ACS Biomater Sci Eng. 2025 Jan 15. doi: 10.1021/acsbiomaterials.4c01503. Online ahead of print.

ABSTRACT

Mechanical properties of engineered connective tissues are critical for their success, yet modern sensors that measure physical qualities of tissues for quality control are invasive and destructive. The goal of this work was to develop a noncontact, nondestructive method to measure mechanical attributes of engineered skin substitutes during production without disturbing the sterile culture packaging. We optimized a digital holographic vibrometry (DHV) system to measure the mechanical behavior of Apligraf living cellular skin substitute through the clear packaging in multiple conditions: resting on solid agar as when the tissue is shipped, on liquid media in which it is grown, and freely suspended in air as occurs when the media is removed for feeding. We utilized full-field measurement to assess the complete surface deformation pattern to compare with vibration theory and found the patterns observed in air showed the closest behavior to theory. To simulate the effects of the actual culture dish geometry and the trilayer composition of the tissue on the porous membrane support, we employed finite element (FE) analysis. To simulate changes in thickness and stiffness that may occur with manufacturing process variations, we dried samples over time and observed measurable increases in the fundamental mode frequency which could be predicted by altering the thickness of the tissue layers in the FE model. However, quantitative estimates of the engineered tissue stiffness based on vibration theory are unrealistically high due to the signal being dominated by the stiff underlying membrane on which the tissue is cultured. Thus, although DHV is not able to specifically quantify the thickness or modulus or identify small spot defects, it has the potential to be used assess the overall properties of a tissue in-line and noninvasively for quality control.

PMID:39813060 | DOI:10.1021/acsbiomaterials.4c01503

J Speech Lang Hear Res. 2025 Jan 15:1-31. doi: 10.1044/2024_JSLHR-23-00740. Online ahead of print.

ABSTRACT

PURPOSE: In this review article, we present an extensive overview of recent developments in the area of dysarthric speech research. One of the key objectives of speech technology research is to improve the quality of life of its users, as evidenced by the focus of current research trends on creating inclusive conversational interfaces that cater to pathological speech, out of which dysarthric speech is an important example. Applications of speech technology research for dysarthric speech demand a clear understanding of the acoustics of dysarthric speech as well as of speech technologies, including machine learning and deep neural networks for speech processing.

METHOD: We review studies pertaining to speech technology and dysarthric speech. Specifically, we discuss dysarthric speech corpora, acoustic analysis, intelligibility assessment, and automatic speech recognition. We also delve into deep learning approaches for automatic assessment and recognition of dysarthric speech. Ethics committee or institutional review board did not apply to this study.

CONCLUSIONS: Overcoming the challenge of limited data and exploring new avenues in data collection, artificial intelligence-powered analysis and teletherapy hold immense potential for significant advancements in dysarthria research. To make longer and faster strides, researchers typically rely on existing research and data on a global scale. Therefore, it is imperative to consolidate the existing research and present it in a form that can serve as a basis for future work. In this review article, we have reviewed the contributions of speech technologists to the area of dysarthric speech with a focus on acoustic analysis, speech features, and techniques used. By focusing on the existing research and future directions, researchers can develop more effective tools and interventions to improve communication, quality of life, and overall well-being for people with dysarthria.

PMID:39813019 | DOI:10.1044/2024_JSLHR-23-00740

Clin Exp Nephrol. 2025 Jan 15. doi: 10.1007/s10157-024-02616-1. Online ahead of print.

ABSTRACT

BACKGROUND: Chronic kidney disease (CKD) represents a significant public health challenge, with rates consistently on the rise. Enhancing kidney function prediction could contribute to the early detection, prevention, and management of CKD in clinical practice. We aimed to investigate whether deep learning techniques, especially those suitable for processing missing values, can improve the accuracy of predicting future renal function compared to traditional statistical method, using the Japan Chronic Kidney Disease Database (J-CKD-DB), a nationwide multicenter CKD registry.

METHODS: From the J-CKD-DB-Ex, a prospective longitudinal study within the J-CKD-DB, we selected individuals who had at least two eGFR measurements recorded between 12 and 20 months apart (n = 22,929 CKD patients). We used the multiple linear regression model as a conventional statistical method, and the Feed Forward Neural Network (FFNN) and Gated Recurrent Unit (GRU)-D (decay) models as deep learning techniques. We compared the prediction accuracies of each model for future eGFR based on the existing data using the root mean square error (RMSE).

