise, artifacts, and overall image quality), made a diagnosis for uterine leiomyomas, and scored diagnostic confidence. For the uterus and piriformis muscle, quantitative imaging analysis was also performed. Wilcoxon signed rank tests were used to compare the two sets of T2W images. Results: In the qualitative analysis, DL-BH showed similar or significantly higher scores for all features than conventional T2W imaging (P <0.05). In the quantitative analysis, the noise in the uterus was lower in DL-BH, but the noise in the muscle was lower in conventional T2W imaging. In the uterus and muscle, the signal-to-noise ratio was significantly lower in DL-BH than in conventional T2W imaging (P <0.001). The diagnostic performance of the two sets of T2W images was not different for uterine leiomyoma. Conclusions: DL-accelerated sagittal T2W imaging obtained with three breath-holds demonstrated superior or comparable image quality to conventional T2W imaging with no significant difference in diagnostic performance for uterine leiomyomas. [ABSTRACT FROM AUTHOR]

In response to the optimal extraction of DCT coefficients in facial images, the author proposes a DCT coefficient extraction method based on discriminant analysis. Based on the discriminant analysis of DCT coefficients, the DCT coefficients with high discriminant values are selected as features. Comparing the DPA based discrete cosine coefficient selection method proposed by the author with the traditional Zigzag discrete cosine coefficient selection method, experiments were conducted on the ORL face database and the Yale face database, respectively. The recognition performance on the ORL face database was higher than that on the Yale face database, as the facial image expression and lighting changes in the ORL database were relatively few, making it suitable for extracting key features. In response to the problem that the speech parameter MFCC is greatly affected by noise and can only reflect the static characteristics of speech, the author extracted gamma pass filtering cepstrum coefficients with human auditory characteristics and gamma pass sliding differential cepstrum coefficients that can reflect the dynamic characteristics of speech based on gamma tone filters and sliding differential cepstrum. In the NUST603 speech database, under pure background, the recognition rate based on GFSDCC features reached 89.88%, and the recognition effect based on GFCC features was 87.52%, which is 4.66% and 2.36% higher than that based on MFCC features. In noisy environments, the average recognition rates of speaker recognition systems based on GFCC and GFSDCC are 56.06% and 59.07%, while the average recognition rates of speaker recognition systems based on MFCC speech features are 53.89%, 2.17% and 5.18% higher, respectively. The gain in this recognition effect comes from the characteristics of the auditory model, as the Gammatone filter effectively reflects the noise resistance of the human auditory system. [ABSTRACT FROM AUTHOR]

Emotion recognition systems from speech signals are realized with the help of acoustic or spectral features. Acoustic analysis is the extraction of digital features from speech files using digital signal processing methods. Another method is the analysis of time-frequency images of speech using image processing. The size of the features obtained by acoustic analysis is in the thousands. Therefore, classification complexity increases and causes variation in classification accuracy. In feature selection, features unrelated to emotions are extracted from the feature space and are expected to contribute to the classifier performance. Traditional feature selection methods are mostly based on statistical analysis. Another feature selection method is the use of metaheuristic algorithms to detect and remove irrelevant features from the feature set. In this study, we compare the performance of metaheuristic feature selection algorithms for speech emotion recognition. For this purpose, a comparative analysis was performed on four different datasets, eight metaheuristics and three different classifiers. The results of the analysis show that the classification accuracy increases when the feature size is reduced. For all datasets, the highest accuracy was achieved with the support vector machine. The highest accuracy for the EMO-DB, EMOVA, eNTERFACE'05 and SAVEE datasets is 88.1%, 73.8%, 73.3% and 75.7%, respectively. [ABSTRACT FROM AUTHOR]

In order to achieve the goal of dynamically adjusting daily passenger flow to effectively control the overall efficiency of the transportation system, this study constructs a real-time monitoring and prediction system for subway passenger flow based on front-end voice processing technology and support vector machine models. The study first conducted a railway passenger flow analysis, and then used a support vector machine model to construct a preliminary prediction system. In order to achieve global optimization, the study also introduced particle swarm optimization algorithm to construct an optimization prediction model based on PSO-SVM. The results show that the proposed PSO-SVM method has undergone 48 iterations of training, and the predicted values closely match the actual passenger flow curve. The maximum RE error is 2%, and the overall prediction error is 98%. The decision coefficient of PSO-SVM is 0.998932. Therefore, this indicates that it has high performance and feasibility in predicting and controlling passenger flow during peak hours of urban rail transit. [ABSTRACT FROM AUTHOR]

Background: Cone-beam computed tomography (CBCT) has been introduced for breast-specimen imaging to identify a free resection margin of abnormal tissues in breast conservation. As well-known, typical micro CT consumes long acquisition and computation times. One simple solution to reduce the acquisition scan time is to decrease of the number of projections, but this method generates streak artifacts on breast specimen images. Furthermore, the presence of a metallic-needle marker on a breast specimen causes metal artifacts that are prominently visible in the images. In this work, we propose a deep learning-based approach for suppressing both streak and metal artifacts in CBCT. Methods: In this work, sinogram datasets acquired from CBCT and a small number of projections containing metal objects were used. The sinogram was first modified by removing metal objects and up sampling in the angular direction. Then, the modified sinogram was initialized by linear interpolation and synthesized by a modified neural network model based on a U-Net structure. To obtain the reconstructed images, the synthesized sinogram was reconstructed using the traditional filtered backprojection (FBP) approach. The remaining residual artifacts on the images were further handled by another neural network model, ResU-Net. The corresponding denoised image was combined with the extracted metal objects in the same data positions to produce the final results. Results: The image quality of the reconstructed images from the proposed method was improved better than the images from the conventional FBP, iterative reconstruction (IR), sinogram with linear interpolation, denoise with ResU-Net, sinogram with U-Net. The proposed method yielded 3.6 times higher contrast-to-noise ratio, 1.3 times higher peak signal-to-noise ratio, and 1.4 times higher structural similarity index (SSIM) than the traditional technique. Soft tissues around the marker on the images showed good improvement, and the mainly severe artifacts on the images were significantly reduced and regulated by the proposed. method. Conclusions: Our proposed method performs well reducing streak and metal artifacts in the CBCT reconstructed images, thus improving the overall breast specimen images. This would be beneficial for clinical use. [ABSTRACT FROM AUTHOR]

