Small. 2025 Apr 9:e2410258. doi: 10.1002/smll.202410258. Online ahead of print.

ABSTRACT

The human intestine, a biomechanically active organ, generates cyclic mechanical forces crucial for maintaining its health and functions. Yet, the physiological impact of these forces on gut microbiota dynamics remains largely unexplored. In this study, we investigate how cyclic intestinal motility influences the dynamics of gut microbial communities within a 3D gut-like structure (µGut). To enable the study, a tunable flow Gut Microbiome-on-a-Chip (tfGMoC) is developed that recapitulates the cyclic expansion and compression of intestinal motility while allowing high-magnification imaging of microbial communities within a 3D stratified, biomimetic gut epithelium. Using deep learning-based microbial analysis, it is found that hydrodynamic forces organize microbial communities by promoting distinct spatial exploration behaviors in microorganisms with varying motility characteristics. Empirical evidence demonstrates the impact of gut motility forces in maintaining a balanced gut microbial composition, enhancing both the diversity and stability of the community - key factors for a healthy microbiome. This study, leveraging the new tfGMoC platform, uncovers previously unknown effects of intestinal motility on modulating gut microbial behaviors and community organizations. This will be critical for a deeper understanding of host-microbiome interactions in the emerging field of microbiome therapeutics.

PMID:40201941 | DOI:10.1002/smll.202410258

Head Neck Tumor Segm MR Guid Appl (2024). 2025;15273:204-213. doi: 10.1007/978-3-031-83274-1_15. Epub 2025 Mar 3.

ABSTRACT

Radiation therapy is one of the most frequently applied cancer treatments worldwide, especially in the context of head and neck cancer. Today, MRI-guided radiation therapy planning is becoming increasingly popular due to good soft tissue contrast, lack of radiation dose delivered to the patient, and the capability of performing functional imaging. However, MRI-guided radiation therapy requires segmenting of the cancer both before and during radiation therapy. So far, the segmentation was often performed manually by experienced radiologists, however, recent advances in deep learning-based segmentation suggest that it may be possible to perform the segmentation automatically. Nevertheless, the task is arguably more difficult when using MRI compared to e.g. PET-CT because even manual segmentation of head and neck cancer in MRI volumes is challenging and time-consuming. The importance of the problem motivated the researchers to organize the HNTSMRG challenge with the aim of developing the most accurate segmentation methods, both before and during MRI-guided radiation therapy. In this work, we benchmark several different state-of-the-art segmentation architectures to verify whether the recent advances in deep encoder-decoder architectures are impactful for low data regimes and low-contrast tasks like segmenting head and neck cancer in magnetic resonance images. We show that for such cases the traditional residual UNetbased method outperforms (DSC = 0.775/0.701) recent advances such as UNETR (DSC = .617/0.657), SwinUNETR (DSC = 0.757/0.700), or SegMamba (DSC = 0.708/0.683). The proposed method (lWM team) achieved a mean aggregated Dice score on the closed test set at the level of 0.771 and 0.707 for the pre- and mid-therapy segmentation tasks, scoring 14th and 6th place, respectively. The results suggest that proper data preparation, objective function, and preprocessing are more influential for the segmentation of head and neck cancer than deep network architecture.

PMID:40201773 | PMC:PMC11977277 | DOI:10.1007/978-3-031-83274-1_15

Head Neck Tumor Segm MR Guid Appl (2024). 2025;15273:259-273. doi: 10.1007/978-3-031-83274-1_21. Epub 2025 Mar 3.