RESULTS: The RMSE values were 7.5 for multiple regression analysis, 7.9 for FFNN model, and 7.6 mL/min/1.73 m2 for GRU-D model. In the subgroup analysis according to CKD stages, lower RMSE values were observed in higher stages for all models.

CONCLUSION: Our result demonstrate the predictive accuracy of future eGFR based on the existing dataset in the J-CKD-DB-Ex. The accuracy was not improved by applying deep learning techniques compared to conventional statistical methods.

PMID:39813007 | DOI:10.1007/s10157-024-02616-1

Neuroradiology. 2025 Jan 15. doi: 10.1007/s00234-024-03536-3. Online ahead of print.

ABSTRACT

PURPOSE: We aimed to validate a clinically available artificial intelligence (AI) model to assist general radiologists in the detection of intracranial aneurysm (IA) in a multi-reader multi-case (MRMC) study, and to explore its performance in routine clinical settings.

METHODS: Two distinct cohorts of head CT angiography (CTA) data were assembled to validate an AI model. Cohort 1, comprising gold-standard consecutive CTA cases, was used in an MRMC study involving six board-certified general radiologists. Cohort 2, representing clinical CTA cases, was used to simulate a routine clinical setting. Following these evaluations, an algorithmic audit was conducted to identify any unusual or unexpected behaviors exhibited by the model.

RESULTS: Cohort 1 consisted of 131 CTA cases, while Cohort 2 included 515 CTA cases. In the MRMC study, the AI-assisted strategy demonstrated a significant improvement in aneurysm diagnostic performance, with the area under the receiver operating characteristic curve increasing from 0.815 (95%CI: 0.754-0.875) to 0.875 (95%CI: 0.831-0.921; p = 0.008). In the AI-based first-reader study, 60.4% of the CTA cases were identified as negative by the AI, with a high negative predictive value of 0.994 (95%CI: 0.977-0.999). The algorithmic audit highlighted two issues for improvement: the accurate detection of tiny aneurysms and the effective exclusion of false-positive lesions.

CONCLUSION: This study highlights the clinical utility of a high-performance AI model in detecting IAs, significantly improving general radiologists' diagnostic performance with the potential to reduce their workload in routine clinical practice. The algorithmic audit offers insights to guide the development and validation of future AI models.

PMID:39812775 | DOI:10.1007/s00234-024-03536-3

Neuroinformatics. 2025 Jan 15;23(1):8. doi: 10.1007/s12021-024-09697-z.

ABSTRACT

Intracranial atherosclerotic stenosis (ICAS) and intracranial aneurysms are prevalent conditions in the cerebrovascular system. ICAS causes a narrowing of the arterial lumen, thereby restricting blood flow, while aneurysms involve the ballooning of blood vessels. Both conditions can lead to severe outcomes, such as stroke or vessel rupture, which can be fatal. Early detection is crucial for effective intervention. In this study, we introduced a method that combines classical computer vision techniques with deep learning to detect intracranial aneurysms and ICAS in time-of-flight magnetic resonance angiography images. The process began with skull-stripping, followed by an affine transformation to align the images to a common atlas space. We then focused on the region of interest, including the circle of Willis, by cropping the relevant area. A segmentation algorithm was used to isolate the arteries, after which a patch-wise residual neural network was applied across the image. A voting mechanism was then employed to identify the presence of atrophies. Our method achieved accuracies of 76.5% for aneurysms and 82.4% for ICAS. Notably, when occlusions were not considered, the accuracy for ICAS detection improved to 85.7%. While the algorithm performed well for localized pathological findings, it was less effective at detecting occlusions, which involved long-range dependencies in the MRIs. This limitation was due to the architectural design of the patch-wise deep learning approach. Regardless, this can, in the future, be mitigated in a multi-scale patch-wise algorithm.

PMID:39812766 | DOI:10.1007/s12021-024-09697-z

Insights Imaging. 2025 Jan 15;16(1):20. doi: 10.1186/s13244-024-01865-8.

ABSTRACT

OBJECTIVE: To evaluate the feasibility of utilizing artificial intelligence (AI)-predicted biparametric MRI (bpMRI) image features for predicting the aggressiveness of prostate cancer (PCa).

MATERIALS AND METHODS: A total of 878 PCa patients from 4 hospitals were retrospectively collected, all of whom had pathological results after radical prostatectomy (RP). A pre-trained AI algorithm was used to select suspected PCa lesions and extract lesion features for model development. The study evaluated five prediction methods, including (1) A clinical-imaging model of clinical features and image features of suspected PCa lesions selected by AI algorithm, (2) the PIRADS category, (3) a conventional radiomics model, (4) a deep-learning bases radiomics model, and (5) biopsy pathology.