The objective of the current study was to develop and evaluate a DEep learning‐based rapid Spiral Image REconstruction (DESIRE) and deep learning (DL)‐based segmentation approach to quantify the left ventricular ejection fraction (LVEF) for high‐resolution spiral real‐time cine imaging, including 2D balanced steady‐state free precession imaging at 1.5 T and gradient echo (GRE) imaging at 1.5 and 3 T. A 3D U‐Net–based image reconstruction network and 2D U‐Net–based image segmentation network were proposed and evaluated. Low‐rank plus sparse (L+S) served as the reference for the image reconstruction network and manual contouring of the left ventricle was the reference of the segmentation network. To assess the image reconstruction quality, structural similarity index, peak signal‐to‐noise ratio, normalized root‐mean‐square error, and blind grading by two experienced cardiologists (5: excellent; 1: poor) were performed. To assess the segmentation performance, quantification of the LVEF on GRE imaging at 3 T was compared with the quantification from manual contouring. Excellent performance was demonstrated by the proposed technique. In terms of image quality, there was no difference between L+S and the proposed DESIRE technique. For quantification analysis, the proposed DL method was not different to the manual segmentation method (p > 0.05) in terms of quantification of LVEF. The reconstruction time for DESIRE was ~32 s (including nonuniform fast Fourier transform [NUFFT]) per dynamic series (40 frames), while the reconstruction time of L+S with GPU acceleration was approximately 3 min. The DL segmentation takes less than 5 s. In conclusion, the proposed DL‐based image reconstruction and quantification techniques enabled 1‐min image reconstruction for the whole heart and quantification with automatic reconstruction and quantification of the left ventricle function for high‐resolution spiral real‐time cine imaging with excellent performance. [ABSTRACT FROM AUTHOR]

This research suggests an efficient method that is better suited for speech processing applications to retrieve the accurate fundamental frequencies from speech in noisy conditions. For this objective, a fundamental frequency extraction algorithm that is tolerant to the non-stationary changes of the amplitude and frequency of the input signal is presented. Moreover, autocorrelation function is used as a pre-processor, which uses the power spectrum of the input signal to reduce the noise characteristics of the speech signals. In order to increase the accuracy of the fundamental frequency extraction, the focus has been on maintaining the speech harmonics in their original state and suppressing the noise elements involved in the noisy speech signal. The two stages that make up the suggested fundamental frequency extraction approach are producing the autocorrelation function of the speech signal as a pre-processor and applying it with the spectral subtraction method. As per the experiment results, the proposed technique appears to be better in noisy situations than other existing state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Introduction: Speaker diarization is an essential preprocessing step for diagnosing cognitive impairments from speech-based Montreal cognitive assessments (MoCA). Methods: This paper proposes three enhancements to the conventional speaker diarization methods for such assessments. The enhancements tackle the challenges of diarizing MoCA recordings on two fronts. First, multi-scale channel interdependence speaker embedding is used as the front-end speaker representation for overcoming the acoustic mismatch caused by far-field microphones. Specifically, a squeeze-and-excitation (SE) unit and channel-dependent attention are added to Res2Net blocks for multi-scale feature aggregation. Second, a sequence comparison approach with a holistic view of the whole conversation is applied to measure the similarity of short speech segments in the conversation, which results in a speaker-turn aware scoring matrix for the subsequent clustering step. Third, to further enhance the diarization performance, we propose incorporating a pairwise similarity measure so that the speaker-turn aware scoring matrix contains both local and global information across the segments. Results: Evaluations on an interactive MoCA dataset show that the proposed enhancements lead to a diarization system that outperforms the conventional x-vector/PLDA systems under language-, age-, and microphone-mismatch scenarios. Discussion: The results also show that the proposed enhancements can help hypothesize the speaker-turn timestamps, making the diarization method amendable to datasets without timestamp information. [ABSTRACT FROM AUTHOR]

The Spoofing attacks poses serious threats to Automatic Speaker verification (ASV) systems. Replay attack is one of such spoofing attacks which is a low effort yet high-risk attack due to ease of availability as well as widespread usage of smart phones, recording and playback devices. In replay spoofing attack, the attacker replays a pre-recorded speech samples of a genuine speaker to get unauthorized access. This study is focused on experimental analysis of different features with Gaussian Mixture Model (GMM) for development of countermeasures system for detection of replay spoofing attacks on ASV systems. GMM is a standard widely used as well as effective classifier for baseline countermeasures in ASV systems. The results are compared based on Equal Error Rate (EER) metric. This work presents a comparison of various spectral features for the most common spoofing attack i.e. replay spoofing attack detection based on its performance over the released ASVspoof challenge dataset of 2017, which is a benchmark dataset dedicated to address replay spoofing attacks only. [ABSTRACT FROM AUTHOR]

Introduction Noise obscures speech signal, causing auditory masking. The effects of this masking can be observed through the cortical auditory evoked potentials (CAEPs). White noise, in turn, has an effect on the auditory cortex, interfering, for example, with lexical decision making. Objective To analyze the effect of simultaneous masking by contralateral white noise on CAEPs elicited by speech stimuli. Methods Cross-sectional observational analytical study carried out with 15 participants of both sexes, who were submitted to CAEPs in two conditions: 1) without noise; 2) with white noise at 100 dBSPL intensity, contralaterally and simultaneously. To compare these conditions, the Student t test or the Wilcoxon test were used, depending on the sample normality. Differences with p values < 0.05 were considered significant. Results : When white noise was presented contralaterally and simultaneously to the CAEPs with speech stimulus, an increase in P1, N1 and P2 wave latencies was observed. P1 and P2 amplitudes and N1-P2 peak to peak amplitude also increased, unlike N1 amplitude, which decreased. The differences were significant for P1 and P2 wave latencies and for P2 wave amplitude. Conclusion The simultaneous masking effect was observed from the morphological alterations of the CAEPs with speech stimulus when white noise was presented in the contralateral ear. There was a significant increase in P1 and P2 wave latencies, as well as in P2 wave amplitude. [ABSTRACT FROM AUTHOR]