ABSTRACT

Automated segmentation of gross tumor volumes (GTVp) and lymph nodes (GTVn) in head and neck cancer using MRI presents a critical challenge with significant potential to enhance radiation oncology workflows. In this study, we developed a deep learning pipeline based on the SegResNet architecture, integrated into the Auto3DSeg framework, to achieve fully-automated segmentation on pre-treatment (pre-RT) and mid-treatment (mid-RT) MRI scans as part of the DLaBella29 team submission to the HNTS-MRG 2024 challenge. For Task 1, we used an ensemble of six SegResNet models with predictions fused via weighted majority voting. The models were pre-trained on both pre-RT and mid-RT image-mask pairs, then fine-tuned on pre-RT data, without any pre-processing. For Task 2, an ensemble of five SegResNet models was employed, with predictions fused using majority voting. Pre-processing for Task 2 involved setting all voxels more than 1 cm from the registered pre-RT masks to background (value 0), followed by applying a bounding box to the image. Post-processing for both tasks included removing tumor predictions smaller than 175-200 mm3 and node predictions under 50-60 mm3. Our models achieved testing DSCagg scores of 0.72 and 0.82 for GTVn and GTVp in Task 1 (pre-RT MRI) and testing DSCagg scores of 0.81 and 0.49 for GTVn and GTVp in Task 2 (mid-RT MRI). This study underscores the feasibility and promise of deep learning-based auto-segmentation for improving clinical workflows in radiation oncology, particularly in adaptive radiotherapy. Future efforts will focus on refining mid-RT segmentation performance and further investigating the clinical implications of automated tumor delineation.

PMID:40201772 | PMC:PMC11978229 | DOI:10.1007/978-3-031-83274-1_21

Head Neck Tumor Segm MR Guid Appl (2024). 2025;15273:36-49. doi: 10.1007/978-3-031-83274-1_2. Epub 2025 Mar 3.

ABSTRACT

Radiation therapy (RT) is a vital part of treatment for head and neck cancer, where accurate segmentation of gross tumor volume (GTV) is essential for effective treatment planning. This study investigates the use of pre-RT tumor regions and local gradient maps to enhance mid-RT tumor segmentation for head and neck cancer in MRI-guided adaptive radiotherapy. By leveraging pre-RT images and their segmentations as prior knowledge, we address the challenge of tumor localization in mid-RT segmentation. A gradient map of the tumor region from the pre-RT image is computed and applied to mid-RT images to improve tumor boundary delineation. Our approach demonstrated improved segmentation accuracy for both primary GTV (GTVp) and nodal GTV (GTVn), though performance was limited by data constraints. The final DSC agg scores from the challenge's test set evaluation were 0.534 for GTVp, 0.867 for GTVn, and a mean score of 0.70. This method shows potential for enhancing segmentation and treatment planning in adaptive radiotherapy. Team: DCPT-Stine's group.

PMID:40201771 | PMC:PMC11977786 | DOI:10.1007/978-3-031-83274-1_2

BJR Artif Intell. 2024 Nov 13;1(1):ubae016. doi: 10.1093/bjrai/ubae016. eCollection 2024 Jan.

ABSTRACT

Breast cancer is one of the most common and deadly cancers in women. Triple-negative breast cancer (TNBC) accounts for approximately 10%-15% of breast cancer diagnoses and is an aggressive molecular breast cancer subtype associated with important challenges in its diagnosis, treatment, and prognostication. This poses an urgent need for developing more effective and personalized imaging biomarkers for TNBC. Towards this direction, artificial intelligence (AI) for radiologic imaging holds a prominent role, leveraging unique advantages of radiologic breast images, being used routinely for TNBC diagnosis, staging, and treatment planning, and offering high-resolution whole-tumour visualization, combined with the immense potential of AI to elucidate anatomical and functional properties of tumours that may not be easily perceived by the human eye. In this review, we synthesize the current state-of-the-art radiologic imaging applications of AI in assisting TNBC diagnosis, treatment, and prognostication. Our goal is to provide a comprehensive overview of radiomic and deep learning-based AI developments and their impact on advancing TNBC management over the last decade (2013-2024). For completeness of the review, we start with a brief introduction of AI, radiomics, and deep learning. Next, we focus on clinically relevant AI-based diagnostic, predictive, and prognostic models for radiologic breast images evaluated in TNBC. We conclude with opportunities and future directions for AI towards advancing diagnosis, treatment response predictions, and prognostic evaluations for TNBC.

