School of Life Sciences, University of Chinese Academy of Sciences, Beijing 100049, China # These authors contributed equally to this work. Email: andrewxiao@bupt.edu.cn;liuliqing@ibp.ac.cn;huj@ibp.ac.cn; feisun@ibp.ac.cn; yangge@ucas.edu.cn;xutao@ibp.ac.cn ABSTRACT Recent advances in 2D large - scale and 3D volume electron microscopy have stimulated the rapid development of nanoscale functional analysis at the tissue and organ levels. To meet the requirements of characterizing intracellular organelle s and their interactions within defined cellular cohorts at tissue level, we have developed DeepOrganelle. It adopts a lightweighted Mask2Former frameworks as a universal segmentor and is capable of segmenting and extracting organelles within different cell types, performing statistical quantitative analysis, as well as visualizing and quantifying the spatial distribution of organelle morphologies and interactions across different cell types at tissue scales. Using DeepOrganelle, we systemically perform cross - scale quantification of membrane contact sites( MCSs) dynamics across the progression of the seminiferous epithelial cycle, covering 12 distinct developmental stages and 24 statuses of germ cells . Noticeably, it discovers a waved pattern of mitochondria(Mito) - endoplasmic reticulum(ER) contact with a significant increase specifically at Stage X pachytene preceding the transition to diplotene, which aligns well with a newly reported experiment that mitochondrial metabolic proteins like PDHA2 are essential for this transition by maintaining ATP supply for double - strand break (DSB) repair.

Semantic segmentation of electron microscopy (EM) images of biological samples remains a challenge in the life sciences. EM data captures details of biological structures, sometimes with such complexity that even human observers can find it overwhelming. We introduce ε-Seg, a method based on hierarchical variational autoencoders (HVAEs), employing center-region masking, sparse label contrastive learning (CL), a Gaussian mixture model (GMM) prior, and clustering-free label prediction. Center-region masking and the inpainting loss encourage the model to learn robust and representative embeddings to distinguish the desired classes, even if training labels are sparse (0.05% of the total image data or less). For optimal performance, we employ CL and a GMM prior to shape the latent space of the HVAE such that encoded input patches tend to cluster wrt. the semantic classes we wish to distinguish. Finally, instead of clustering latent embeddings for semantic segmentation, we propose a MLP semantic segmentation head to directly predict class labels from latent embeddings. We show empirical results of ε-Seg and baseline methods on 2 dense EM datasets of biological tissues and demonstrate the applicability of our method also on fluorescence microscopy data. Our results show that ε-Seg is capable of achieving competitive sparsely-supervised segmentation results on complex biological image data, even if only limited amounts of training labels are available.

Live-cell imaging of multiple subcellular structures is essential for understanding subcellular dynamics. However, the conventional multi-color sequential fluorescence microscopy suffers from significant imaging delays and limited number of subcellular structure separate labeling, resulting in substantial limitations for real-time live-cell research applications. Here, we present the Adaptive Explainable Multi-Structure Network (AEMS-Net), a deep-learning framework that enables simultaneous prediction of two subcellular structures from a single image. The model normalizes staining intensity and prioritizes critical image features by integrating attention mechanisms and brightness adaptation layers. Leveraging the Kolmogorov-Arnold representation theorem, our model decomposes learned features into interpretable univariate functions, enhancing the explainability of complex subcellular morphologies. We demonstrate that AEMS-Net allows real-time recording of interactions between mitochondria and microtubules, requiring only half the conventional sequential-channel imaging procedures. Notably, this approach achieves over 30% improvement in imaging quality compared to traditional deep learning methods, establishing a new paradigm for long-term, interpretable live-cell imaging that advances the ability to explore subcellular dynamics.