Background: The perceived impact of hearing loss varies considerably among those affected due to the heterogeneous types of hearing loss, their diverse etiologies, and the different rehabilitation possibilities. Therefore, assessing listening skills in a daily context using questionnaires is essential. This study aimed to investigate the validity and reliability of the adapted version of the Speech, Spatial and Qualities of Hearing Scale 12 (SSQ12) in the Romanian language. Materials and Methods: The SSQ12 is a 12-item self-reporting questionnaire that assesses a range of everyday listening situations. The internal consistency, test–retest reliability, and validity of the r-SSQ12 questionnaire resulting from the adaptation of the original scale were investigated. Results: The responses of 183 subjects aged between 11 and 79 years were evaluated. In total, 121 subjects had hearing loss (19 adolescents), and 62 subjects had normal hearing (11 adolescents). Significant differences were observed in the means of the overall score and for individual items between normal-hearing subjects and subjects with hearing loss. The SSQ12 had high internal consistency (Cronbach's alpha = 0.97), and the test–retest scores were highly correlated. Conclusions: The SSQ12 scale can be used to investigate the self-reporting of hearing quality in both general populations to identify hearing disorders and populations with hearing loss. [ABSTRACT FROM AUTHOR]

The image source method (ISM) is often used to simulate room acoustics due to its ease of use and computational efficiency. The standard ISM is limited to simulations of room impulse responses between point sources and omnidirectional receivers. In this work, the ISM is extended using spherical harmonic directivity coefficients to include acoustic diffraction effects. These effects occur in practice when transducers are mounted on audio devices of finite spatial extent, e.g., modern smart speakers with loudspeakers and microphones. The proposed method is verified using finite element simulations of various loudspeaker and microphone configurations in a shoebox-shaped room. It is shown that the accuracy of the proposed method is related to the sizes, shapes, number, and positions of the devices inside a room. A simplified version of the proposed method, which can significantly reduce computational effort, is also presented. The proposed method and its simplified version can simulate room transfer functions more accurately than currently available image source methods and can aid the development and evaluation of speech and acoustic signal processing algorithms, including speech enhancement, acoustic scene analysis, and acoustic parameter estimation. [ABSTRACT FROM AUTHOR]

This research relates to the field of speech technologies, where the key issue is the optimization of speech signal processing under conditions of a prior uncertainty of its fine structure. The problem of automatic (objective) analysis of the speaker's voice timbre using a speech signal of finite duration is considered. It is proposed to use a universal information-theoretic approach to solve it. Based on the Kullback-Leibler divergence, an expression was obtained to describe the asymptotically optimal decision statistic for differentiating speech signals by the voice timbre. The author highlights a serious obstacle during practical implementation of such statistics, namely: synchronization of the sequence of observations with the pitch of speech signals. To overcome the described obstacle, an objective measure of timbre-based differences in speech signals is proposed in terms of the acoustic theory of speech production and its "acoustic tube" type model of the speaker's vocal tract. The possibilities of practical implementation of a new measure based on an adaptive recursive filter are considered. A full-scale experiment was set up and carried out. The experimental results confirmed two main properties of the proposed measure: high sensitivity to differences in speech signals in terms of voice timbre and invariance with respect to the fundamental pitch frequency. The obtained results can be used when designing and studying digital speech processing systems tuned to the speaker's voice, for example, digital voice communication systems, biometric and biomedical systems, etc. [ABSTRACT FROM AUTHOR]

Gaining the ability to audit the behavior of deep learning (DL) denoising models is of crucial importance to prevent potential hallucinations and adversarial clinical consequences. We present a preliminary version of AntiHalluciNet, which is designed to predict spurious structural components embedded in the residual noise from DL denoising models in low-dose CT and assess its feasibility for auditing the behavior of DL denoising models. We created a paired set of structure-embedded and pure noise images and trained AntiHalluciNet to predict spurious structures in the structure-embedded noise images. The performance of AntiHalluciNet was evaluated by using a newly devised residual structure index (RSI), which represents the prediction confidence based on the presence of structural components in the residual noise image. We also evaluated whether AntiHalluciNet could assess the image fidelity of a denoised image by using only a noise component instead of measuring the SSIM, which requires both reference and test images. Then, we explored the potential of AntiHalluciNet for auditing the behavior of DL denoising models. AntiHalluciNet was applied to three DL denoising models (two pre-trained models, RED-CNN and CTformer, and a commercial software, ClariCT.AI [version 1.2.3]), and whether AntiHalluciNet could discriminate between the noise purity performances of DL denoising models was assessed. AntiHalluciNet demonstrated an excellent performance in predicting the presence of structural components. The RSI values for the structural-embedded and pure noise images measured using the 50% low-dose dataset were 0.57 ± 31 and 0.02 ± 0.02, respectively, showing a substantial difference with a p-value < 0.0001. The AntiHalluciNet-derived RSI could differentiate between the quality of the degraded denoised images, with measurement values of 0.27, 0.41, 0.48, and 0.52 for the 25%, 50%, 75%, and 100% mixing rates of the degradation component, which showed a higher differentiation potential compared with the SSIM values of 0.9603, 0.9579, 0.9490, and 0.9333. The RSI measurements from the residual images of the three DL denoising models showed a distinct distribution, being 0.28 ± 0.06, 0.21 ± 0.06, and 0.15 ± 0.03 for RED-CNN, CTformer, and ClariCT.AI, respectively. AntiHalluciNet has the potential to predict the structural components embedded in the residual noise from DL denoising models in low-dose CT. With AntiHalluciNet, it is feasible to audit the performance and behavior of DL denoising models in clinical environments where only residual noise images are available. [ABSTRACT FROM AUTHOR]

