Magn Reson Med. 2024 Feb 11. doi: 10.1002/mrm.30021. Online ahead of print.

ABSTRACT

PURPOSE: We introduced a novel reconstruction network, jointly unrolled cross-domain optimization-based spatio-temporal reconstruction network (JUST-Net), aimed at accelerating 3D multi-echo gradient-echo (mGRE) data acquisition and improving the quality of resulting myelin water imaging (MWI) maps.

METHOD: An unrolled cross-domain spatio-temporal reconstruction network was designed. The main idea is to combine frequency and spatio-temporal image feature representations and to sequentially implement convolution layers in both domains. The k-space subnetwork utilizes shared information from adjacent frames, whereas the image subnetwork applies separate convolutions in both spatial and temporal dimensions. The proposed reconstruction network was evaluated for both retrospectively and prospectively accelerated acquisition. Furthermore, it was assessed in simulation studies and real-world cases with k-space corruptions to evaluate its potential for motion artifact reduction.

RESULTS: The proposed JUST-Net enabled highly reproducible and accelerated 3D mGRE acquisition for whole-brain MWI, reducing the acquisition time from fully sampled 15:23 to 2:22 min within a 3-min reconstruction time. The normalized root mean squared error of the reconstructed mGRE images increased by less than 4.0%, and the correlation coefficients for MWI showed a value of over 0.68 when compared to the fully sampled reference. Additionally, the proposed method demonstrated a mitigating effect on both simulated and clinical motion-corrupted cases.

CONCLUSION: The proposed JUST-Net has demonstrated the capability to achieve high acceleration factors for 3D mGRE-based MWI, which is expected to facilitate widespread clinical applications of MWI.

PMID:38342983 | DOI:10.1002/mrm.30021

Small. 2024 Feb 11:e2311036. doi: 10.1002/smll.202311036. Online ahead of print.

ABSTRACT

Wearable devices play an indispensable role in modern life, and the human body contains multiple wasted energies available for wearable devices. This study proposes a self-sensing and self-powered wearable system (SS-WS) based on scavenging waist motion energy and knee negative energy. The proposed SS-WS consists of a three-degree-of-freedom triboelectric nanogenerator (TDF-TENG) and a negative energy harvester (NEH). The TDF-TENG is driven by waist motion energy and the generated triboelectric signals are processed by deep learning for recognizing the human motion. The triboelectric signals generated by TDF-TENG can accurately recognize the motion state after processing based on Gate Recurrent Unit deep learning model. With double frequency up-conversion, the NEH recovers knee negative energy generation for powering wearable devices. A model wearing the single energy harvester can generate the power of 27.01 mW when the movement speed is 8 km h-1 , and the power density of NEH reaches 0.3 W kg-1 at an external excitation condition of 3 Hz. Experiments and analysis prove that the proposed SS-WS can realize self-sensing and effectively power wearable devices.

PMID:38342584 | DOI:10.1002/smll.202311036

J Acoust Soc Am. 2024 Feb 1;155(2):1198-1215. doi: 10.1121/10.0024353.

ABSTRACT

Speech and language development are early indicators of overall analytical and learning ability in children. The preschool classroom is a rich language environment for monitoring and ensuring growth in young children by measuring their vocal interactions with teachers and classmates. Early childhood researchers are naturally interested in analyzing naturalistic vs controlled lab recordings to measure both quality and quantity of such interactions. Unfortunately, present-day speech technologies are not capable of addressing the wide dynamic scenario of early childhood classroom settings. Due to the diversity of acoustic events/conditions in such daylong audio streams, automated speaker diarization technology would need to be advanced to address this challenging domain for segmenting audio as well as information extraction. This study investigates alternate deep learning-based lightweight, knowledge-distilled, diarization solutions for segmenting classroom interactions of 3-5 years old children with teachers. In this context, the focus on speech-type diarization which classifies speech segments as being either from adults or children partitioned across multiple classrooms. Our lightest CNN model achieves a best F1-score of ∼76.0% on data from two classrooms, based on dev and test sets of each classroom. It is utilized with automatic speech recognition-based re-segmentation modules to perform child-adult diarization. Additionally, F1-scores are obtained for individual segments with corresponding speaker tags (e.g., adult vs child), which provide knowledge for educators on child engagement through naturalistic communications. The study demonstrates the prospects of addressing educational assessment needs through communication audio stream analysis, while maintaining both security and privacy of all children and adults. The resulting child communication metrics have been used for broad-based feedback for teachers with the help of visualizations.