PMID:40201726 | PMC:PMC11974408 | DOI:10.1093/bjrai/ubae016

Data Brief. 2025 Mar 3;59:111434. doi: 10.1016/j.dib.2025.111434. eCollection 2025 Apr.

ABSTRACT

The development of advanced coffee bean classification techniques depends on the availability of high quality datasets. Coffee bean quality is influenced by various factors, including bean size, shape, colour, and defects such as fungal damage, full black, full sour, broken beans, and insect damage. Constructing an accurate and reliable ground truth dataset for coffee bean classification is a challenging and labour intensive process. To address this need, we introduce the Coffee Beans Dataset (CBD) which contains 450 high-resolution images sampled across 9 distinct coffee bean grades A, AA, AAA, AB, C, PB-I, PB-II, BITS and BULK with 50 images per class. These samples were sourced from Wayanad, Kerala, reflecting the region's diverse coffee bean quality .This dataset is specifically designed to support machine learning and deep learning models for coffee bean classification and grading. By providing a comprehensive and diverse dataset, we aim to address key challenges in coffee quality assessment and improvement in classification accuracy. When tested using the EfficientNet-B0 model, the model achieved a high accuracy of 100%, demonstrating its potential to enhance automated coffee bean grading systems. The CBD serves as a valuable resource for researchers and industry professionals, promot-ing innovation in coffee quality monitoring and classification algorithms.

PMID:40201542 | PMC:PMC11978365 | DOI:10.1016/j.dib.2025.111434

Front Radiol. 2025 Mar 25;5:1503625. doi: 10.3389/fradi.2025.1503625. eCollection 2025.

ABSTRACT

INTRODUCTION: Lumbar spine magnetic resonance imaging (MRI) plays a critical role in diagnosing and planning treatment for spinal conditions such as degenerative disc disease, spinal canal stenosis, and disc herniation. Measuring the cross-sectional area of the dural sac (DSCA) is a key factor in evaluating the severity of spinal canal narrowing. Traditionally, radiologists perform this measurement manually, which is both time-consuming and susceptible to errors. Advances in deep learning, particularly convolutional neural networks (CNNs) like the U-Net architecture, have demonstrated significant potential in the analysis of medical images. This study evaluates the efficacy of deep learning models for automating DSCA measurements in lumbar spine MRIs to enhance diagnostic precision and alleviate the workload of radiologists.

METHODS: For algorithm development and assessment, we utilized two extensive, anonymized online datasets: the "Lumbar Spine MRI Dataset" and the SPIDER-MRI dataset. The combined dataset comprised 683 lumbar spine MRI scans for training and testing, with an additional 50 scans reserved for external validation. We implemented and assessed three deep learning models-U-Net, Attention U-Net, and MultiResUNet-using 5-fold cross-validation. The models were trained on T1-weighted axial MRI images and evaluated on metrics such as accuracy, precision, recall, F1-score, and mean absolute error (MAE).

RESULTS: All models exhibited a high correlation between predicted and actual DSCA values. The MultiResUNet model achieved superior results, with a Pearson correlation coefficient of 0.9917 and an MAE of 23.7032 mm2 on the primary dataset. This high precision and reliability were consistent in external validation, where the MultiResUNet model attained an accuracy of 99.95%, a recall of 0.9989, and an F1-score of 0.9393. Bland-Altman analysis revealed that most discrepancies between predicted and actual DSCA values fell within the limits of agreement, further affirming the robustness of these models.

DISCUSSION: This study demonstrates that deep learning models, particularly MultiResUNet, offer high accuracy and reliability in the automated segmentation and calculation of DSCA in lumbar spine MRIs. These models hold significant potential for improving diagnostic accuracy and reducing the workload of radiologists. Despite some limitations, such as the restricted dataset size and reliance on T1-weighted images, this study provides valuable insights into the application of deep learning in medical imaging. Future research should include larger, more diverse datasets and additional image weightings to further validate and enhance the generalizability and clinical utility of these models.