RESULTS: In the externally validated dataset, the deep learning-based radiomics model showed the highest area under the curve (AUC 0.700 to 0.791). It exceeded the clinical-imaging model (AUC 0.597 to 0.718), conventional radiomic model (AUC 0.566 to 0.632), PIRADS score (AUC 0.554 to 0.613), and biopsy pathology (AUC 0.537 to 0.578). The AUC predicted by the model did not show a statistically significant difference among the three externally verified hospitals (p > 0.05).

CONCLUSION: Deep-learning radiomics models utilizing AI-extracted image features from bpMRI images can potentially be used to predict PCa aggressiveness, demonstrating a generalized ability for external validation.

CRITICAL RELEVANCE STATEMENT: Predicting the aggressiveness of prostate cancer (PCa) is important for formulating the best treatment plan for patients. The radiomic model based on deep learning is expected to provide an objective and non-invasive method for evaluating the aggressiveness of PCa.

KEY POINTS: Predicting the aggressiveness of PCa is important for patients to obtain the best treatment options. The deep learning-based radiomics model can predict the aggressiveness of PCa with high accuracy. The model has good universality when tested on multiple external datasets.

PMID:39812752 | DOI:10.1186/s13244-024-01865-8

Neuroinformatics. 2025 Jan 15;23(1):7. doi: 10.1007/s12021-024-09712-3.

ABSTRACT

This study presents a thorough bibliometric analysis of Neuroinformatics over the past 20 years, offering insights into the journal's evolution at the intersection of neuroscience and computational science. Using advanced tools such as VOS viewer and methodologies like co-citation analysis, bibliographic coupling, and keyword co-occurrence, we examine trends in publication, citation patterns, and the journal's influence. Our analysis reveals enduring research themes like neuroimaging, data sharing, machine learning, and functional connectivity, which form the core of Neuroinformatics. These themes highlight the journal's role in addressing key challenges in neuroscience through computational methods. Emerging topics like deep learning, neuron reconstruction, and reproducibility further showcase the journal's responsiveness to technological advances. We also track the journal's rising impact, marked by a substantial growth in publications and citations, especially over the last decade. This growth underscores the relevance of computational approaches in neuroscience and the high-quality research the journal attracts. Key bibliometric indicators, such as publication counts, citation analysis, and the h-index, spotlight contributions from leading authors, papers, and institutions worldwide, particularly from the USA, China, and Europe. These metrics provide a clear view of the scientific landscape and collaboration patterns driving progress. This analysis not only celebrates Neuroinformatics's rich history but also offers strategic insights for future research, ensuring the journal remains a leader in innovation and advances both neuroscience and computational science.

PMID:39812741 | DOI:10.1007/s12021-024-09712-3

Eur Radiol Exp. 2025 Jan 15;9(1):8. doi: 10.1186/s41747-024-00535-0.

ABSTRACT

BACKGROUND: Hemorrhagic transformation (HT) is a complication of reperfusion therapy following acute ischemic stroke (AIS). We aimed to develop and validate a model for predicting HT and its subtypes with poor prognosis-parenchymal hemorrhage (PH), including PH-1 (hematoma within infarcted tissue, occupying < 30%) and PH-2 (hematoma occupying ≥ 30% of the infarcted tissue)-in AIS patients following intravenous thrombolysis (IVT) based on noncontrast computed tomography (NCCT) and clinical data.

METHODS: In this six-center retrospective study, clinical and imaging data from 445 consecutive IVT-treated AIS patients were collected (01/2018-06/2023). The training cohort comprised 344 patients from five centers, and the test cohort included 101 patients from the sixth center. A clinical model was developed using eXtreme Gradient Boosting, an NCCT-based imaging model was created using deep learning, and an ensemble model integrated both models. Comparison with existing clinical scores (MSS, SEDAN, GRASPS) was performed using the DeLong test.

RESULTS: Of the 445 individuals, 202 (45.4%) had HT, 79 (17.8%) had hemorrhagic infarction, and 123 (27.6%) had PH. In the test cohort, the area under the receiver operating characteristic curve (AUROC) of the clinical, imaging, and ensemble model for HT prediction was 0.877, 0.920, and 0.937, respectively. The ensemble model for HT prediction outperformed MSS, SEDAN, and GRASPS scores (p ≤ 0.023). The ensemble model predicted PH and PH-2 with AUROC of 0.858 and 0.806, respectively.