PMID:38341746 | DOI:10.1121/10.0024353

J Acoust Soc Am. 2024 Feb 1;155(2):1048-1059. doi: 10.1121/10.0024750.

ABSTRACT

A method is presented for estimating and reconstructing the sound field within a room using physics-informed neural networks. By incorporating a limited set of experimental room impulse responses as training data, this approach combines neural network processing capabilities with the underlying physics of sound propagation, as articulated by the wave equation. The network's ability to estimate particle velocity and intensity, in addition to sound pressure, demonstrates its capacity to represent the flow of acoustic energy and completely characterise the sound field with only a few measurements. Additionally, an investigation into the potential of this network as a tool for improving acoustic simulations is conducted. This is due to its proficiency in offering grid-free sound field mappings with minimal inference time. Furthermore, a study is carried out which encompasses comparative analyses against current approaches for sound field reconstruction. Specifically, the proposed approach is evaluated against both data-driven techniques and elementary wave-based regression methods. The results demonstrate that the physics-informed neural network stands out when reconstructing the early part of the room impulse response, while simultaneously allowing for complete sound field characterisation in the time domain.

PMID:38341739 | DOI:10.1121/10.0024750

Eur J Pharm Sci. 2024 Feb 8:106724. doi: 10.1016/j.ejps.2024.106724. Online ahead of print.

ABSTRACT

BACKGROUND: Recent studies, based on clinical data, have identified sex and age as significant factors associated with an increased risk of long COVID. These two factors align with the two post-COVID-19 clusters identified by a deep learning algorithm in computed tomography (CT) lung scans: Cluster 1 (C1), comprising predominantly females with small airway diseases, and Cluster 2 (C2), characterized by older individuals with fibrotic-like patterns. This study aims to assess the distributions of inhaled aerosols in these clusters.

METHODS: 140 COVID survivors examined around 112 days post-diagnosis, along with 105 uninfected, non-smoking healthy controls, were studied. Their demographic data and CT scans at full inspiration and expiration were analyzed using a combined imaging and modeling approach. A subject-specific CT-based computational model analysis was utilized to predict airway resistance and particle deposition among C1 and C2 subjects. The cluster-specific structure and function relationships were explored.

RESULTS: In C1 subjects, distinctive features included airway narrowing, a reduced homothety ratio of daughter over parent branch diameter, and increased airway resistance. Airway resistance was concentrated in the distal region, with a higher fraction of particle deposition in the proximal airways. On the other hand, C2 subjects exhibited airway dilation, an increased homothety ratio, reduced airway resistance, and a shift of resistance concentration towards the proximal region, allowing for deeper particle penetration into the lungs.

CONCLUSIONS: This study revealed unique mechanistic phenotypes of airway resistance and particle deposition in the two post-COVID-19 clusters. The implications of these findings for inhaled drug delivery effectiveness and susceptibility to air pollutants were explored.

PMID:38340875 | DOI:10.1016/j.ejps.2024.106724

Cell Rep Med. 2024 Feb 2:101419. doi: 10.1016/j.xcrm.2024.101419. Online ahead of print.

ABSTRACT

Federated learning (FL) is a distributed machine learning framework that is gaining traction in view of increasing health data privacy protection needs. By conducting a systematic review of FL applications in healthcare, we identify relevant articles in scientific, engineering, and medical journals in English up to August 31st, 2023. Out of a total of 22,693 articles under review, 612 articles are included in the final analysis. The majority of articles are proof-of-concepts studies, and only 5.2% are studies with real-life application of FL. Radiology and internal medicine are the most common specialties involved in FL. FL is robust to a variety of machine learning models and data types, with neural networks and medical imaging being the most common, respectively. We highlight the need to address the barriers to clinical translation and to assess its real-world impact in this new digital data-driven healthcare scene.