PMID:40201339 | PMC:PMC11975661 | DOI:10.3389/fradi.2025.1503625

Front Med (Lausanne). 2025 Mar 25;12:1563153. doi: 10.3389/fmed.2025.1563153. eCollection 2025.

ABSTRACT

OBJECTIVE: Establish a deep learning ultrasound radiomics model based on destruction-replenishment contrast-enhanced ultrasound (DR-CEUS) for the early prediction of acute kidney injury (SA-AKI).

METHOD: This paper proposes a deep learning ultrasound radiomics model (DLUR). Deep learning models were separately established using ResNet18, ResNet50, ResNext18, and ResNext50 networks. Based on the features extracted from the fully connected layers of the optimal model, a deep learning ultrasound radiomics model (DLUR) was established using three classification models (built with 3 classifiers). The predictive performance of the best DLUR model was compared with the visual assessments of two groups of ultrasound physicians with varying levels of experience. The performance of each model and the ultrasound physicians was evaluated by assessing the receiver operating characteristic (ROC) curves. The area under the curve (AUC), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were subsequently calculated.

RESULTS: Compared to the ResNet18 model, the DLUR model based on logistic regression (DLUR-LR) demonstrated the best predictive performance, showing a Net Reclassification Improvement (NRI) value of 0.210 (p < 0.05). The Integrated Discrimination Improvement (IDI) value for the corresponding stage was 0.169 (p < 0.05). Additionally, the performance of the DLUR-LR model also surpassed that of senior ultrasound physicians (AUC, 0.921 vs. 0.829, p < 0.05).

CONCLUSION: By combining deep learning and ultrasound radiomics, a deep learning ultrasound radiomics model with outstanding predictive efficiency and robustness has demonstrated excellent capability in the early prediction of acute kidney injury (SA-AKI).

PMID:40201329 | PMC:PMC11975892 | DOI:10.3389/fmed.2025.1563153

Front Med (Lausanne). 2025 Mar 25;12:1555676. doi: 10.3389/fmed.2025.1555676. eCollection 2025.

ABSTRACT

BACKGROUND: Thyroid disease continues to be one of the most prevalent disease groups worldwide, with its frequency and distribution being impacted by numerous factors. Significant progress has been achieved in recent years in thyroid nodules, largely due to the advent of novel detection and diagnostic techniques. This study aims to scrutinize the top 100 most frequently cited articles in thyroid nodule research, utilizing bibliometric analysis to identify trends, highlight critical focal points, and lay a groundwork for forthcoming investigations.

METHODS: A comprehensive literature search was carried out using the SCI-E database, and all the recorded results were downloaded in plain text format for detailed analysis. The key terms analyzed with VOSviewer 1.6.18, CiteSpace 6.3r1, bibliometrix in R Studio (v.4.4.1), and Microsoft Excel 2021 software include country, institution, author, journal, and keywords.

RESULTS: The publication timeframe extends from 1 January 2003 to 31 December 2021, reaching a peak citation count of 9,100. Notably, the United States leads in the number of published articles, with Harvard University standing out as a prestigious institution. These articles were featured in 45 diverse journals, with THYROID leading in publication volume. Nikiforov Yuri E. was the most prolific first author, appearing 10 times. Keyword analysis highlighted traditional research themes such as "fine needle aspiration," "carcinogens," and "management." However, "deep learning" has surfaced as a significant area of focus in recent studies.

CONCLUSION: This study has extracted the bibliometric characteristics of the top 100 most-cited articles pertaining to TNs, providing an invaluable reference for upcoming studies. Through meticulous analysis, it has been determined that the primary research concentrations encompass the diagnosis of benign or malignant TNs, the management of TNs, and the subsequent monitoring of TNs, with deep learning emerging as a pivotal area of exploration.

