J Neural Eng. 2024 Jun 17. doi: 10.1088/1741-2552/ad593c. Online ahead of print.

ABSTRACT

Brain decoding is a field of computational neuroscience that aims to infer mental states or internal representations of perceptual inputs from measurable brain activity. In this study, we propose a novel approach to brain decoding that relies on semantic and contextual similarity. We use several fMRI datasets of natural images as stimuli and create a deep learning decoding pipeline inspired by the bottom-up and top-down processes in human vision.

Our pipeline includes a linear brain-to-feature model that maps fMRI activity to semantic visual stimuli features. We assume that the brain projects visual information onto a space that is homeomorphic to the latent space represented the last layer of a pretrained neural network, which summarizes and highlights similarities and differences between concepts. These features are categorized in the latent space using a nearest-neighbor strategy, and the results are used to retrieve images or condition a generative latent diffusion model to create novel images. We demonstrate semantic classification and image retrieval on three different fMRI datasets, GOD (vision perception and imagination), BOLD5000 and NSD. In all cases a simple mapping between fMRI and a deep semantic representation of the visual stimulus resulted in meaningful classification and retrieved or generated images. We assessed quality using quantitative metrics and a human evaluation experiment that reproduces the multiplicity of conscious and unconscious criteria that humans use to evaluate image similarity. Our method achieved correct evaluation in over 80% of the test set.

In summary, our study proposes a novel approach to brain decoding that relies on semantic and contextual similarity. Our results demonstrate that measurable neural correlates can be linearly mapped onto the latent space of a neural network to synthesize images that match the original content. The findings have implications for both cognitive neuroscience and artificial intelligence.

PMID:38885689 | DOI:10.1088/1741-2552/ad593c

J Neural Eng. 2024 Jun 17. doi: 10.1088/1741-2552/ad593a. Online ahead of print.

ABSTRACT

Abstract-Objective: Brain-computer interfaces (BCIs) are technologies that bypass damaged or disrupted neural pathways and directly decode brain signals to perform intended actions. BCIs for speech have the potential to restore communication by decoding the intended speech directly. Many studies have demonstrated promising results using invasive micro-electrode arrays (MEA) and electrocorticography (ECoG). However, the use of stereoelectroencephalography (sEEG) for speech decoding has not been fully recognized.
Methods: In this research, recently-released sEEG data were used to decode Dutch words spoken by participants. We decoded speech waveforms from sEEG data using advanced deep-learning methods. Three methods were implemented: a linear regression method, a sequence-to-sequence model (seq2seq), and a transformer model.
Results: Our RNN-based seq2seq and transformer models outperformed the linear regression significantly, while no significant difference was found between the two deep-learning methods. Further investigation on individual electrodes showed that the same decoding result can be obtained using only few of the electrodes.
Conclusion: This study demonstrated that decoding speech from sEEG signals is possible, and the location of the electrodes is critical to the decoding performance.

PMID:38885688 | DOI:10.1088/1741-2552/ad593a

Spectrochim Acta A Mol Biomol Spectrosc. 2024 Jun 13;320:124655. doi: 10.1016/j.saa.2024.124655. Online ahead of print.

ABSTRACT

Rapid and quantitative detection of malachite green (MG) in aquaculture products is very important for safety assurance in food supply. Here, we develop a point-of-care testing (POCT) platform that combines a flexible and transparent surface-enhanced Raman scattering (SERS) substrate with deep learning network for achieving rapid and quantitative detection of MG in fish. The flexible and transparent SERS substrate was prepared by depositing silver (Ag) film on the polydimethylsiloxane (PDMS) film using laser molecular beam epitaxy (LMBE) technique. The wrinkled Ag NPs@PDMS film exhibits high SERS activity, excellent reproducibility and good mechanical stability. Additionally, the fast in situ detection of MG residues onfishscales was achieved by using the wrinkled Ag NPs/PDMS film and a portable Raman spectrometer, with a minimum detectable concentration of 10-6 M. Subsequently, a one-dimensional convolutional neural network (1D CNN) model was constructed for rapid quantification of MG concentration. The results demonstrated that the 1D CNN quantitative analysis model possessed superior predictive performance, with a coefficient of determination (R2) of 0.9947 and a mean squared error (MSE) of 0.0104. The proposed POCT platform, integrating a transparent flexible SERS substrate, a portable Raman spectrometer and a 1D CNN model, provides an efficient strategy for rapid identification and quantitative analysis of MG in fish.