PMID:38340728 | DOI:10.1016/j.xcrm.2024.101419

Phys Med. 2024 Feb 9;119:103304. doi: 10.1016/j.ejmp.2024.103304. Online ahead of print.

ABSTRACT

Precise delineation of brain glioblastoma or tumor through segmentation is pivotal in the diagnosis, formulating treatment strategies, and evaluating therapeutic progress in patients. Precisely identifying brain glioblastoma within multimodal MRI scans poses a significant challenge in the field of medical image analysis as different intensity profiles are observed across the sub-regions, reflecting diverse tumor biological properties. For segmenting glioblastoma areas, convolutional neural networks have displayed astounding performance in recent years. This paper introduces an innovative methodology for brain glioblastoma segmentation by combining the Dense-Attention 3D U-Net network with a fusion strategy and the focal tversky loss function. By fusing information from multiple resolution segmentation maps, our model enhances its ability to discern intricate tumor boundaries. Incorporating the focal tversky loss function, we effectively emphasize critical regions and mitigate class imbalance. Recursive Convolution Block 2 is proposed after fusion to ensure efficient utilization of all accessible features while maintaining rapid convergence. The network's effectiveness is assessed using diverse datasets BraTS 2020 and BraTS 2021. Results show comparable dice similarity coefficient compared to other methods with increased efficiency and segmentation performance. Additionally, the architecture achieved an average dice similarity coefficient of 82.4% and an average hausdorff distance (HD95) of 10.426, which demonstrated consistent performance improvement compared to baseline models like U-Net, Attention U-Net, V-Net and Res U-Net and indicating the effectiveness of proposed architecture.

PMID:38340694 | DOI:10.1016/j.ejmp.2024.103304

Comput Med Imaging Graph. 2024 Feb 9;113:102356. doi: 10.1016/j.compmedimag.2024.102356. Online ahead of print.

ABSTRACT

The extraction of abdominal structures using deep learning has recently experienced a widespread interest in medical image analysis. Automatic abdominal organ and vessel segmentation is highly desirable to guide clinicians in computer-assisted diagnosis, therapy, or surgical planning. Despite a good ability to extract large organs, the capacity of U-Net inspired architectures to automatically delineate smaller structures remains a major issue, especially given the increase in receptive field size as we go deeper into the network. To deal with various abdominal structure sizes while exploiting efficient geometric constraints, we present a novel approach that integrates into deep segmentation shape priors from a semi-overcomplete convolutional auto-encoder (S-OCAE) embedding. Compared to standard convolutional auto-encoders (CAE), it exploits an over-complete branch that projects data onto higher dimensions to better characterize anatomical structures with a small spatial extent. Experiments on abdominal organs and vessel delineation performed on various publicly available datasets highlight the effectiveness of our method compared to state-of-the-art, including U-Net trained without and with shape priors from a traditional CAE. Exploiting a semi-overcomplete convolutional auto-encoder embedding as shape priors improves the ability of deep segmentation models to provide realistic and accurate abdominal structure contours.

PMID:38340573 | DOI:10.1016/j.compmedimag.2024.102356

BMC Oral Health. 2024 Feb 10;24(1):212. doi: 10.1186/s12903-024-03993-5.

ABSTRACT

BACKGROUND: Oral cancer is a life-threatening malignancy, which affects the survival rate and quality of life of patients. The aim of this systematic review was to review deep learning (DL) studies in the diagnosis and prognostic prediction of oral cancer.

METHODS: This systematic review was conducted following the PRISMA guidelines. Databases (Medline via PubMed, Google Scholar, Scopus) were searched for relevant studies, from January 2000 to June 2023.