Objectives: To assess image quality and liver metastasis detection of reduced-dose dual-energy CT (DECT) with deep learning image reconstruction (DLIR) compared to standard-dose single-energy CT (SECT) with DLIR or iterative reconstruction (IR). Methods: In this prospective study, two groups of 40 participants each underwent abdominal contrast-enhanced scans with full-dose SECT (120-kVp images, DLIR and IR algorithms) or reduced-dose DECT (40- to 60-keV virtual monochromatic images [VMIs], DLIR algorithm), with 122 and 106 metastases, respectively. Groups were matched by age, sex ratio, body mass index, and cross-sectional area. Noise power spectrum of liver images and task-based transfer function of metastases were calculated to assess the noise texture and low-contrast resolution. The image noise, signal-to-noise ratios (SNR) of liver and portal vein, liver-to-lesion contrast-to-noise ratio (LLR), lesion conspicuity, lesion detection rate, and the subjective image quality metrics were compared between groups on 1.25-mm reconstructed images. Results: Compared to 120-kVp images with IR, 40- and 50-keV VMIs with DLIR showed similar noise texture and LLR, similar or higher image noise and low-contrast resolution, improved SNR and lesion conspicuity, and similar or better perceptual image quality. When compared to 120-kVp images with DLIR, 50-keV VMIs with DLIR had similar low-contrast resolution, SNR, LLR, lesion conspicuity, and perceptual image quality but lower frequency noise texture and higher image noise. For the detection of hepatic metastases, reduced-dose DECT by 34% maintained observer lesion detection rates. Conclusion: DECT assisted with DLIR enables a 34% dose reduction for detecting hepatic metastases while maintaining comparable perceptual image quality to full-dose SECT. Clinical relevance statement: Reduced-dose dual-energy CT with deep learning image reconstruction is as accurate as standard-dose single-energy CT for the detection of liver metastases and saves more than 30% of the radiation dose. Key Points: • The 40- and 50-keV virtual monochromatic images (VMIs) with deep learning image reconstruction (DLIR) improved lesion conspicuity compared with 120-kVp images with iterative reconstruction while providing similar or better perceptual image quality. • The 50-keV VMIs with DLIR provided comparable perceptual image quality and lesion conspicuity to 120-kVp images with DLIR. • The reduction of radiation by 34% by DLIR in low-keV VMIs is clinically sufficient for detecting low-contrast hepatic metastases. [ABSTRACT FROM AUTHOR]

Visual speech plays a powerful role in facilitating auditory speech processing and has been a publicly noticed topic with the wide usage of face masks during the COVID-19 pandemic. In a previous magnetoencephalography study, we showed that occluding the mouth area significantly impairs neural speech tracking. To rule out the possibility that this deterioration is because of degraded sound quality, in the present follow-up study, we presented participants with audiovisual (AV) and audio-only (A) speech. We further independently manipulated the trials by adding a face mask and a distractor speaker. Our results clearly show that face masks only affect speech tracking in AV conditions, not in A conditions. This shows that face masks indeed primarily impact speech processing by blocking visual speech and not by acoustic degradation. We can further highlight how the spectrogram, lip movements and lexical units are tracked on a sensor level. We can show visual benefits for tracking the spectrogram especially in the multi-speaker condition. While lip movements only show additional improvement and visual benefit over tracking of the spectrogram in clear speech conditions, lexical units (phonemes and word onsets) do not show visual enhancement at all. We hypothesize that in young normal hearing individuals, information from visual input is less used for specific feature extraction, but acts more as a general resource for guiding attention. [ABSTRACT FROM AUTHOR]

The pioneering AI-Based Speech Analysis Framework presented in this research paper was painstakingly created to help people overcome linguistic obstacles, notably in the context of English language communication. Through a thorough speech analysis, the framework's multimodal approach enables real-time evaluation of emotional state, fluency, stress levels, and even identification recognition. This framework delivers a sophisticated and perceptive interpretation of spoken language by utilizing cutting-edge artificial intelligence approaches, hence promoting an enhanced and successful communication experience. The study is focused on four significant sub-objectives, each of which advances the main objective of encouraging increased self-awareness and communication: First, by detecting subtly emotional indicators embedded in the voice, the framework transforms emotional assessment. The AI algorithms identify emotional patterns, such as enthusiasm, trepidation, or tranquility. This in-the-moment emotional analysis creates opportunities for tailored communication techniques and a greater understanding of the speaker's feelings. The framework also introduces a novel way for assessing fluency levels using voice analysis. It analyzes various facets of speech, such as pace, intonation, and lexical decisions, giving language learners immediate feedback on their level of linguistic proficiency. This makes it easier to make focused improvements and to move more easily toward effective communication. The framework also discusses the complex relationship between stress and good communication. It measures stress levels through vocal pattern analysis, offering light on instances of heightened tension or anxiety when speaking. Such knowledge enables people to overcome stress-related hurdles and enhance communication. The framework's capacity to accurately identify people based on distinctive voice traits lies at the heart of its innovation. Language limitations are no obstacle to this identity recognition technology, which provides an effective and secure method of identification in a variety of settings. Voice-based identification detection accelerates procedures and promotes inclusion in a variety of settings, including work settings and public services. The development of an AI-Based Speech Analysis Framework that reveals fresh angles in language evaluation and communication improvement is the culmination of this research. It not only encourages self-improvement but also highlights the revolutionary potential of AI in redefining language landscapes and promoting true connections by merging emotional, fluency, stress analysis, and identity identification through voice. [ABSTRACT FROM AUTHOR]

Target speaker separation aims to separate the speech components of the target speaker from mixed speech and remove extraneous components such as noise. In recent years, deep learning-based speech separation methods have made significant breakthroughs and have gradually become mainstream. However, these existing methods generally face problems with system latency and performance upper limits due to the large model size. To solve these problems, this paper proposes improvements in the network structure and training methods to enhance the model's performance. A lightweight target speaker separation network based on long-short-term memory (LSTM) is proposed, which can reduce the model size and computational delay while maintaining the separation performance. Based on this, a target speaker separation method based on joint training is proposed to achieve the overall training and optimization of the target speaker separation system. Joint loss functions based on speaker registration and speaker separation are proposed for joint training of the network to further improve the system's performance. The experimental results show that the lightweight target speaker separation network proposed in this paper has better performance while being lightweight, and joint training of the target speaker separation network with our proposed loss function can further improve the separation performance of the original model. [ABSTRACT FROM AUTHOR]