PMID:40201321 | PMC:PMC11975563 | DOI:10.3389/fmed.2025.1555676

J Neurotrauma. 2025 Apr 9. doi: 10.1089/neu.2024.0301. Online ahead of print.

ABSTRACT

An extensive library of symptom inventories has been developed over time to measure clinical symptoms of traumatic brain injury (TBI), but this variety has led to several long-standing issues. Most notably, results drawn from different settings and studies are not comparable. This creates a fundamental problem in TBI diagnostics and outcome prediction, namely that it is not possible to equate results drawn from distinct tools and symptom inventories. Here, we present an approach using semantic textual similarity (STS) to link symptoms and scores across previously incongruous symptom inventories by ranking item text similarities according to their conceptual likeness. We tested the ability of four pretrained deep learning models to screen thousands of symptom description pairs for related content-a challenging task typically requiring expert panels. Models were tasked to predict symptom severity across four different inventories for 6,607 participants drawn from 16 international data sources. The STS approach achieved 74.8% accuracy across five tasks, outperforming other models tested. Correlation and factor analysis found the properties of the scales were broadly preserved under conversion. This work suggests that incorporating contextual, semantic information can assist expert decision-making processes, yielding broad gains for the harmonization of TBI assessment.

PMID:40200899 | DOI:10.1089/neu.2024.0301

STAR Protoc. 2025 Apr 7;6(2):103738. doi: 10.1016/j.xpro.2025.103738. Online ahead of print.

ABSTRACT

Functional impact of noncoding variants can be predicted using computational approaches. Although predictive scores can be insightful, implementing the scores for a custom variant set and associating scores with complex traits require multiple phases of analysis. Here, we present a protocol for prioritizing variants by generating deep-learning-predicted functional scores and relating them with brain traits. We describe steps for score prediction, statistical comparison, phenotype correlation, and functional enrichment analysis. This protocol can be generalized to different models and phenotypes. For complete details on the use and execution of this protocol, please refer to Mondragon-Estrada et al.1.

PMID:40198216 | DOI:10.1016/j.xpro.2025.103738

J Gerontol B Psychol Sci Soc Sci. 2025 Apr 8:gbaf065. doi: 10.1093/geronb/gbaf065. Online ahead of print.

ABSTRACT

OBJECTIVE: Alzheimer's disease and related dementias significantly impact older adults' quality of life. The clock-drawing test (CDT) is a widely used dementia screening tool due to its ease of administration and effectiveness. However, manual CDT-coding in large-scale studies can be time-intensive and prone to coding errors and is typically limited to ordinal responses. In this study, we developed a continuous CDT score using a deep learning neural network (DLNN) and evaluated its ability to classify participants as having dementia or not.

METHODS: Using a nationally representative sample of older adults from the National Health and Aging Trends Study (NHATS), we trained deep learning models on CDT images to generate both ordinal and continuous scores. Using a modified NHATS dementia classification algorithm as a benchmark, we computed the Area Under the Receiver Operating Characteristic Curve for each scoring approach. Thresholds were determined by balancing sensitivity and specificity, and demographic-specific thresholds were compared to a uniform threshold for classification accuracy.

RESULTS: Continuous CDT scores provided more granular thresholds than ordinal scores for dementia classification, which vary by demographic characteristics. Lower thresholds were identified for Black individuals, those with lower education, and those ages 90 or older. Compared to ordinal scores, continuous scores also allowed for a more balanced sensitivity and specificity.

DISCUSSION: This study demonstrates the potential of continuous CDT generated by DLNN to enhance dementia classification. By identifying demographic-specific thresholds, it offers a more inclusive and adaptive approach, which could lead to improved guidelines for using CDT in dementia screening.

PMID:40197801 | DOI:10.1093/geronb/gbaf065

J Chem Phys. 2025 Apr 14;162(14):144101. doi: 10.1063/5.0257558.