RESULTS: Fifty-four qualified for inclusion, including diagnostic (n = 51), and prognostic prediction (n = 3). Thirteen studies showed a low risk of biases in all domains, and 40 studies low risk for concerns regarding applicability. The performance of DL models was reported of the accuracy of 85.0-100%, F1-score of 79.31 - 89.0%, Dice coefficient index of 76.0 - 96.3% and Concordance index of 0.78-0.95 for classification, object detection, segmentation, and prognostic prediction, respectively. The pooled diagnostic odds ratios were 2549.08 (95% CI 410.77-4687.39) for classification studies.

CONCLUSIONS: The number of DL studies in oral cancer is increasing, with a diverse type of architectures. The reported accuracy showed promising DL performance in studies of oral cancer and appeared to have potential utility in improving informed clinical decision-making of oral cancer.

PMID:38341571 | DOI:10.1186/s12903-024-03993-5

NPJ Digit Med. 2024 Feb 10;7(1):32. doi: 10.1038/s41746-024-01033-8.

NO ABSTRACT

PMID:38341516 | DOI:10.1038/s41746-024-01033-8

Sci Data. 2024 Feb 10;11(1):184. doi: 10.1038/s41597-024-03005-9.

ABSTRACT

Fluorescent Neuronal Cells v2 is a collection of fluorescence microscopy images and the corresponding ground-truth annotations, designed to foster innovative research in the domains of Life Sciences and Deep Learning. This dataset encompasses three image collections wherein rodent neuronal cell nuclei and cytoplasm are stained with diverse markers to highlight their anatomical or functional characteristics. Specifically, we release 1874 high-resolution images alongside 750 corresponding ground-truth annotations for several learning tasks, including semantic segmentation, object detection and counting. The contribution is two-fold. First, thanks to the variety of annotations and their accessible formats, we anticipate our work will facilitate methodological advancements in computer vision approaches for segmentation, detection, feature extraction, unsupervised and self-supervised learning, transfer learning, and related areas. Second, by enabling extensive exploration and benchmarking, we hope Fluorescent Neuronal Cells v2 will catalyze breakthroughs in fluorescence microscopy analysis and promote cutting-edge discoveries in life sciences.

PMID:38341463 | DOI:10.1038/s41597-024-03005-9

Nat Commun. 2024 Feb 10;15(1):1253. doi: 10.1038/s41467-024-45589-1.

ABSTRACT

Deep Learning (DL) can predict biomarkers from cancer histopathology. Several clinically approved applications use this technology. Most approaches, however, predict categorical labels, whereas biomarkers are often continuous measurements. We hypothesize that regression-based DL outperforms classification-based DL. Therefore, we develop and evaluate a self-supervised attention-based weakly supervised regression method that predicts continuous biomarkers directly from 11,671 images of patients across nine cancer types. We test our method for multiple clinically and biologically relevant biomarkers: homologous recombination deficiency score, a clinically used pan-cancer biomarker, as well as markers of key biological processes in the tumor microenvironment. Using regression significantly enhances the accuracy of biomarker prediction, while also improving the predictions' correspondence to regions of known clinical relevance over classification. In a large cohort of colorectal cancer patients, regression-based prediction scores provide a higher prognostic value than classification-based scores. Our open-source regression approach offers a promising alternative for continuous biomarker analysis in computational pathology.

PMID:38341402 | DOI:10.1038/s41467-024-45589-1

Ultrasound Med Biol. 2024 Feb 9:S0301-5629(23)00417-9. doi: 10.1016/j.ultrasmedbio.2023.12.023. Online ahead of print.

ABSTRACT

OBJECTIVE: Valvular heart diseases (VHDs) pose a significant public health burden, and deciding the best treatment strategy necessitates accurate assessment of heart valve function. Transthoracic echocardiography (TTE) is the key modality to evaluate VHDs, but the lack of standardized quantitative measurements leads to subjective and time-consuming assessments. We aimed to use deep learning to automate the extraction of mitral valve (MV) leaflets and annular hinge points from echocardiograms of the MV, improving standardization and reducing workload in quantitative assessment of MV disease.