CONCLUSION: Developing and validating an integrated model that can predict HT and its subtypes in AIS patients following IVT based on NCCT and clinical data is feasible.

RELEVANCE STATEMENT: The clinical, imaging, and ensemble models based on noncontrast CT and clinical data outperformed existing clinical scores in predicting hemorrhagic transformation of AIS and its subtypes with poor prognosis, facilitating personalized treatment decisions.

KEY POINTS: The models demonstrated the capability to predict hemorrhagic transformation of acute ischemic stroke quickly, accurately, and reliably. The proposed models outperformed existing clinical scores in predicting hemorrhagic transformation. The ensemble model provided risk assessment of parenchymal hemorrhage and parenchymal hemorrhage-2 outperforming existing clinical scores.

PMID:39812734 | DOI:10.1186/s41747-024-00535-0

Osteoporos Int. 2025 Jan 15. doi: 10.1007/s00198-024-07378-w. Online ahead of print.

ABSTRACT

This study utilized deep learning for bone mineral density (BMD) prediction and classification using biplanar X-ray radiography (BPX) images from Huashan Hospital Medical Checkup Center. Results showed high accuracy and strong correlation with quantitative computed tomography (QCT) results. The proposed models offer potential for screening patients at a high risk of osteoporosis and reducing unnecessary radiation and costs.

PURPOSE: To explore the feasibility of using a hybrid deep learning framework (HDLF) to establish a model for BMD prediction and classification based on BPX images. This study aimed to establish an automated tool for screening patients at a high risk of osteoporosis.

METHODS: A total of 906 BPX scans from 453 subjects were included in this study, with QCT results serving as the reference standard. The training-validation set:independent test set ratio was 4:1. The L1-L3 vertebral bodies were manually annotated by experienced radiologists, and the HDLF was established to predict BMD and diagnose abnormality based on BPX images and clinical information. The performance metrics of the models were calculated and evaluated.

RESULTS: The R 2 values of the BMD prediction regression model in the independent test set based on BPX images and multimodal data (BPX images and clinical information) were 0.77 and 0.79, respectively. The Pearson correlation coefficients were 0.88 and 0.89, respectively, with P-values < 0.001. Bland-Altman analysis revealed no significant difference between the predictions of the models and QCT results. The classification model achieved the highest AUC of 0.97 based on multimodal data in the independent test set, with an accuracy of 0.93, sensitivity of 0.84, specificity of 0.96, and F1 score of 0.93.

CONCLUSION: This study demonstrates that deep learning neural networks applied to BPX images can accurately predict BMD and perform classification diagnoses, which can reduce the radiation risk, economic consumption, and time consumption associated with specialized BMD measurement.

PMID:39812675 | DOI:10.1007/s00198-024-07378-w

Cancer Med. 2025 Jan;14(2):e70574. doi: 10.1002/cam4.70574.

ABSTRACT

BACKGROUND: Medical images play an important role in diagnosis and treatment of pediatric solid tumors. The field of radiology, pathology, and other image-based diagnostics are getting increasingly important and advanced. This indicates a need for advanced image processing technology such as Deep Learning (DL).

AIM: Our review focused on the use of DL in multidisciplinary imaging in pediatric surgical oncology.

METHODS: A search was conducted within three databases (Pubmed, Embase, and Scopus), and 2056 articles were identified. Three separate screenings were performed for each identified subfield.

RESULTS: In total, we identified 36 articles, divided between radiology (n = 22), pathology (n = 9), and other image-based diagnostics (n = 5). Four types of tasks were identified in our review: classification, prediction, segmentation, and synthesis. General statements about the studies'' performance could not be made due to the inhomogeneity of the included studies. To implement DL in pediatric clinical practice, both technical validation and clinical validation are of uttermost importance.

CONCLUSION: In conclusion, our review provided an overview of all DL research in the field of pediatric surgical oncology. The more advanced status of DL in adults should be used as guide to move the field of DL in pediatric oncology further, to keep improving the outcomes of children with cancer.

PMID:39812075 | DOI:10.1002/cam4.70574

J Educ Health Promot. 2024 Nov 29;13:456. doi: 10.4103/jehp.jehp_141_24. eCollection 2024.