PMID:38885572 | DOI:10.1016/j.saa.2024.124655

Biomed Eng Online. 2024 Jun 17;23(1):55. doi: 10.1186/s12938-024-01250-y.

ABSTRACT

BACKGROUND: Schizophrenia (SZ), a psychiatric disorder for which there is no precise diagnosis, has had a serious impact on the quality of human life and social activities for many years. Therefore, an advanced approach for accurate treatment is required.

NEW METHOD: In this study, we provide a classification approach for SZ patients based on a spatial-temporal residual graph convolutional neural network (STRGCN). The model primarily collects spatial frequency features and temporal frequency features by spatial graph convolution and single-channel temporal convolution, respectively, and blends them both for the classification learning, in contrast to traditional approaches that only evaluate temporal frequency information in EEG and disregard spatial frequency features across brain regions.

RESULTS: We conducted extensive experiments on the publicly available dataset Zenodo and our own collected dataset. The classification accuracy of the two datasets on our proposed method reached 96.32% and 85.44%, respectively. In the experiment, the dataset using delta has the best classification performance in the sub-bands.

COMPARISON WITH EXISTING METHODS: Other methods mainly rely on deep learning models dominated by convolutional neural networks and long and short time memory networks, lacking exploration of the functional connections between channels. In contrast, the present method can treat the EEG signal as a graph and integrate and analyze the temporal frequency and spatial frequency features in the EEG signal.

CONCLUSION: We provide an approach to not only performs better than other classic machine learning and deep learning algorithms on the dataset we used in diagnosing schizophrenia, but also understand the effects of schizophrenia on brain network features.

PMID:38886737 | DOI:10.1186/s12938-024-01250-y

BMC Med Imaging. 2024 Jun 17;24(1):147. doi: 10.1186/s12880-024-01323-3.

ABSTRACT

Diagnosing brain tumors is a complex and time-consuming process that relies heavily on radiologists' expertise and interpretive skills. However, the advent of deep learning methodologies has revolutionized the field, offering more accurate and efficient assessments. Attention-based models have emerged as promising tools, focusing on salient features within complex medical imaging data. However, the precise impact of different attention mechanisms, such as channel-wise, spatial, or combined attention within the Channel-wise Attention Mode (CWAM), for brain tumor classification remains relatively unexplored. This study aims to address this gap by leveraging the power of ResNet101 coupled with CWAM (ResNet101-CWAM) for brain tumor classification. The results show that ResNet101-CWAM surpassed conventional deep learning classification methods like ConvNet, achieving exceptional performance metrics of 99.83% accuracy, 99.21% recall, 99.01% precision, 99.27% F1-score and 99.16% AUC on the same dataset. This enhanced capability holds significant implications for clinical decision-making, as accurate and efficient brain tumor classification is crucial for guiding treatment strategies and improving patient outcomes. Integrating ResNet101-CWAM into existing brain classification software platforms is a crucial step towards enhancing diagnostic accuracy and streamlining clinical workflows for physicians.

PMID:38886661 | DOI:10.1186/s12880-024-01323-3

Sci Rep. 2024 Jun 17;14(1):13914. doi: 10.1038/s41598-024-63955-3.