METHODS: We annotated the MV leaflets and annulus points in 2931 images from 127 patients. We propose an approach for segmenting the annotated features using Attention UNet with deep supervision and weight scheduling of the attention coefficients to enforce saliency surrounding the MV. The derived segmentation masks were used to extract quantitative biomarkers for specific MV leaflet scallops throughout the heart cycle.

RESULTS: Evaluation performance was summarized using a Dice score of 0.63 ± 0.14, annulus error of 3.64 ± 2.53 and leaflet angle error of 8.7 ± 8.3°. Leveraging Attention UNet with deep supervision robustness of clinically relevant metrics was improved compared with UNet, reducing standard deviations by 2.7° (angle error) and 0.73 mm (annulus error). We correctly identified cases of MV prolapse, cases of stenosis and healthy references from a clinical material using the derived biomarkers.

CONCLUSION: Robust deep learning segmentation and tracking of MV morphology and motion is possible by leveraging attention gates and deep supervision, and holds promise for enhancing VHD diagnosis and treatment monitoring.

PMID:38341361 | DOI:10.1016/j.ultrasmedbio.2023.12.023

J Affect Disord. 2024 Feb 8:S0165-0327(24)00332-X. doi: 10.1016/j.jad.2024.02.023. Online ahead of print.

ABSTRACT

The hypothalamus is a well-established core structure in the sleep-wake cycle. While previous studies have not consistently found whole hypothalamus volume changes in chronic insomnia disorder (CID), differences may exist at the smaller substructural level of the hypothalamic nuclei. The study aimed to investigate the differences in total and subfield hypothalamic volumes, between CID patients and healthy controls (HCs) in vivo, through an advanced deep learning-based automated segmentation tool. A total of 150 patients with CID and 155 demographically matched HCs underwent T1-weighted structural magnetic resonance scanning. We utilized FreeSurfer v7.2 for automated segmentation of the hypothalamus and its five nuclei. Additionally, correlation and causal mediation analyses were performed to investigate the association between hypothalamic volume changes, insomnia symptom severity, and hypothalamus-pituitary-adrenal (HPA) axis-related blood biomarkers. CID patients exhibited larger volumes in the right anterior inferior, left anterior superior, and left posterior subunits of the hypothalamus compared to HCs. Moreover, we observed a positive association between blood corticotropin-releasing hormone (CRH) levels and insomnia severity, with anterior inferior hypothalamus (a-iHyp) hypertrophy mediating this relationship. In conclusion, we found significant volume increases in several hypothalamic subfield regions in CID patients, highlighting the central role of the HPA axis in the pathophysiology of insomnia.

PMID:38341156 | DOI:10.1016/j.jad.2024.02.023

Behav Brain Res. 2024 Feb 8:114900. doi: 10.1016/j.bbr.2024.114900. Online ahead of print.

ABSTRACT

Alzheimer's disease is a progressive neurological disorder characterized by brain atrophy and cell death, leading to cognitive decline and impaired functioning. Previous research has primarily focused on using cross-sectional data for Alzheimer's disease identification, but analyzing longitudinal sequential MR images is crucial for improved diagnostic accuracy and understanding disease progression. However, existing deep learning models face challenges in learning spatial and temporal features from such data. To address these challenges, this study presents a novel hybrid DenseNet-BiLSTM method for Alzheimer's disease prediction using longitudinal MRI analysis. The proposed framework combines Convolutional DenseNet for spatial information extraction and joined BiLSTM layers for capturing temporal characteristics and relationships between longitudinal images at different time points. This approach overcomes issues like overfitting, vanishing gradients, and incomplete patient data. We evaluated the model on 684 longitudinal MRI images from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database, including normal controls, individuals with mild cognitive impairment, and Alzheimer's disease patients. The results demonstrate high classification accuracy, with 95.28% for AD/CN, 88.19% for NC/MCI, 83.51% for sMCI/pMCI, and 92.14% for MCI/AD. These findings highlight the substantial improvement in Alzheimer's disease diagnosis achieved through the utilization of longitudinal MRI images. The contributions of this study lie in both the deep learning and medical domains. In the deep learning domain, our hybrid framework effectively learns spatial and temporal features from longitudinal data, addressing the challenges associated with multi-dimensional and sequential time series data. In the medical domain, our study emphasizes the importance of analyzing baseline and longitudinal MR images for accurate diagnosis and understanding disease progression.