Objective: A study of supervised automated classification of the cervical vertebrae maturation (CVM) stages using deep learning (DL) network is presented. A parallel structured deep convolutional neural network (CNN) with a pre‐processing layer that takes X‐ray images and the age as the input is proposed. Methods: A total of 1018 cephalometric radiographs were labelled and classified according to the CVM stages. The images were separated according to gender for better model‐fitting. The images were cropped to extract the cervical vertebrae automatically using an object detector. The resulting images and the age inputs were used to train the proposed DL model: AggregateNet with a set of tunable directional edge enhancers. After the features of the images were extracted, the age input was concatenated to the output feature vector. To have the parallel network not overfit, data augmentation was used. The performance of our CNN model was compared with other DL models, ResNet20, Xception, MobileNetV2 and custom‐designed CNN model with the directional filters. Results: The proposed innovative model that uses a parallel structured network preceded with a pre‐processing layer of edge enhancement filters achieved a validation accuracy of 82.35% in CVM stage classification on female subjects, 75.0% in CVM stage classification on male subjects, exceeding the accuracy achieved with the other DL models investigated. The effectiveness of the directional filters is reflected in the improved performance attained in the results. If AggregateNet is used without directional filters, the test accuracy decreases to 80.0% on female subjects and to 74.03% on male subjects. Conclusion: AggregateNet together with the tunable directional edge filters is observed to produce higher accuracy than the other models that we investigated in the fully automated determination of the CVM stages. [ABSTRACT FROM AUTHOR]

Even prior to producing their first words, infants are developing a sophisticated speech processing system, with robust word recognition present by 4–6 months of age. These emergent linguistic skills, observed with behavioural investigations, are likely to rely on increasingly sophisticated neural underpinnings. The infant brain is known to robustly track the speech envelope, however previous cortical tracking studies were unable to demonstrate the presence of phonetic feature encoding. Here we utilise temporal response functions computed from electrophysiological responses to nursery rhymes to investigate the cortical encoding of phonetic features in a longitudinal cohort of infants when aged 4, 7 and 11 months, as well as adults. The analyses reveal an increasingly detailed and acoustically invariant phonetic encoding emerging over the first year of life, providing neurophysiological evidence that the pre-verbal human cortex learns phonetic categories. By contrast, we found no credible evidence for age-related increases in cortical tracking of the acoustic spectrogram. To understand speech, our brains have to learn the different types of sounds that constitute words, including syllables, stress patterns and smaller sound elements, such as phonetic categories. Here, the authors provide evidence that at 7 months, the infant brain learns reliably to detect invariant phonetic categories. [ABSTRACT FROM AUTHOR]

Speakers with dysarthria often struggle to accurately pronounce words and effectively communicate with others. Automatic speech recognition (ASR) is a powerful tool for extracting the content from speakers with dysarthria. However, the narrow concept of ASR typically only covers technologies that process acoustic modality signals. In this paper, we broaden the scope of this concept that the generalized concept of ASR for dysarthric speech. Our survey discussed the systems encompassed acoustic modality processing, articulatory movements processing and audio-visual modality fusion processing in the application of recognizing dysarthric speech. Contrary to previous surveys on dysarthric speech recognition, we have conducted a systematic review of the advancements in this field. In particular, we introduced state-of-the-art technologies to supplement the survey of recent research during the era of multi-modality fusion in dysarthric speech recognition. Our survey found that audio-visual fusion technologies perform better than traditional ASR technologies in the task of dysarthric speech recognition. However, training audio-visual fusion models requires more computing resources, and the available data corpus for dysarthric speech is limited. Despite these challenges, state-of-the-art technologies show promising potential for further improving the accuracy of dysarthric speech recognition in the future. [ABSTRACT FROM AUTHOR]

This article presents the research work on improving speech recognition systems for the morphologically complex Malayalam language using subword tokens for language modeling. The speech recognition system is built using a deep neural network–hidden Markov model (DNN-HMM)-based automatic speech recognition (ASR). We propose a novel method, syllable-byte pair encoding (S-BPE), that combines linguistically informed syllable tokenization with the data-driven tokenization method of byte pair encoding (BPE). The proposed method ensures words are always segmented at valid pronunciation boundaries. On a text corpus that has been divided into tokens using the proposed method, we construct statistical n-gram language models and assess the modeling effectiveness in terms of both information-theoretic and corpus linguistic metrics. A comparative study of the proposed method with other data-driven (BPE, Morfessor, and Unigram), linguistic (Syllable), and baseline (Word) tokenization algorithms is also presented. Pronunciation lexicons of subword tokenized units are built with pronunciation described as graphemes. We develop ASR systems employing the subword tokenized language models and pronunciation lexicons. The resulting ASR models are comprehensively evaluated to answer the research questions regarding the impact of subword tokenization algorithms on language modeling complexity and on ASR performance. Our study highlights the strong performance of the hybrid S-BPE tokens, achieving a notable 10.6% word error rate (WER), which represents a substantial 16.8% improvement over the baseline word-level ASR system. The ablation study has revealed that the performance of S-BPE segmentation, which initially underperformed compared to syllable tokens with lower amounts of textual data for language modeling, exhibited steady improvement with the increase in LM training data. The extensive ablation study indicates that there is a limited advantage in raising the n-gram order of the language model beyond n = 3 . Such an increase results in considerable model size growth without significant improvements in WER. The implementation of the algorithm and all associated experiments are available under an open license, allowing for reproduction, adaptation, and reuse. [ABSTRACT FROM AUTHOR]

The most prominent acoustic features in speech are intensity modulations, represented by the amplitude envelope of speech. Synchronization of neural activity with these modulations supports speech comprehension. As the acoustic modulation of speech is related to the production of syllables, investigations of neural speech tracking commonly do not distinguish between lower‐level acoustic (envelope modulation) and higher‐level linguistic (syllable rate) information. Here we manipulated speech intelligibility using noise‐vocoded speech and investigated the spectral dynamics of neural speech processing, across two studies at cortical and subcortical levels of the auditory hierarchy, using magnetoencephalography. Overall, cortical regions mostly track the syllable rate, whereas subcortical regions track the acoustic envelope. Furthermore, with less intelligible speech, tracking of the modulation rate becomes more dominant. Our study highlights the importance of distinguishing between envelope modulation and syllable rate and provides novel possibilities to better understand differences between auditory processing and speech/language processing disorders. Natural speech consists of both acoustic sound modulations and linguistic features, such as the rate at which syllables are expressed. Current investigations of neural speech tracking do not distinguish between low‐level acoustic and linguistic information. Instead, the low‐frequency acoustic modulations of speech are commonly simply equated to the production of syllables. However, while the temporal scale of the low‐level acoustic modulations of speech is surprisingly stable across various speech settings (different speakers, languages, and speaking conditions) the production rate of syllables can differ across the same settings. We show that cortical regions track the syllabic rate, whereas subcortical regions track the modulation rate of speech. Furthermore, with declining intelligibility tracking of the modulation rate becomes overall more dominant. [ABSTRACT FROM AUTHOR]