ABSTRACT

We investigate the locality of magnetic response in polycyclic aromatic molecules using a novel deep-learning approach. Our method employs graph neural networks (GNNs) with a graph-of-rings representation to predict nucleus independent chemical shifts (NICS) in the space around the molecule. We train a series of models, each time reducing the size of the largest molecules used in training. The accuracy of prediction remains high (MAE < 0.5 ppm), even when training the model only on molecules with up to four rings, thus providing strong evidence for the locality of magnetic response. To overcome the known problem of generalization of GNNs, we implement a k-hop expansion strategy and succeed in achieving accurate predictions for molecules with up to 15 rings (almost 4 times the size of the largest training example). Our findings have implications for understanding the magnetic response in complex molecules and demonstrate a promising approach to overcoming GNN scalability limitations. Furthermore, the trained models enable rapid characterization, without the need for more expensive DFT calculations.

PMID:40197568 | DOI:10.1063/5.0257558

BMC Genomics. 2025 Apr 7;26(1):350. doi: 10.1186/s12864-025-11511-2.

ABSTRACT

BACKGROUND: Clustering scRNA-seq data plays a vital role in scRNA-seq data analysis and downstream analyses. Many computational methods have been proposed and achieved remarkable results. However, there are several limitations of these methods. First, they do not fully exploit cellular features. Second, they are developed based on gene expression information and lack of flexibility in integrating intercellular relationships. Finally, the performance of these methods is affected by dropout event.

RESULTS: We propose a novel deep learning (DL) model based on attention autoencoder and zero-inflated (ZI) layer, namely scAMZI, to cluster scRNA-seq data. scAMZI is mainly composed of SimAM (a Simple, parameter-free Attention Module), autoencoder, ZINB (Zero-Inflated Negative Binomial) model and ZI layer. Based on ZINB model, we introduce autoencoder and SimAM to reduce dimensionality of data and learn feature representations of cells and relationships between cells. Meanwhile, ZI layer is used to handle zero values in the data. We compare the performance of scAMZI with nine methods (three shallow learning algorithms and six state-of-the-art DL-based methods) on fourteen benchmark scRNA-seq datasets of various sizes (from hundreds to tens of thousands of cells) with known cell types. Experimental results demonstrate that scAMZI outperforms competing methods.

CONCLUSIONS: scAMZI outperforms competing methods and can facilitate downstream analyses such as cell annotation, marker gene discovery, and cell trajectory inference. The package of scAMZI is made freely available at https://doi.org/10.5281/zenodo.13131559 .

PMID:40197174 | DOI:10.1186/s12864-025-11511-2

Radiology. 2025 Apr;315(1):e240775. doi: 10.1148/radiol.240775.

ABSTRACT

Background Acute ischemic stroke (AIS) is a major cause of morbidity and mortality, requiring swift and precise clinical decisions based on neuroimaging. Recent advances in deep learning-based computer vision and language artificial intelligence (AI) models have demonstrated transformative performance for several stroke-related applications. Purpose To evaluate deep learning applications for imaging in AIS in adult patients, providing a comprehensive overview of the current state of the technology and identifying opportunities for advancement. Materials and Methods A systematic literature review was conducted following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. A comprehensive search of four databases from January 2016 to January 2024 was performed, targeting deep learning applications for imaging of AIS, including automated detection of large vessel occlusion and measurement of Alberta Stroke Program Early CT Score. Articles were selected based on predefined inclusion and exclusion criteria, focusing on convolutional neural networks and transformers. The top-represented areas were addressed, and the relevant information was extracted and summarized. Results Of 380 studies included, 171 (45.0%) focused on stroke lesion segmentation, 129 (33.9%) on classification and triage, 31 (8.2%) on outcome prediction, 15 (3.9%) on generative AI and large language models, and 11 (2.9%) on rapid or low-dose imaging specific to stroke applications. Detailed data extraction was performed for 68 studies. Public AIS datasets are also highlighted, for researchers developing AI models for stroke imaging. Conclusion Deep learning applications have permeated AIS imaging, particularly for stroke lesion segmentation. However, challenges remain, including the need for standardized protocols and test sets, larger public datasets, and performance validation in real-world settings. © RSNA, 2025 Supplemental material is available for this article.