ABSTRACT

BACKGROUND: Reflection is one of the main components of the medical sciences curriculum. It is one of the learner-centered educational strategies, leading to deep learning, and is necessary to attain professional capabilities. A pertinent challenge is how to assess reflection. This study was conducted to design and assess psychometric characteristics of medical sciences students' reflection skills self-assessment questionnaire (RSSAQ) in Persian.

MATERIALS AND METHODS: This is a methodological explorative study conducted at our University of Medical Sciences. First, an item pool was collected from both the literature review (previously designed questionnaires and existent models of reflection) and experts' and researchers' perspectives. Then the initial version of the questionnaire was presented to 19 experts and 50 students to assess the face and content validity. To assess the reliability, 48 students filled out the questionnaire twice at a one-week interval. To assess the construct validity, exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were done. For doing so, 151 students filled out the questionnaire. The data was analyzed using IBM SPSS (Statistical Package for the Social Sciences) Statistics for Windows version 16 and Analysis of Moment Structures (AMOS).

RESULTS: The content validity index (CVI), content validity ratio (CVR), and impact score (IS) for the questionnaire came out to be 0.91, 0.75, and 4.68, respectively. Regarding the reliability, Cronbach's alpha and intraclass correlation coefficient (ICC) were, respectively, 0.75 and 0.8 (with a 95 percent confidence interval). Regarding the construct validity, three factors were extracted, labeled as "readiness to reflect" (RTR), "reflection in action" (RIA), and "reflection on action" (ROA). It became clear that this questionnaire can predict 36.8 percent of variations in reflective behavior or process in students. CFA determined that there is a positive and significant correlation between RIA and ROA factors. However, the RTR factor has a negative correlation with the other factors and vice versa.

CONCLUSION: The questionnaire designed in this study for reflection self-assessment had acceptable psychometric characteristics and can be applied in curriculum planning, educational evaluations, and designing educational interventions.

PMID:39811846 | PMC:PMC11731446 | DOI:10.4103/jehp.jehp_141_24

Alzheimers Dement (Amst). 2025 Jan 14;17(1):e70041. doi: 10.1002/dad2.70041. eCollection 2025 Jan-Mar.

ABSTRACT

INTRODUCTION: Diagnostic performance of optical coherence tomography (OCT) to detect Alzheimer's disease (AD) and mild cognitive impairment (MCI) remains limited. We aimed to develop a deep-learning algorithm using OCT to detect AD and MCI.

METHODS: We performed a cross-sectional study involving 228 Asian participants (173 cases/55 controls) for model development and testing on 68 Asian (52 cases/16 controls) and 85 White (39 cases/46 controls) participants. Features from OCT were used to develop an ensemble trilateral deep-learning model.

RESULTS: The trilateral model significantly outperformed single non-deep learning models in Asian (area under the curve [AUC] = 0.91 vs. 0.71-0.72, p = 0.022-0.032) and White (AUC = 0.84 vs. 0.58-0.75, p = 0.056- < 0.001) populations. However, its performance was comparable to that of the trilateral statistical model (AUCs similar, p > 0.05).

DISCUSSION: Both multimodal approaches, using deep learning or traditional statistical models, show promise for AD and MCI detection. The choice between these models may depend on computational resources, interpretability preferences, and clinical needs.

HIGHLIGHTS: A deep-learning algorithm was developed to detect Alzheimer's disease (AD) and mild cognitive impairment (MCI) using OCT images.The combined model outperformed single OCT parameters in both Asian and White cohorts.The study demonstrates the potential of OCT-based deep-learning algorithms for AD and MCI detection.

PMID:39811700 | PMC:PMC11730192 | DOI:10.1002/dad2.70041

Osteoarthr Cartil Open. 2024 Nov 30;7(1):100551. doi: 10.1016/j.ocarto.2024.100551. eCollection 2025 Mar.

ABSTRACT

OBJECTIVE: Lower limb malalignment can complicate symptoms and accelerate knee osteoarthritis (OA), necessitating consideration in study population selection. In this study, we develop and validate a deep learning model that classifies leg alignment as "normal" or "malaligned" from knee antero-posterior (AP)/postero-anterior (PA) radiographs alone, using an adjustable hip-knee-ankle (HKA) angle threshold.

MATERIAL AND METHODS: We utilized 8878 digital radiographs, including 6181 AP/PA full-leg x-rays (LLRs) and 2697 AP/PA knee x-rays (2292 with positioning frame, 405 without). The model's evaluation involved two steps: In step 1, the model's predictions on knee images cropped from LLRs were compared against the ground truth from the original LLRs. In step 2, the model was tested on knee AP radiographs, using corresponding same-day LLRs as a proxy for ground truth.