PMID:38341100 | DOI:10.1016/j.bbr.2024.114900

Biosystems. 2024 Feb 8:105142. doi: 10.1016/j.biosystems.2024.105142. Online ahead of print.

ABSTRACT

Single-cell analysis (SCA) improves the detection of cancer, the immune system, and chronic diseases from complicated biological processes. SCA techniques generate high-dimensional, innovative, and complex data, making traditional analysis difficult and impractical. In the different cell types, conventional cell sequencing methods have signal transformation and disease detection limitations. To overcome these challenges, various deep learning techniques (DL) have outperformed standard state-of-the-art computer algorithms in SCA techniques. This review discusses DL application in SCA and presents a detailed study on improving SCA data processing and analysis. Firstly, we introduced fundamental concepts and critical points of cell analysis techniques, which illustrate the application of SCA. Secondly, various effective DL strategies apply to SCA to analyze data and provide significant results from complex data sources. Finally, we explored DL as a future direction in SCA and highlighted new challenges and opportunities for the rapidly evolving field of single-cell omics.

PMID:38340976 | DOI:10.1016/j.biosystems.2024.105142

Magn Reson Imaging. 2024 Feb 8:S0730-725X(24)00038-9. doi: 10.1016/j.mri.2024.02.006. Online ahead of print.

ABSTRACT

PURPOSE: To assess the utility of deep learning (DL)-based image reconstruction with the combination of compressed sensing (CS) denoising cycle by comparing images reconstructed by conventional CS-based method without DL in fat-suppressed (Fs)-contrast enhanced (CE) three-dimensional (3D) T1-weighted images (T1WIs) of the head and neck.

MATERIALS AND METHODS: We retrospectively analyzed the cases of 39 patients who had undergone head and neck Fs-CE 3D T1WI applying reconstructions based on conventional CS and CS augmented by DL, respectively. In the qualitative assessment, we evaluated overall image quality, visualization of anatomical structures, degree of artifacts, lesion conspicuity, and lesion edge sharpness based on a five-point system. In the quantitative assessment, we calculated the signal-to-noise ratios (SNRs) of the lesion and the posterior neck muscle and the contrast-to-noise ratio (CNR) between the lesion and the adjacent muscle.

RESULTS: For all items of the qualitative analysis, significantly higher scores were awarded to images with DL-based reconstruction (p < 0.001). In the quantitative analysis, DL-based reconstruction resulted in significantly higher values for both the SNR of lesions (p < 0.001) and posterior neck muscles (p < 0.001). Significantly higher CNRs were also observed in images with DL-based reconstruction (p < 0.001).

CONCLUSION: DL-based image reconstruction integrating into the CS-based denoising cycle offered superior image quality compared to the conventional CS method. This technique will be useful for the assessment of patients with head and neck disease.

PMID:38340971 | DOI:10.1016/j.mri.2024.02.006

Int J Biol Macromol. 2024 Feb 8:129810. doi: 10.1016/j.ijbiomac.2024.129810. Online ahead of print.

ABSTRACT

The prevalence of metabolic diseases is increasing at a frightening rate year by year. The burgeoning development of deep learning enables drug design to be more efficient, selective, and structurally novel. The critical relevance of Histone deacetylase 11 (HDAC11) to the pathogenesis of several metabolic diseases makes it a promising drug target for curbing metabolic disorders. The present study aims to design new specific HDAC11 inhibitors for the treatment of metabolic diseases. Deep learning was performed to learn the properties of existing HDAC11 inhibitors and yield a novel compound library containing 23,122 molecules. Subsequently, the compound library was screened by ADMET properties, Lipinski & Veber rules, traditional machine classification models, and molecular docking, and 10 compounds were screened as candidate HDAC11 inhibitors. The stability of the 10 new molecules was further evaluated by deploying RMSD, RMSF, MM/GBSA, free energy landscape mapping, and PCA analysis in molecular dynamics simulations. As a result, ten compounds, Cpd_17556, Cpd_2184, Cpd_8907, Cpd_7771, Cpd_14959, Cpd_7108, Cpd_12383, Cpd_13153, Cpd_14500and Cpd_21811, were characterized as good HDAC11 inhibitors and are expected to be promising drug candidates for metabolic disorders, and further in vitro, in vivo and clinical trials to demonstrate in the future.