Objectives: To examine a compressed sensing artificial intelligence (CSAI) framework to accelerate image acquisition in non-contrast-enhanced whole-heart bSSFP coronary magnetic resonance (MR) angiography. Methods: Thirty healthy volunteers and 20 patients with suspected coronary artery disease (CAD) scheduled for coronary computed tomography angiography (CCTA) were enrolled. Non-contrast-enhanced coronary MR angiography was performed with CSAI, compressed sensing (CS), and sensitivity encoding (SENSE) methods in healthy participants and with CSAI in patients. Acquisition time, subjective image quality score, and objective image quality measurement (blood pool homogeneity, signal-to-noise ratio [SNR], and contrast-to-noise ratio [CNR]) were compared among the three protocols. The diagnostic performance of CASI coronary MR angiography for predicting significant stenosis (≥ 50% diameter stenosis) on CCTA was evaluated. The Friedman test was performed to compare the three protocols. Results: Acquisition time was significantly shorter in the CSAI and CS groups than in the SENSE group (10.2 ± 3.2 min vs. 10.9 ± 2.9 min vs. 13.0 ± 4.1 min, p < 0.001). However, the CSAI approach had the highest image quality scores, blood pool homogeneity, mean SNR value, and mean CNR value (all p < 0.001) compared with the CS and SENSE approaches. The sensitivity, specificity, and accuracy of CSAI coronary MR angiography per patient were 87.5% (7/8), 91.7% (11/12), and 90.0% (18/20); those per vessel were 81.8% (9/11), 93.9% (46/49), and 91.7% (55/60); and those per segment were 84.6% (11/13), 98.0% (244/249), and 97.3% (255/262), respectively. Conclusions: CSAI yielded superior image quality within a clinically feasible acquisition time in healthy participants and patients with suspected CAD. Clinical relevance statement: The non-invasive and radiation-free CSAI framework could be a promising tool for rapid screening and comprehensive examination of the coronary vasculature in patients with suspected CAD. Key Points: • This prospective study showed that CSAI enables a reduction in acquisition time by 22% with superior diagnostic image quality compared with the SENSE protocol. • CSAI replaces the wavelet transform with a CNN as a sparsifying transform in the CS algorithm, achieving high coronary MR image quality with reduced noise. • CSAI achieved per-patient sensitivity of 87.5% (7/8) and specificity of 91.7% (11/12) respectively for detecting significant coronary stenosis. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Purpose: To develop a deep learning‐based method, dubbed Denoising CEST Network (DECENT), to fully exploit the spatiotemporal correlation prior to CEST image denoising. Methods: DECENT is composed of two parallel pathways with different convolution kernel sizes aiming to extract the global and spectral features embedded in CEST images. Each pathway consists of a modified U‐Net with residual Encoder‐Decoder network and 3D convolution. Fusion pathway with 1 × 1 × 1 convolution kernel is utilized to concatenate two parallel pathways, and the output of DECENT is noise‐reduced CEST images. The performance of DECENT was validated in numerical simulations, egg white phantom experiments, and ischemic mouse brain and human skeletal muscle experiments in comparison with existing state‐of‐the‐art denoising methods. Results: Rician noise was added to CEST images to mimic a low SNR situation for numerical simulation, egg white phantom experiment, and mouse brain experiments, while human skeletal muscle experiments were of inherently low SNR. From the denoising results evaluated by peak SNR (PSNR) and structural similarity index (SSIM), the proposed deep learning‐based denoising method (DECENT) can achieve better performance compared to existing CEST denoising methods such as NLmCED, MLSVD, and BM4D, avoiding complicated parameter tuning or time‐consuming iterative processes. Conclusions: DECENT can well exploit the prior spatiotemporal correlation knowledge of CEST images and restore the noise‐free images from their noisy observations, outperforming state‐of‐the‐art denoising methods. [ABSTRACT FROM AUTHOR]

Purpose: The inherently poor SNR of MRS measurements presents a significant hurdle to its clinical application. Denoising by machine or deep learning (DL) was proposed as a remedy. It is investigated whether such denoising leads to lower estimate uncertainties or whether it essentially reduces noise in signal‐free areas only. Methods: Noise removal based on supervised DL with U‐nets was implemented using simulated 1H MR spectra of human brain in two approaches: (1) via time‐frequency domain spectrograms and (2) using 1D spectra as input. Quality of denoising was evaluated in three ways: (1) by an adapted fit quality score, (2) by traditional model fitting, and (3) by quantification via neural networks. Results: Visually appealing spectra were obtained; hinting that denoising is well‐suited for MRS. However, an adapted denoising score showed that noise removal is inhomogeneous and more efficient for signal‐free areas. This was confirmed by quantitative analysis of traditional fit results as well as DL quantitation following DL denoising. DL denoising, although apparently successful as judged by mean squared errors, led to substantially biased estimates in both implementations. Conclusion: The implemented DL‐based denoising techniques may be useful for display purposes, but do not help quantitative evaluations, confirming expectations based on estimation theory: Cramér Rao lower bounds defined by the original data and the appropriate fitting model cannot be circumvented in an unbiased way for single data sets, unless additional prior knowledge can be incurred in the form of parameter restrictions/relations or applicable substates. [ABSTRACT FROM AUTHOR]