ABSTRACT

This research paper presents a comprehensive investigation into the utilization of color image processing technologies and deep learning algorithms in the development of a robot vision system specifically designed for 8-ball billiards. The sport of billiards, with its various games and ball arrangements, presents unique challenges for robotic vision systems. The proposed methodology addresses these challenges through two main components: object detection and ball pattern recognition. Initially, a robust algorithm is employed to detect the billiard balls using color space transformation and thresholding techniques. This is followed by determining the position of the billiard table through strategic cropping and isolation of the primary table area. The crucial phase involves the intricate task of recognizing ball patterns to differentiate between solid and striped balls. To achieve this, a modified convolutional neural network is utilized, leveraging the Xception network optimized by an innovative algorithm known as the Improved Chaos African Vulture Optimization (ICAVO) algorithm. The ICAVO algorithm enhances the Xception network's performance by efficiently exploring the solution space and avoiding local optima. The results of this study demonstrate a significant enhancement in recognition accuracy, with the Xception/ICAVO model achieving remarkable recognition rates for both solid and striped balls. This paves the way for the development of more sophisticated and efficient billiards robots. The implications of this research extend beyond 8-ball billiards, highlighting the potential for advanced robotic vision systems in various applications. The successful integration of color image processing, deep learning, and optimization algorithms shows the effectiveness of the proposed methodology. This research has far-reaching implications that go beyond just billiards. The cutting-edge robotic vision technology can be utilized for detecting and tracking objects in different sectors, transforming industrial automation and surveillance setups. By combining color image processing, deep learning, and optimization algorithms, the system proves its effectiveness and flexibility. The innovative approach sets the stage for creating advanced and productive robotic vision systems in various industries.

PMID:38886386 | DOI:10.1038/s41598-024-63955-3

Nat Commun. 2024 Jun 17;15(1):5163. doi: 10.1038/s41467-024-49516-2.

ABSTRACT

As the most abundant organic substances in nature, carbohydrates are essential for life. Understanding how carbohydrates regulate proteins in the physiological and pathological processes presents opportunities to address crucial biological problems and develop new therapeutics. However, the diversity and complexity of carbohydrates pose a challenge in experimentally identifying the sites where carbohydrates bind to and act on proteins. Here, we introduce a deep learning model, DeepGlycanSite, capable of accurately predicting carbohydrate-binding sites on a given protein structure. Incorporating geometric and evolutionary features of proteins into a deep equivariant graph neural network with the transformer architecture, DeepGlycanSite remarkably outperforms previous state-of-the-art methods and effectively predicts binding sites for diverse carbohydrates. Integrating with a mutagenesis study, DeepGlycanSite reveals the guanosine-5'-diphosphate-sugar-recognition site of an important G-protein coupled receptor. These findings demonstrate DeepGlycanSite is invaluable for carbohydrate-binding site prediction and could provide insights into molecular mechanisms underlying carbohydrate-regulation of therapeutically important proteins.

PMID:38886381 | DOI:10.1038/s41467-024-49516-2

Sci Rep. 2024 Jun 17;14(1):13894. doi: 10.1038/s41598-024-64265-4.

ABSTRACT

Stroke is one of the major causes of death worldwide, and is closely associated with atherosclerosis of the carotid artery. Panoramic radiographs (PRs) are routinely used in dental practice, and can be used to visualize carotid artery calcification (CAC). The purpose of this study was to automatically and robustly classify and segment CACs with large variations in size, shape, and location, and those overlapping with anatomical structures based on deep learning analysis of PRs. We developed a cascaded deep learning network (CACSNet) consisting of classification and segmentation networks for CACs on PRs. This network was trained on ground truth data accurately determined with reference to CT images using the Tversky loss function with optimized weights by balancing between precision and recall. CACSNet with EfficientNet-B4 achieved an AUC of 0.996, accuracy of 0.985, sensitivity of 0.980, and specificity of 0.988 in classification for normal or abnormal PRs. Segmentation performances for CAC lesions were 0.595 for the Jaccard index, 0.722 for the Dice similarity coefficient, 0.749 for precision, and 0.756 for recall. Our network demonstrated superior classification performance to previous methods based on PRs, and had comparable segmentation performance to studies based on other imaging modalities. Therefore, CACSNet can be used for robust classification and segmentation of CAC lesions that are morphologically variable and overlap with surrounding structures over the entire posterior inferior region of the mandibular angle on PRs.

PMID:38886356 | DOI:10.1038/s41598-024-64265-4

J Imaging Inform Med. 2024 Jun 17. doi: 10.1007/s10278-023-00919-5. Online ahead of print.