RESULTS: The model effectively classified alignment, with step one achieving sensitivity and specificity of 0.92 for a threshold of 7.5°, and 0.90 and 0.85 for 5°. For positioning frame images, step two showed a sensitivity of 0.85 and specificity of 0.81 for 7.5°, and 0.79 and 0.74 for 5°. For non-positioning frame images, sensitivity and specificity were 0.91 and 0.83 for 7.5°, and 0.9 and 0.86 for 5°.

CONCLUSION: The model developed in this study accurately classifies lower limb malalignment from AP/PA knee radiographs using adjustable thresholds, offering a practical alternative to LLRs. This can enhance the precision of study population selection and patient management.

PMID:39811691 | PMC:PMC11729668 | DOI:10.1016/j.ocarto.2024.100551

Data Brief. 2024 Dec 11;58:111221. doi: 10.1016/j.dib.2024.111221. eCollection 2025 Feb.

ABSTRACT

Contemporary research in 3D object detection for autonomous driving primarily focuses on identifying standard entities like vehicles and pedestrians. However, the need for large, precisely labelled datasets limits the detection of specialized and less common objects, such as Emergency Medical Service (EMS) and law enforcement vehicles. To address this, we leveraged the Car Learning to Act (CARLA) simulator to generate and fairly distribute rare EMS vehicles, automatically labelling these objects in 3D point cloud data. This enriched dataset, organized in the KITTI 3D object detection benchmark format by the Karlsruhe Institute of Technology and the Toyota Technological Institute, improves its utility for training and evaluating autonomous vehicle systems. To bridge the gap between simulated and real-world scenarios, our methodology integrates a wide range of scenarios simulation in CARLA, including variations in weather conditions, human presence, and different environmental settings. This approach enhances the realism and robustness of the dataset, making it more applicable to practical autonomous driving scenarios. The data provided in this article offers a valuable resource for researchers, industry professionals, and stakeholders interested in advancing autonomous vehicle technologies and improving emergency vehicle detection. Furthermore, this dataset contributes to broader efforts in road safety and the development of AI systems capable of handling specialized vehicle identification in real-world applications.

PMID:39811523 | PMC:PMC11730950 | DOI:10.1016/j.dib.2024.111221

Data Brief. 2024 Dec 19;58:111244. doi: 10.1016/j.dib.2024.111244. eCollection 2025 Feb.

ABSTRACT

A comprehensive dataset on lemon leaf disease can surely bring a lot of potentials into the development of agricultural research and the improvement of disease management strategies. This dataset was developed from 1354 raw images taken with professional agricultural specialist guidance from July to September 2024 in Charpolisha, Jamalpur, and further enhanced with augmented techniques, adding 9000 images. The augmentation process involves a set of techniques-flipping, rotation, zooming, shifting, adding noise, shearing, and brightening-to increase variety for different lemon leaf condition representations. Each of these images was standardized to 800 × 800 pixels resolution, so that consistency may be maintained among the dataset. All images were labelled in the nine prefixed categories: anthracnose, bacterial blight, citrus canker, curl virus, deficiency leaf, dry leaf, healthy leaf, sooty mould, and spider mites. In the present study, a DenseNet-121 architecture was used, where 20 % of the dataset was kept for validation and the remaining 80 % for training. A trained model with a batch size of 32 was trained for 30 epochs, achieving an accuracy of 98.56 % with augmentation, and 96.19 % without it. The dataset will not only act as a benchmark in developing accurate machine learning models for early disease detection, but it will also contribute to the cause of sustainable lemon cultivation practices by facilitating timely and effective disease management interventions.

PMID:39811522 | PMC:PMC11732584 | DOI:10.1016/j.dib.2024.111244

Data Brief. 2024 Dec 10;58:111216. doi: 10.1016/j.dib.2024.111216. eCollection 2025 Feb.

ABSTRACT

Epipremnum aureum, sometimes known as the Money Plant, is a popular houseplant known for its hearts-shaped leaves and durability. Commonly referred to as Golden Pothos or Devil's Ivy, it is also appreciated for its ornamental value and air cleaning ability. They say that these plants are attractive to many people owing to their tolerance to several conditions and easy care, therefore, it is no surprise that they are found in many households and workplaces. Money Plants are hardy, but like any other plant they can also be infected by various diseases, which may render them less attractive, or even unattractive. This work encompasses bacterial wilt, manganese poisoning aspects and together with a healthy leaves aspect presents all prevalent masses and offer a comprehensive image of diseases. A dataset of 224 × 224 pixel images is utilized to accomplish this work with the intention to further enhance support in Ornamental Horticulture practices and diagnose more accurately. This work not only contributes ideas and approaches in understanding the field of plants pathology but also stresses on the fact how image processing can be beneficial in looking after plants. The dataset serves as a solid foundation for deep learning approaches into Ornamental Agriculture and provides useful insights for researchers studying the cultivation of money plants.