PMID:40197098 | DOI:10.1148/radiol.240775

J Chem Inf Model. 2025 Apr 8. doi: 10.1021/acs.jcim.5c00051. Online ahead of print.

ABSTRACT

Deep learning has revolutionized difficult tasks in chemistry and biology, yet existing language models often treat these domains separately, relying on concatenated architectures and independently pretrained weights. These approaches fail to fully exploit the shared atomic foundations of molecular and protein sequences. Here, we introduce T5ProtChem, a unified model based on the T5 architecture, designed to simultaneously process molecular and protein sequences. Using a new pretraining objective, ProtiSMILES, T5ProtChem bridges the molecular and protein domains, enabling efficient, generalizable protein-chemical modeling. The model achieves a state-of-the-art performance in tasks such as binding affinity prediction and reaction prediction, while having a strong performance in protein function prediction. Additionally, it supports novel applications, including covalent binder classification and sequence-level adduct prediction. These results demonstrate the versatility of unified language models for drug discovery, protein engineering, and other interdisciplinary efforts in computational biology and chemistry.

PMID:40197028 | DOI:10.1021/acs.jcim.5c00051

J Chem Theory Comput. 2025 Apr 8. doi: 10.1021/acs.jctc.5c00409. Online ahead of print.

ABSTRACT

We present a deep learning approach for analyzing two-dimensional scattering data of semiflexible polymers under external forces. In our framework, scattering functions are compressed into a three-dimensional latent space using a Variational Autoencoder (VAE), and two converter networks establish a bidirectional mapping between the polymer parameters (bending modulus, stretching force, and steady shear) and the scattering functions. The training data are generated using off-lattice Monte Carlo simulations to avoid the orientational bias inherent in lattice models, ensuring robust sampling of polymer conformations. The feasibility of this bidirectional mapping is demonstrated by the organized distribution of polymer parameters in the latent space. By integrating the converter networks with the VAE, we obtain a generator that produces scattering functions from given polymer parameters and an inferrer that directly extracts polymer parameters from scattering data. While the generator can be utilized in a traditional least-squares fitting procedure, the inferrer produces comparable results in a single pass and operates 3 orders of magnitude faster. This approach offers a scalable automated tool for polymer scattering analysis and provides a promising foundation for extending the method to other scattering models, experimental validation, and the study of time-dependent scattering data.

PMID:40197011 | DOI:10.1021/acs.jctc.5c00409

Front Cell Dev Biol. 2025 Mar 24;13:1549811. doi: 10.3389/fcell.2025.1549811. eCollection 2025.

ABSTRACT

BACKGROUND: Hepatocellular carcinoma (HCC) and cholangiocarcinoma (CCA) represent the primary liver cancer types. Traditional diagnostic techniques, reliant on radiologist interpretation, are both time-intensive and often inadequate for detecting the less prevalent CCA. There is an emergent need to explore automated diagnostic methods using deep learning to address these challenges.

METHODS: This study introduces HTRecNet, a novel deep learning framework for enhanced diagnostic precision and efficiency. The model incorporates sophisticated data augmentation strategies to optimize feature extraction, ensuring robust performance even with constrained sample sizes. A comprehensive dataset of 5,432 histopathological images was divided into 5,096 for training and validation, and 336 for external testing. Evaluation was conducted using five-fold cross-validation and external validation, applying metrics such as accuracy, area under the receiver operating characteristic curve (AUC), and Matthews correlation coefficient (MCC) against established clinical benchmarks.