ABSTRACT

Deep learning has significantly advanced the field of radiology-based disease diagnosis, offering enhanced accuracy and efficiency in detecting various medical conditions through the analysis of complex medical images such as X-rays. This technology's ability to discern subtle patterns and anomalies has proven invaluable for swift and accurate disease identification. The relevance of deep learning in radiology has been particularly highlighted during the COVID-19 pandemic, where rapid and accurate diagnosis is crucial for effective treatment and containment. However, recent research has uncovered vulnerabilities in deep learning models when exposed to adversarial attacks, leading to incorrect predictions. In response to this critical challenge, we introduce a novel approach that leverages total variation minimization to combat adversarial noise within X-ray images effectively. Our focus narrows to COVID-19 diagnosis as a case study, where we initially construct a classification model through transfer learning designed to accurately classify lung X-ray images encompassing no pneumonia, COVID-19 pneumonia, and non-COVID pneumonia cases. Subsequently, we extensively evaluated the model's susceptibility to targeted and un-targeted adversarial attacks by employing the fast gradient sign gradient (FGSM) method. Our findings reveal a substantial reduction in the model's performance, with the average accuracy plummeting from 95.56 to 19.83% under adversarial conditions. However, the experimental results demonstrate the exceptional efficacy of the proposed denoising approach in enhancing the performance of diagnosis models when applied to adversarial examples. Post-denoising, the model exhibits a remarkable accuracy improvement, surging from 19.83 to 88.23% on adversarial images. These promising outcomes underscore the potential of denoising techniques to fortify the resilience and reliability of AI-based COVID-19 diagnostic systems, laying the foundation for their successful deployment in clinical settings.

PMID:38886292 | DOI:10.1007/s10278-023-00919-5

J Imaging Inform Med. 2024 Jun 17. doi: 10.1007/s10278-024-01165-z. Online ahead of print.

ABSTRACT

Two significant obstacles hinder the advancement of Radiology AI. The first is the challenge of overfitting, where small training data sets can result in unreliable outcomes. The second challenge is the need for more generalizability, the lack of which creates difficulties in implementing the technology across various institutions and practices. A recent innovation, deep neuroevolution (DNE), has been introduced to tackle the overfitting issue by training on small data sets and producing accurate predictions. However, the generalizability of DNE has yet to be proven. This paper strives to overcome this barrier by demonstrating that DNE can achieve satisfactory results in diverse external validation sets. The main innovation of the work is thus showing that DNE can generalize to varied outside data. Our example use case is predicting brain metastasis from neuroblastoma, emphasizing the importance of AI with limited data sets. Despite image collection and labeling advancements, rare diseases will always constrain data availability. We optimized a convolutional neural network (CNN) with DNE to demonstrate generalizability. We trained the CNN with 60 MRI images and tested it on a separate diverse collection of images from over 50 institutions. For comparison, we also trained with the more traditional stochastic gradient descent (SGD) method, with the two variants of (1) training from scratch and (2) transfer learning. Our results show that DNE demonstrates excellent generalizability with 97% accuracy on the heterogeneous testing set, while neither form of SGD could reach 60% accuracy. DNE's ability to generalize from small training sets to external and diverse testing sets suggests that it or similar approaches may play an integral role in improving the clinical performance of AI.

PMID:38886289 | DOI:10.1007/s10278-024-01165-z

Insights Imaging. 2024 Jun 18;15(1):150. doi: 10.1186/s13244-024-01733-5.

ABSTRACT

OBJECTIVES: Synchronous colorectal cancer peritoneal metastasis (CRPM) has a poor prognosis. This study aimed to create a radiomics-boosted deep learning model by PET/CT image for risk assessment of synchronous CRPM.

METHODS: A total of 220 colorectal cancer (CRC) cases were enrolled in this study. We mapped the feature maps (Radiomic feature maps (RFMs)) of radiomic features across CT and PET image patches by a 2D sliding kernel. Based on ResNet50, a radiomics-boosted deep learning model was trained using PET/CT image patches and RFMs. Besides that, we explored whether the peritumoral region contributes to the assessment of CRPM. In this study, the performance of each model was evaluated by the area under the curves (AUC).