Purpose: To develop a method for building MRI reconstruction neural networks robust to changes in signal‐to‐noise ratio (SNR) and trainable with a limited number of fully sampled scans. Methods: We propose Noise2Recon, a consistency training method for SNR‐robust accelerated MRI reconstruction that can use both fully sampled (labeled) and undersampled (unlabeled) scans. Noise2Recon uses unlabeled data by enforcing consistency between model reconstructions of undersampled scans and their noise‐augmented counterparts. Noise2Recon was compared to compressed sensing and both supervised and self‐supervised deep learning baselines. Experiments were conducted using retrospectively accelerated data from the mridata three‐dimensional fast‐spin‐echo knee and two‐dimensional fastMRI brain datasets. All methods were evaluated in label‐limited settings and among out‐of‐distribution (OOD) shifts, including changes in SNR, acceleration factors, and datasets. An extensive ablation study was conducted to characterize the sensitivity of Noise2Recon to hyperparameter choices. Results: In label‐limited settings, Noise2Recon achieved better structural similarity, peak signal‐to‐noise ratio, and normalized‐RMS error than all baselines and matched performance of supervised models, which were trained with 14×$$ 14\times $$ more fully sampled scans. Noise2Recon outperformed all baselines, including state‐of‐the‐art fine‐tuning and augmentation techniques, among low‐SNR scans and when generalizing to OOD acceleration factors. Augmentation extent and loss weighting hyperparameters had negligible impact on Noise2Recon compared to supervised methods, which may indicate increased training stability. Conclusion: Noise2Recon is a label‐efficient reconstruction method that is robust to distribution shifts, such as changes in SNR, acceleration factors, and others, with limited or no fully sampled training data. [ABSTRACT FROM AUTHOR]

Speech processing is today's trending topic in the digital industry for making authentication to keep aware of unauthorized ones. However, analyzing the signal feature using conventional filtering or the neural models is insufficient due to the present signal's several noisy features. Hence, incorporating the different noise elimination filters has maximized the algorithm complexity in verifying the user speech signal. So, the present study built a novel Chimp-based recursive Speech Identification (CbRSI) system for the speech processing domain to verify the authenticated users through the speech signal data. To make signal processing the most straightforward task was activated the filtering function to recognize and neglect the noisy features. Consequently, the filtered audio data is imported as the classification phase input then the features-selecting process is performed. Finally, authenticated users were found and validated the performance by matching the analyzed signal features with the saved audio features. Hence, a novel CbRSI earned the finest user verification exactness score of 98.2%, which is the most satisfactory outcome compared to past studies. Therefore, the implemented solution is the most required framework for verifying authenticated users. [ABSTRACT FROM AUTHOR]

Background: This study aimed to develop a deep learning (DL) algorithm that enhances the quality of a single-frame enface OCTA scan to make it comparable to 4-frame averaged scan without the need for the repeated acquisitions required for averaging. Methods: Each of the healthy eyes and eyes from diabetic subjects that were prospectively enrolled in this cross-sectional study underwent four repeated 6 × 6 mm macular scans (PLEX Elite 9000 SS-OCT), and the repeated scans of each eye were co-registered to produce 4-frame averages. This prospective dataset of original (single-frame) enface scans and their corresponding averaged scans was divided into a training dataset and a validation dataset. In the training dataset, a DL algorithm (named pseudoaveraging) was trained using original scans as input and 4-frame averages as target. In the validation dataset, the pseudoaveraging algorithm was applied to single-frame scans to produce pseudoaveraged scans, and the single-frame and its corresponding averaged and pseudoaveraged scans were all qualitatively compared. In a separate retrospectively collected dataset of single-frame scans from eyes of diabetic subjects, the DL algorithm was applied, and the produced pseudoaveraged scan was qualitatively compared against its corresponding original. Results: This study included 39 eyes that comprised the prospective dataset (split into 5 eyes for training and 34 eyes for validating the DL algorithm), and 105 eyes that comprised the retrospective test dataset. Of the total 144 study eyes, 58% had any level of diabetic retinopathy (with and without diabetic macular edema), and the rest were from healthy eyes or eyes of diabetic subjects but without diabetic retinopathy and without macular edema. Grading results in the validation dataset showed that the pseudoaveraged enface scan ranked best in overall scan quality, background noise reduction, and visibility of microaneurysms (p < 0.05). Averaged scan ranked best for motion artifact reduction (p < 0.05). Grading results in the test dataset showed that pseudoaveraging resulted in enhanced small vessels, reduction of background noise, and motion artifact in 100%, 82%, and 98% of scans, respectively. Rates of false-positive/-negative perfusion were zero. Conclusion: Pseudoaveraging is a feasible DL approach to more efficiently improve enface OCTA scan quality without introducing notable image artifacts. [ABSTRACT FROM AUTHOR]

Introduction: Diagnosing Alzheimer's disease (AD) lesions via visual examination of Electroencephalography (EEG) signals poses a considerable challenge. This has prompted the exploration of deep learning techniques, such as Convolutional Neural Networks (CNNs) and Visual Transformers (ViTs), for AD prediction. However, the classification performance of CNN-based methods has often been deemed inadequate. This is primarily attributed to CNNs struggling with extracting meaningful lesion signals from the complex and noisy EEG data. Methods: In contrast, ViTs have demonstrated proficiency in capturing global signal patterns. In light of these observations, we propose a novel approach to enhance AD risk assessment. Our proposition involves a hybrid architecture, merging the strengths of CNNs and ViTs to compensate for their respective feature extraction limitations. Our proposed Dual-Branch Feature Fusion Network (DBN) leverages both CNN and ViT components to acquire texture features and global semantic information from EEG signals. These elements are pivotal in capturing dynamic electrical signal changes in the cerebral cortex. Additionally, we introduce Spatial Attention (SA) and Channel Attention (CA) blocks within the network architecture. These attention mechanisms bolster the model's capacity to discern abnormal EEG signal patterns from the amalgamated features. To make well-informed predictions, we employ a two-factor decision-making mechanism. Specifically, we conduct correlation analysis on predicted EEG signals from the same subject to establish consistency. Results: This is then combined with results from the Clinical Neuropsychological Scale (MMSE) assessment to comprehensively evaluate the subject's susceptibility to AD. Our experimental validation on the publicly available OpenNeuro database underscores the efficacy of our approach. Notably, our proposed method attains an impressive 80.23% classification accuracy in distinguishing between AD, Frontotemporal dementia (FTD), and Normal Control (NC) subjects. Discussion: This outcome outperforms prevailing state-of-the-artmethodologies in EEG-based AD prediction. Furthermore, our methodology enables the visualization of salient regions within pathological images, providing invaluable insights for interpreting and analyzing AD predictions. [ABSTRACT FROM AUTHOR]