PMID:38340912 | DOI:10.1016/j.ijbiomac.2024.129810

Environ Sci Pollut Res Int. 2024 Feb 10. doi: 10.1007/s11356-024-32280-7. Online ahead of print.

ABSTRACT

The main goal of this research is the interpretability of deep learning (DL) model output (e.g., CNN and LSTM) used to map land susceptibility to subsidence hazard by means of different techniques. For this purpose, an inventory map of land subsidence (LS) is prepared based on fieldwork and a record of LS presence points, and 16 features controlling LS were mapped. Thereafter, 11 effective features controlling LS were identified by means of a particle swarm optimization (PSO) algorithm, which was then used as input in the CNN and LSTM predictive models. To address the inherent black box nature of DL models, six interpretation methods (feature interaction, permutation importance plot (PFIM), bar plot, SHapley Additive exPlanations (SHAP) main plot, heatmap plot, and waterfall plot) were used to interpret the predictive model outputs. Both models (CNN and LSTM) had AUC > 90 and therefore provided excellent accuracy for mapping LS hazard. According to the most accurate model-the CNN predictive model-the range from very low to very high hazard classes occupied 20%, 20%, 25%, 16.3%, and 18.7% of the study area, respectively. According to three plots (bar plot, SHAP main plot, and heatmap plot), which were constructed based on the SHAP value, distance from the well, GDR and DEM were identified as the three most important features with the highest impact on the DL model output. The results of the waterfall plot indicate two effective features consisting of distance from the well and coarse fragment, and two effective features comprising landuse and DEM, contributed negatively and positively to LS, respectively. Overall, these explanation techniques can address critical concerns with respect to the interpretability of sophisticated DL predictive models.

PMID:38340298 | DOI:10.1007/s11356-024-32280-7

Brain Imaging Behav. 2024 Feb 10. doi: 10.1007/s11682-024-00857-y. Online ahead of print.

ABSTRACT

While one can characterize mental health using questionnaires, such tools do not provide direct insight into the underlying biology. By linking approaches that visualize brain activity to questionnaires in the context of individualized prediction, we can gain new insights into the biology and behavioral aspects of brain health. Resting-state fMRI (rs-fMRI) can be used to identify biomarkers of these conditions and study patterns of abnormal connectivity. In this work, we estimate mental health quality for individual participants using static functional network connectivity (sFNC) data from rs-fMRI. The deep learning model uses the sFNC data as input to predict four categories of mental health quality and visualize the neural patterns indicative of each group. We used guided gradient class activation maps (guided Grad-CAM) to identify the most discriminative sFNC patterns. The effectiveness of this model was validated using the UK Biobank dataset, in which we showed that our approach outperformed four alternative models by 4-18% accuracy. The proposed model's performance evaluation yielded a classification accuracy of 76%, 78%, 88%, and 98% for the excellent, good, fair, and poor mental health categories, with poor mental health accuracy being the highest. The findings show distinct sFNC patterns across each group. The patterns associated with excellent mental health consist of the cerebellar-subcortical regions, whereas the most prominent areas in the poor mental health category are in the sensorimotor and visual domains. Thus the combination of rs-fMRI and deep learning opens a promising path for developing a comprehensive framework to evaluate and measure mental health. Moreover, this approach had the potential to guide the development of personalized interventions and enable the monitoring of treatment response. Overall this highlights the crucial role of advanced imaging modalities and deep learning algorithms in advancing our understanding and management of mental health.

PMID:38340285 | DOI:10.1007/s11682-024-00857-y