RESULTS: The AUCs of the radiomics-boosted deep learning model in the training, internal, external, and all validation datasets were 0.926 (95% confidence interval (CI): 0.874-0.978), 0.897 (95% CI: 0.801-0.994), 0.885 (95% CI: 0.795-0.975), and 0.889 (95% CI: 0.823-0.954), respectively. This model exhibited consistency in the calibration curve, the Delong test and IDI identified it as the most predictive model.

CONCLUSIONS: The radiomics-boosted deep learning model showed superior estimated performance in preoperative prediction of synchronous CRPM from pre-treatment PET/CT, offering potential assistance in the development of more personalized treatment methods and follow-up plans.

CRITICAL RELEVANCE STATEMENT: The onset of synchronous colorectal CRPM is insidious, and using a radiomics-boosted deep learning model to assess the risk of CRPM before treatment can help make personalized clinical treatment decisions or choose more sensitive follow-up plans.

KEY POINTS: Prognosis for patients with CRPM is bleak, and early detection poses challenges. The synergy between radiomics and deep learning proves advantageous in evaluating CRPM. The radiomics-boosted deep-learning model proves valuable in tailoring treatment approaches for CRC patients.

PMID:38886244 | DOI:10.1186/s13244-024-01733-5

Clin Oral Investig. 2024 Jun 18;28(7):381. doi: 10.1007/s00784-024-05781-5.

ABSTRACT

OBJECTIVES: Tooth extraction is one of the most frequently performed medical procedures. The indication is based on the combination of clinical and radiological examination and individual patient parameters and should be made with great care. However, determining whether a tooth should be extracted is not always a straightforward decision. Moreover, visual and cognitive pitfalls in the analysis of radiographs may lead to incorrect decisions. Artificial intelligence (AI) could be used as a decision support tool to provide a score of tooth extractability.

MATERIAL AND METHODS: Using 26,956 single teeth images from 1,184 panoramic radiographs (PANs), we trained a ResNet50 network to classify teeth as either extraction-worthy or preservable. For this purpose, teeth were cropped with different margins from PANs and annotated. The usefulness of the AI-based classification as well that of dentists was evaluated on a test dataset. In addition, the explainability of the best AI model was visualized via a class activation mapping using CAMERAS.

RESULTS: The ROC-AUC for the best AI model to discriminate teeth worthy of preservation was 0.901 with 2% margin on dental images. In contrast, the average ROC-AUC for dentists was only 0.797. With a 19.1% tooth extractions prevalence, the AI model's PR-AUC was 0.749, while the dentist evaluation only reached 0.589.

CONCLUSION: AI models outperform dentists/specialists in predicting tooth extraction based solely on X-ray images, while the AI performance improves with increasing contextual information.

CLINICAL RELEVANCE: AI could help monitor at-risk teeth and reduce errors in indications for extractions.

PMID:38886242 | DOI:10.1007/s00784-024-05781-5

Brief Bioinform. 2024 May 23;25(4):bbae284. doi: 10.1093/bib/bbae284.

ABSTRACT

Morphological profiling is a valuable tool in phenotypic drug discovery. The advent of high-throughput automated imaging has enabled the capturing of a wide range of morphological features of cells or organisms in response to perturbations at the single-cell resolution. Concurrently, significant advances in machine learning and deep learning, especially in computer vision, have led to substantial improvements in analyzing large-scale high-content images at high throughput. These efforts have facilitated understanding of compound mechanism of action, drug repurposing, characterization of cell morphodynamics under perturbation, and ultimately contributing to the development of novel therapeutics. In this review, we provide a comprehensive overview of the recent advances in the field of morphological profiling. We summarize the image profiling analysis workflow, survey a broad spectrum of analysis strategies encompassing feature engineering- and deep learning-based approaches, and introduce publicly available benchmark datasets. We place a particular emphasis on the application of deep learning in this pipeline, covering cell segmentation, image representation learning, and multimodal learning. Additionally, we illuminate the application of morphological profiling in phenotypic drug discovery and highlight potential challenges and opportunities in this field.