Many speech processing systems' crucial frontends include speech enhancement. Single-channel speech enhancement experiences a number of technological challenges. Due to the advent of cloud-based technology and the use of deep learning systems in big data, deep neural networks in particular have recently been seen as a potent means for complex classification and regression. In this work, spectral gating noise filter is combined with deep neural network U-Net to enhance the performance of speech enhancement network. Further, for performance analysis three distinct objective functions namely, Mean Square Error, Huber Loss and Mean Absolute Error are considered as loss functions. In addition, comparison of three different optimizers Adam, Adagrad and Stochastic Gradient Descent is presented. Proposed system is tested and evaluated on LibriSpeech and NOIZEUS datasets and compared to other state-of-the-art systems. It demonstrates that, in comparison to other state-of-the-art models, the proposed network outperformed them with PESQ scores of 2.737420 for training and 2.67857 for testing, along with better generalization ability. [ABSTRACT FROM AUTHOR]

With the rapid development of multimedia technology, it is very important to improve the accuracy of sparse signal reconstruction for image or speech based on compressed sensing technology. In this work, a matching pursuit algorithm for sparse signal reconstruction based on Jaccard coefficient and backtracking strategy is proposed to improve the accurate reconstruction probability. Jaccard coefficient matching criterion and inner product matching criterion are combined to preserve more features of the original vector. A mean-based thresholding strategy is used to selects atoms adaptively, and the backtracking method is introduced to filter out the wrong atoms. Experimental results show that the reconstruction performance of the proposed algorithm is much better than other greedy algorithms when the sparsity K is large (K>35). For speech signals, the proposed algorithm has better reconstruction performance than other algorithms for both male and female voice signals. Under different compression ratios, the AFSNR and the PESQ scores of the reconstructed speech by the proposed algorithm are increased by at least 1.5348 dB and 0.5629, respectively, especially showing superior performance under larger compression ratio. The proposed algorithm has bright application prospect in intelligent speech interaction with accurate reconstruction of speech signal under large compression ratios. [ABSTRACT FROM AUTHOR]

Chemical exchange saturation transfer (CEST) MRI is a promising molecular imaging technique but suffers from long scan times and complicated processing. CEST was recently combined with magnetic resonance fingerprinting (MRF) to address these shortcomings. However, the CEST‐MRF signal depends on multiple acquisition and tissue parameters so selecting an optimal acquisition schedule is challenging. In this work, we propose a novel dual‐network deep learning framework to optimize the CEST‐MRF acquisition schedule. The quality of the optimized schedule was assessed in a digital brain phantom and compared with alternate deep learning optimization approaches. The effect of schedule length on the reconstruction error was also investigated. A healthy subject was scanned with optimized and random schedules and with a conventional CEST sequence for comparison. The optimized schedule was also tested in a subject with metastatic renal cell carcinoma. Reproducibility was assessed via test–retest experiments and the concordance correlation coefficient calculated for white matter (WM) and grey matter (GM). The optimized schedule was 12% shorter but yielded equal or lower normalized root mean square error for all parameters. The proposed optimization also provided a lower error compared with alternate methodologies. Longer schedules generally yielded lower error. In vivo maps obtained with the optimized schedule showed reduced noise and improved delineation of GM and WM. CEST curves synthesized from the optimized parameters were highly correlated (r = 0.99) with measured conventional CEST. The mean concordance correlation coefficient in WM/GM for all tissue parameters was 0.990/0.978 for the optimized schedule but only 0.979/0.975 for the random schedule. The proposed schedule optimization is widely applicable to MRF pulse sequences and provides accurate and reproducible tissue maps with reduced noise at a shorter scan time than a randomly generated schedule. [ABSTRACT FROM AUTHOR]

Purpose: Water saturation shift referencing (WASSR) Z‐spectra are used commonly for field referencing in chemical exchange saturation transfer (CEST) MRI. However, their analysis using least‐squares (LS) Lorentzian fitting is time‐consuming and prone to errors because of the unavoidable noise in vivo. A deep learning–based single Lorentzian Fitting Network (sLoFNet) is proposed to overcome these shortcomings. Methods: A neural network architecture was constructed and its hyperparameters optimized. Training was conducted on a simulated and in vivo–paired data sets of discrete signal values and their corresponding Lorentzian shape parameters. The sLoFNet performance was compared with LS on several WASSR data sets (both simulated and in vivo 3T brain scans). Prediction errors, robustness against noise, effects of sampling density, and time consumption were compared. Results: LS and sLoFNet performed comparably in terms of RMS error and mean absolute error on all in vivo data with no statistically significant difference. Although the LS method fitted well on samples with low noise, its error increased rapidly when increasing sample noise up to 4.5%, whereas the error of sLoFNet increased only marginally. With the reduction of Z‐spectral sampling density, prediction errors increased for both methods, but the increase occurred earlier (at 25 vs. 15 frequency points) and was more pronounced for LS. Furthermore, sLoFNet performed, on average, 70 times faster than the LS‐method. Conclusion: Comparisons between LS and sLoFNet on simulated and in vivo WASSR MRI Z‐spectra in terms of robustness against noise and decreased sample resolution, as well as time consumption, showed significant advantages for sLoFNet. [ABSTRACT FROM AUTHOR]

Purpose: To propose a novel end‐to‐end deep learning model to quantify absolute metabolite concentrations from in vivo J‐point resolved spectroscopy (JPRESS) without using spectral fitting. Methods: A novel encoder‐decoder‐style neural network was created, which was trained to predict metabolite concentrations and individual component signals concurrently from 3T JPRESS data in the time domain. The training data set contained 100 000 samples created by spin‐density simulations using experimentally used RF pulses. Concentrations, phase, frequencies, linewidths, and T2 relaxation times in the training data set were varied over a large range with uniform distributions. Random synthesized noise and extraneous signals were added to the data set. Two thousand validation samples were created similarly to the training data set but with mean concentrations close to in vivo values. An in vivo test was conducted with 20 samples acquired from the human brain. Results: Both validation and in vivo test results showed that the proposed model successfully predicted metabolite concentrations as well as individual metabolite signals without involving spectral fitting, while extraneous peaks or unregistered signals were filtered out. Compared with the short‐TE spectral fitting by LCModel, the proposed method had the advantage that the undesired correlations between the estimated concentrations and noise levels and between metabolites were eliminated or substantially reduced. Conclusion: The proposed method provides a working deep learning model that directly maps in vivo JPRESS data to metabolite concentrations. Because spectral fitting is not used, the trained model does not depend on the assumptions associated with parameter tuning when applied to in vivo data. [ABSTRACT FROM AUTHOR]