PMID:39811518 | PMC:PMC11729688 | DOI:10.1016/j.dib.2024.111216

Biol Methods Protoc. 2025 Jan 6;10(1):bpae097. doi: 10.1093/biomethods/bpae097. eCollection 2025.

ABSTRACT

A mixture-of-experts (MoE) approach has been developed to mitigate the poor out-of-distribution (OOD) generalization of deep learning (DL) models for single-sequence-based prediction of RNA secondary structure. The main idea behind this approach is to use DL models for in-distribution (ID) test sequences to leverage their superior ID performances, while relying on physics-based models for OOD sequences to ensure robust predictions. One key ingredient of the pipeline, named MoEFold2D, is automated ID/OOD detection via consensus analysis of an ensemble of DL model predictions without requiring access to training data during inference. Specifically, motivated by the clustered distribution of known RNA structures, a collection of distinct DL models is trained by iteratively leaving one cluster out. Each DL model hence serves as an expert on all but one cluster in the training data. Consequently, for an ID sequence, all but one DL model makes accurate predictions consistent with one another, while an OOD sequence yields highly inconsistent predictions among all DL models. Through consensus analysis of DL predictions, test sequences are categorized as ID or OOD. ID sequences are subsequently predicted by averaging the DL models in consensus, and OOD sequences are predicted using physics-based models. Instead of remediating generalization gaps with alternative approaches such as transfer learning and sequence alignment, MoEFold2D circumvents unpredictable ID-OOD gaps and combines the strengths of DL and physics-based models to achieve accurate ID and robust OOD predictions.

PMID:39811444 | PMC:PMC11729747 | DOI:10.1093/biomethods/bpae097

Heliyon. 2024 Dec 16;11(1):e40855. doi: 10.1016/j.heliyon.2024.e40855. eCollection 2025 Jan 15.

ABSTRACT

INTRODUCTION: The COVID-19 pandemic has created an urgent demand for research, which has spurred the development of enhanced biosafety protocols in biosafety level (BSL)-3 laboratories to safeguard against the risks associated with handling highly contagious pathogens. Laboratory management failures can pose significant hazards.

METHODS: An external system captured images of personnel entering a laboratory, which were then analyzed by an AI-based system to verify their compliance with personal protective equipment (PPE) regulations, thereby introducing an additional layer of protection. A deep learning model was trained to detect the presence of essential PPE items, such as clothing, masks, hoods, double-layer gloves, shoe covers, and respirators, ensuring adherence to World Health Organization (WHO) standards. The internal laboratory management system used a deep learning model to delineate alert zones and monitor compliance with the imposed safety protocols.

RESULTS: The external detection system was trained on a dataset consisting of 4112 images divided into 15 PPE compliance classes. The model achieved an accuracy of 97.52 % and a recall of 97.03 %. The identification results were presented in real time via a visual interface and simultaneously stored on the administrator's dashboard for future reference. We trained the internal management system on 3347 images, achieving 90 % accuracy and 85 % recall. The results were transmitted in JSON format to the internal monitoring system, which triggered alerts in response to violations of safe practices or alert zones. Real-time notifications were sent to the administrators when the safety thresholds were met.

CONCLUSION: The BSL-3 laboratory monitoring system significantly reduces the risk of exposure to pathogens for personnel during laboratory operations. By ensuring the correct use of PPE and enhancing adherence to the imposed safety protocols, this system contributes to maintaining the integrity of BSL-3 facilities and mitigates the risk of personnel becoming infection vectors.

PMID:39811271 | PMC:PMC11730239 | DOI:10.1016/j.heliyon.2024.e40855

Ophthalmol Sci. 2024 Nov 12;5(2):100653. doi: 10.1016/j.xops.2024.100653. eCollection 2025 Mar-Apr.

ABSTRACT

PURPOSE: The diagnosis of fungal keratitis using potassium hydroxide (KOH) smears of corneal scrapings enables initiation of the correct antimicrobial therapy at the point-of-care but requires time-consuming manual examination and expertise. This study evaluates the efficacy of a deep learning framework, dual stream multiple instance learning (DSMIL), in automating the analysis of whole slide imaging (WSI) of KOH smears for rapid and accurate detection of fungal infections.