PMID:38886164 | DOI:10.1093/bib/bbae284

JCO Clin Cancer Inform. 2024 Jun;8:e2300166. doi: 10.1200/CCI.23.00166.

ABSTRACT

PURPOSE: The RECIST guidelines provide a standardized approach for evaluating the response of cancer to treatment, allowing for consistent comparison of treatment efficacy across different therapies and patients. However, collecting such information from electronic health records manually can be extremely labor-intensive and time-consuming because of the complexity and volume of clinical notes. The aim of this study is to apply natural language processing (NLP) techniques to automate this process, minimizing manual data collection efforts, and improving the consistency and reliability of the results.

METHODS: We proposed a complex, hybrid NLP system that automates the process of extracting, linking, and summarizing anticancer therapy and associated RECIST-like responses from narrative clinical text. The system consists of multiple machine learning-/deep learning-based and rule-based modules for diverse NLP tasks such as named entity recognition, assertion classification, relation extraction, and text normalization, to address different challenges associated with anticancer therapy and response information extraction. We then evaluated the system performances on two independent test sets from different institutions to demonstrate its effectiveness and generalizability.

RESULTS: The system used domain-specific language models, BioBERT and BioClinicalBERT, for high-performance therapy mentions identification and RECIST responses extraction and categorization. The best-performing model achieved a 0.66 score in linking therapy and RECIST response mentions, with end-to-end performance peaking at 0.74 after relation normalization, indicating substantial efficacy with room for improvement.

CONCLUSION: We developed, implemented, and tested an information extraction system from clinical notes for cancer treatment and efficacy assessment information. We expect this system will support future cancer research, particularly oncologic studies that focus on efficiently assessing the effectiveness and reliability of cancer therapeutics.

PMID:38885475 | DOI:10.1200/CCI.23.00166

Bioinformatics. 2024 Jun 17:btae378. doi: 10.1093/bioinformatics/btae378. Online ahead of print.

ABSTRACT

MOTIVATION: ADP-ribosylation is a critical modification involved in regulating diverse cellular processes, including chromatin structure regulation, RNA transcription, and cell death. Bacterial ADP-ribosyltransferase toxins (bARTTs) serve as potent virulence factors that orchestrate the manipulation of host cell functions to facilitate bacterial pathogenesis. Despite their pivotal role, the bioinformatic identification of novel bARTTs poses a formidable challenge due to limited verified data and the inherent sequence diversity among bARTT members.

RESULTS: We proposed a deep learning-based model, ARTNet, specifically engineered to predict bARTTs from bacterial genomes. Initially, we introduced an effective data augmentation method to address the issue of data scarcity in training ARTNet. Subsequently, we employed a data optimization strategy by utilizing ART-related domain subsequences instead of the primary full sequences, thereby significantly enhancing the performance of ARTNet. ARTNet achieved a Matthew's correlation coefficient (MCC) of 0.9351 and an F1-score (macro) of 0.9666 on repeated independent test datasets, outperforming three other deep learning models and six traditional machine learning models in terms of time efficiency and accuracy. Furthermore, we empirically demonstrated the ability of ARTNet to predict novel bARTTs across domain superfamilies without sequence similarity. We anticipate that ARTNet will greatly facilitate the screening and identification of novel bARTTs from bacterial genomes.

AVAILABILITY: ARTNet is publicly accessible at http://www.mgc.ac.cn/ARTNet/. The source code of ARTNet is freely available at https://github.com/zhengdd0422/ARTNet/.

SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

PMID:38885365 | DOI:10.1093/bioinformatics/btae378

PLoS One. 2024 Jun 17;19(6):e0292026. doi: 10.1371/journal.pone.0292026. eCollection 2024.