RESULTS: The training and validation cohorts comprised 1,536 images of normal liver tissue, 3,380 of HCC, and 180 of CCA. HTRecNet showed exceptional efficacy, consistently achieving AUC values over 0.99 across all categories. In external testing, the model reached an accuracy of 0.97 and an MCC of 0.95, affirming its reliability in distinguishing between normal, HCC, and CCA tissues.

CONCLUSION: HTRecNet markedly enhances the capability for early and accurate differentiation of HCC and CCA from normal liver tissues. Its high diagnostic accuracy and efficiency position it as an invaluable tool in clinical settings, potentially transforming liver cancer diagnostic protocols. This system offers substantial support for refining diagnostic workflows in healthcare environments focused on liver malignancies.

PMID:40196844 | PMC:PMC11973358 | DOI:10.3389/fcell.2025.1549811

Bioinform Adv. 2025 Apr 2;5(1):vbaf021. doi: 10.1093/bioadv/vbaf021. eCollection 2025.

ABSTRACT

MOTIVATION: Volumetric electron microscopy (VEM) enables nanoscale resolution three-dimensional imaging of biological samples. Identification and labeling of organelles, cells, and other structures in the image volume is required for image interpretation, but manual labeling is extremely time-consuming. This can be automated using deep learning segmentation algorithms, but these traditionally require substantial manual annotation for training and typically these labeled datasets are unavailable for new samples.

RESULTS: We show that transfer learning can help address this challenge. By pretraining on VEM data from multiple mammalian tissues and organelle types and then fine-tuning on a target dataset, we segment multiple organelles at high performance, yet require a relatively small amount of new training data. We benchmark our method on three published VEM datasets and a new rat liver dataset we imaged over a 56×56×11 μ m volume measuring 7000×7000×219 px using serial block face scanning electron microscopy with corresponding manually labeled mitochondria and endoplasmic reticulum structures. We further benchmark our approach against the Segment Anything Model 2 and MitoNet in zero-shot, prompted, and fine-tuned settings.

AVAILABILITY AND IMPLEMENTATION: Our rat liver dataset's raw image volume, manual ground truth annotation, and model predictions are freely shared at github.com/Xrioen/cross-tissue-transfer-learning-in-VEM.

PMID:40196751 | PMC:PMC11974384 | DOI:10.1093/bioadv/vbaf021

bioRxiv [Preprint]. 2025 Mar 26:2025.03.23.644838. doi: 10.1101/2025.03.23.644838.

ABSTRACT

Object tracking in microscopy videos is crucial for understanding biological processes. While existing methods often require fine-tuning tracking algorithms to fit the image dataset, here we explored an alternative paradigm: augmenting the image time-lapse dataset to fit the tracking algorithm. To test this approach, we evaluated whether generative video frame interpolation can augment the temporal resolution of time-lapse microscopy and facilitate object tracking in multiple biological contexts. We systematically compared the capacity of Latent Diffusion Model for Video Frame Interpolation (LDMVFI), Real-time Intermediate Flow Estimation (RIFE), Compression-Driven Frame Interpolation (CDFI), and Frame Interpolation for Large Motion (FILM) to generate synthetic microscopy images derived from interpolating real images. Our testing image time series ranged from fluorescently labeled nuclei to bacteria, yeast, cancer cells, and organoids. We showed that the off-the-shelf frame interpolation algorithms produced bio-realistic image interpolation even without dataset-specific retraining, as judged by high structural image similarity and the capacity to produce segmentations that closely resemble results from real images. Using a simple tracking algorithm based on mask overlap, we confirmed that frame interpolation significantly improved tracking across several datasets without requiring extensive parameter tuning and capturing complex trajectories that were difficult to resolve in the original image time series. Taken together, our findings highlight the potential of generative frame interpolation to improve tracking in time-lapse microscopy across diverse scenarios, suggesting that a generalist tracking algorithm for microscopy could be developed by combining deep learning segmentation models with generative frame interpolation.

PMID:40196554 | PMC:PMC11974701 | DOI:10.1101/2025.03.23.644838