DESIGN: Retrospective observational study.

PARTICIPANTS: Corneal scrapings from 568 patients with suspected fungal keratitis; 51% contained filamentous fungi according to human expert interpretation.

METHODS: Dual stream multiple instance learning was employed to analyze WSI of KOH smears. Due to the extensive size of these images, often exceeding 100 000 pixels, conventional computer vision methods (e.g., convolutional neural networks) are not feasible. Dual stream multiple instance learning segments the WSI into patches for analysis, extracting relevant features from each patch and aggregating these to make a comprehensive slide-level diagnosis while generating heat maps to visualize areas contributing most to the prediction. Fivefold cross-validation was used for training and validation, with a hold-out test set comprising 15% of the total samples.

MAIN OUTCOME MEASURES: Accuracy, sensitivity, specificity, area under the receiver operating characteristic curve (AUC), F1 score, positive predictive value (PPV), and negative predictive value (NPV) in distinguishing fungal from nonfungal slides.

RESULTS: Dual stream multiple instance learning demonstrated an overall AUC of 0.88 with an accuracy of 79% and an F1 score of 0.79 in distinguishing fungal from nonfungal slides, with sensitivity of 85%, specificity of 71%, PPV of 80%, and NPV of 79%. For "consensus cases," where 2 human graders agreed on the slide interpretation, the model achieved an accuracy of 85% and an F1 score of 0.85. For "discrepant cases," the accuracy was 71% with an F1 score of 0.71. The generated heatmaps highlighted regions corresponding to fungal elements. Code and models are open-sourced and available at https://github.com/Redd-Cornea-AI/KOH-Smear-DSMIL.

CONCLUSIONS: The DSMIL framework shows significant promise in automating interpretation of KOH smears. Its capability to handle large, high-resolution WSI data and accurately detect fungal infections, while providing visual explanations through heatmaps, could enhance the scalability of KOH smear interpretation, ultimately reducing the global burden of blindness from infectious keratitis.

FINANCIAL DISCLOSURES: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

PMID:39811263 | PMC:PMC11731208 | DOI:10.1016/j.xops.2024.100653

Cancer Cell Int. 2025 Jan 14;25(1):13. doi: 10.1186/s12935-025-03642-z.

ABSTRACT

BACKGROUND: Patients with lung adenocarcinoma (LUAD) receiving drug treatment often have an unpredictive response and there is a lack of effective methods to predict treatment outcome for patients. Dendritic cells (DCs) play a significant role in the tumor microenvironment and the DCs-related gene signature may be used to predict treatment outcome. Here, we screened for DC-related genes to construct a prognostic signature to predict prognosis and response to immunotherapy in LUAD patients.

METHODS: DC-related biological functions and genes were identified using single-cell RNA sequencing (scRNA-seq) and bulk RNA sequencing. DCs-related gene signature (DCRGS) was constructed using integrated machine learning algorithms. Expression of key genes in clinical samples was examined by real-time q-PCR. Performance of the prognostic model, DCRGS, for the prognostic evaluation, was assessed using a multiple time-dependent receiver operating characteristic (ROC) curve, the R package, "timeROC", and validated using GEO datasets.

RESULTS: Analysis of scRNA-seq data showed that there is a significant upregulation of LGALS9 expression in DCs isolated from malignant pleural effusion samples. Leveraging the Coxboost and random survival forest combination algorithm, we filtered out six DC-related genes on which a prognostic prediction model, DCRGS, was established. A high predictive capability nomogram was constructed by combining DCRGS with clinical features. We found that patients with a high-DCRGS score had immunosuppression, activated tumor-associated pathways, and elevated somatic mutational load and copy number variant load. In contrast, patients in the low-DCRGS subgroup were resistant to chemotherapy but sensitive to the CTLA-4 immune checkpoint inhibitor and targeted therapy.

CONCLUSION: We have innovatively established a deep learning-based prediction model, DCRGS, for the prediction of the prognosis of patients with LUAD. The model possesses a strong prognostic prediction performance with high accuracy and sensitivity and could be clinically useful to guide the management of LUAD. Furthermore, the findings of this study could provide an important reference for individualized clinical treatment and prognostic prediction of patients with LUAD.

PMID:39810206 | DOI:10.1186/s12935-025-03642-z