ABSTRACT

Machine learning (ML) and deep learning (DL) models are being increasingly employed for medical imagery analyses, with both approaches used to enhance the accuracy of classification/prediction in the diagnoses of various cancers, tumors and bloodborne diseases. To date however, no review of these techniques and their application(s) within the domain of white blood cell (WBC) classification in blood smear images has been undertaken, representing a notable knowledge gap with respect to model selection and comparison. Accordingly, the current study sought to comprehensively identify, explore and contrast ML and DL methods for classifying WBCs. Following development and implementation of a formalized review protocol, a cohort of 136 primary studies published between January 2006 and May 2023 were identified from the global literature, with the most widely used techniques and best-performing WBC classification methods subsequently ascertained. Studies derived from 26 countries, with highest numbers from high-income countries including the United States (n = 32) and The Netherlands (n = 26). While WBC classification was originally rooted in conventional ML, there has been a notable shift toward the use of DL, and particularly convolutional neural networks (CNN), with 54.4% of identified studies (n = 74) including the use of CNNs, and particularly in concurrence with larger datasets and bespoke features e.g., parallel data pre-processing, feature selection, and extraction. While some conventional ML models achieved up to 99% accuracy, accuracy was shown to decrease in concurrence with decreasing dataset size. Deep learning models exhibited improved performance for more extensive datasets and exhibited higher levels of accuracy in concurrence with increasingly large datasets. Availability of appropriate datasets remains a primary challenge, potentially resolvable using data augmentation techniques. Moreover, medical training of computer science researchers is recommended to improve current understanding of leucocyte structure and subsequent selection of appropriate classification models. Likewise, it is critical that future health professionals be made aware of the power, efficacy, precision and applicability of computer science, soft computing and artificial intelligence contributions to medicine, and particularly in areas like medical imaging.

PMID:38885231 | DOI:10.1371/journal.pone.0292026

PLoS One. 2024 Jun 17;19(6):e0304798. doi: 10.1371/journal.pone.0304798. eCollection 2024.

ABSTRACT

Drug-drug interaction (DDI) is the combined effects of multiple drugs taken together, which can either enhance or reduce each other's efficacy. Thus, drug interaction analysis plays an important role in improving treatment effectiveness and patient safety. It has become a new challenge to use computational methods to accelerate drug interaction time and reduce its cost-effectiveness. The existing methods often do not fully explore the relationship between the structural information and the functional information of drug molecules, resulting in low prediction accuracy for drug interactions, poor generalization, and other issues. In this paper, we propose a novel method, which is a deep graph contrastive learning model for drug-drug interaction prediction (DeepGCL for brevity). DeepGCL incorporates a contrastive learning component to enhance the consistency of information between different views (molecular structure and interaction network), which means that the DeepGCL model predicts drug interactions by integrating molecular structure features and interaction network topology features. Experimental results show that DeepGCL achieves better performance than other methods in all datasets. Moreover, we conducted many experiments to analyze the necessity of each component of the model and the robustness of the model, which also showed promising results. The source code of DeepGCL is freely available at https://github.com/jzysj/DeepGCL.

PMID:38885206 | DOI:10.1371/journal.pone.0304798

IEEE Trans Pattern Anal Mach Intell. 2024 Jun 17;PP. doi: 10.1109/TPAMI.2024.3415387. Online ahead of print.

ABSTRACT

Deep learning-based solutions have achieved impressive performance in semantic segmentation but often require large amounts of training data with fine-grained annotations. To alleviate such requisition, a variety of weakly supervised annotation strategies have been proposed, among which scribble supervision is emerging as a popular one due to its user-friendly annotation way. However, the sparsity and diversity of scribble annotations make it nontrivial to train a network to produce deterministic and consistent predictions directly. To address these issues, in this paper we propose holistic solutions involving the design of network structure, loss and training procedure, named CC4S to improve Certainty and Consistency for Scribble-Supervised Semantic Segmentation. Specifically, to reduce uncertainty, CC4S embeds a random walk module into the network structure to make neural representations uniformly distributed within similar semantic regions, which works together with a soft entropy loss function to force the network to produce deterministic predictions. To encourage consistency, CC4S adopts self-supervision training and imposes the consistency loss on the eigenspace of the probability transition matrix in the random walk module (we named neural eigenspace). Such self-supervision inherits the category-level discriminability from the neural eigenspace and meanwhile helps the network focus on producing consistent predictions for the salient parts and neglect semantically heterogeneous backgrounds. Finally, to further improve the performance, CC4S uses the network predictions as pseudo-labels and retrains the network with an extra color constraint regularizer on pseudo-labels to boost semantic consistency in color space. Rich experiments demonstrate the excellent performance of CC4S. In particular, under scribble supervision, CC4S achieves comparable performance to those from fully supervised methods. Comprehensive ablation experiments verify the effectiveness of the design choices in CC4S and its robustness under extreme supervision cases, i.e., when scribbles are shrunk proportionally or dropped randomly. The code for this work has been open-sourced at https://github.com/panzhiyi/CC4S.

PMID:38885110 | DOI:10.1109/TPAMI.2024.3415387

IEEE J Biomed Health Inform. 2024 Jun 17;PP. doi: 10.1109/JBHI.2024.3415479. Online ahead of print.

ABSTRACT

Cough is an important symptom in children with acute and chronic respiratory disease. Daily cough is common in Cystic Fibrosis (CF) and increased cough is a symptom of pulmonary exacerbation. To date, cough assessment is primarily subjective in clinical practice and research. Attempts to develop objective, automatic cough counting tools have faced reliability issues in noisy environments and practical barriers limiting long-term use. This single-center pilot study evaluated usability, acceptability and performance of a mechanoacoustic sensor (MAS), previously used for cough classification in adults, in 36 children with CF over brief and multi-day periods in four cohorts. Children whose health was at baseline and who had symptoms of pulmonary exacerbation were included. We trained, validated, and deployed custom deep learning algorithms for accurate cough detection and classification from other vocalization or artifacts with an overall area under the receiver-operator characteristic curve (AUROC) of 0.96 and average precision (AP) of 0.93. Child and parent feedback led to a redesign of the MAS towards a smaller, more discreet device acceptable for daily use in children. Additional improvements optimized power efficiency and data management. The MAS's ability to objectively measure cough and other physiologic signals across clinic, hospital, and home settings is demonstrated, particularly aided by an AUROC of 0.97 and AP of 0.96 for motion artifact rejection. Examples of cough frequency and physiologic parameter correlations with participant-reported outcomes and clinical measurements for individual patients are presented. The MAS is a promising tool in objective longitudinal evaluation of cough in children with CF.

PMID:38885105 | DOI:10.1109/JBHI.2024.3415479

IEEE J Biomed Health Inform. 2024 Jun 17;PP. doi: 10.1109/JBHI.2024.3415000. Online ahead of print.

ABSTRACT

Learning-based methods offer performance leaps over traditional methods in classification analysis of high-dimensional functional MRI (fMRI) data. In this domain, deep-learning models that analyze functional connectivity (FC) features among brain regions have been particularly promising. However, many existing models receive as input temporally static FC features that summarize inter-regional interactions across an entire scan, reducing the temporal sensitivity of classifiers by limiting their ability to leverage information on dynamic FC features of brain activity. To improve the performance of baseline classification models without compromising efficiency, here we propose a novel plug-in based on a graph neural network, GraphCorr, to provide enhanced input features to baseline models. The proposed plug-in computes a set of latent FC features with enhanced temporal information while maintaining comparable dimensionality to static features. Taking brain regions as nodes and blood-oxygen-level-dependent (BOLD) signals as node inputs, GraphCorr leverages a node embedder module based on a transformer encoder to capture dynamic latent representations of BOLD signals. GraphCorr also leverages a lag filter module to account for delayed interactions across nodes by learning correlational features of windowed BOLD signals across time delays. These two feature groups are then fused via a message passing algorithm executed on the formulated graph. Comprehensive demonstrations on three public datasets indicate improved classification performance for several state-of-the-art graph and convolutional baseline models when they are augmented with GraphCorr.

PMID:38885104 | DOI:10.1109/JBHI.2024.3415000