Advances in microscopy imaging enable researchers to visualize structures at the nanoscale level thereby unraveling intricate details of biological organization. However, challenges such as image noise, photobleaching of fluorophores, and low tolerability of biological samples to high light doses remain, restricting temporal resolutions and experiment durations. Reduced laser doses enable longer measurements at the cost of lower resolution and increased noise, which hinders accurate downstream analyses. Here we train a denoising diffusion probabilistic model (DDPM) to predict high-resolution images by conditioning the model on low-resolution information. Additionally, the probabilistic aspect of the DDPM allows for repeated generation of images that tend to further increase the signal-to-noise ratio. We show that our model achieves a performance that is better or similar to the previously best-performing methods, across four highly diverse datasets. Importantly, while any of the previous methods show competitive performance for some, but not all datasets, our method consistently achieves high performance across all four data sets, suggesting high generalizability.

The analysis of dynamic organelles remains a formidable challenge, though key to understanding biological processes. We introduce Nellie, an automated and unbiased pipeline for segmentation, tracking, and feature extraction of diverse intracellular structures. Nellie adapts to image metadata, eliminating user input. Nellie's preprocessing pipeline enhances structural contrast on multiple intracellular scales allowing for robust hierarchical segmentation of sub-organellar regions. Internal motion capture markers are generated and tracked via a radius-adaptive pattern matching scheme, and used as guides for sub-voxel flow interpolation. Nellie extracts a plethora of features at multiple hierarchical levels for deep and customizable analysis. Nellie features a Napari-based GUI that allows for code-free operation and visualization, while its modular open-source codebase invites customization by experienced users. We demonstrate Nellie's wide variety of use cases with two examples: unmixing multiple organelles from a single channel using feature-based classification and training an unsupervised graph autoencoder on mitochondrial multi-mesh graphs to quantify latent space embedding changes following ionomycin treatment.

Scientists have been working for generations to untangle the mysteries of how life began on Earth, and one previously fringe theory just gained a lot of ground. The'RNA World' theory says that the so-called primordial soup of the early Earth was teeming with DNA's single-stranded sister RNA, which carries the instructions for sustaining life. Now, a team of researchers at The Salk Institute have unlocked a crucial piece of that puzzle and even built it in the lab: an obscure but essential class of molecules called RNA polymerase ribozymes. RNA polymerase ribozymes are not well understood, but scientists now suspect that these substances made it possible for RNA to not just replicate but actually evolve in the gel and muck of the early planet. These scatterplots show how, across multiple rounds of evolution, new RNA polymerase ribozymes emerged.

Automated and semi-automated techniques in biomedical electron microscopy (EM) enable the acquisition of large datasets at a high rate. Segmentation methods are therefore essential to analyze and interpret these large volumes of data, which can no longer completely be labeled manually. In recent years, deep learning algorithms achieved impressive results in both pixel-level labeling (semantic segmentation) and the labeling of separate instances of the same class (instance segmentation). In this review, we examine how these algorithms were adapted to the task of segmenting cellular and sub-cellular structures in EM images. The special challenges posed by such images and the network architectures that overcame some of them are described. Moreover, a thorough overview is also provided on the notable datasets that contributed to the proliferation of deep learning in EM. Finally, an outlook of current trends and future prospects of EM segmentation is given, especially in the area of label-free learning.

Drosophila that represents one of the models of neurodegeneration used in the lab to screen for things (both chemically and genetically) that regulate mitophagy. A new study, published in the journal PLOS Biology, suggests that the language used by researchers in describing their results can be utilized to uncover new treatments for Parkinson's disease. The study, led by Angus McQuibban of the University of Toronto in Canada, utilized AI to find an existing anti-cholesterol medication that has the capability to enhance the disposal of mitochondria, which are cellular components responsible for energy production and are affected in Parkinson's disease. The full pathogenic pathway leading to Parkinson's disease (PD) is unknown, but one clear contributor is mitochondrial dysfunction and the inability to dispose of defective mitochondria, a process called mitophagy. At least five genes implicated in PD are linked to impaired mitophagy, either directly or indirectly, and so the authors sought compounds that could enhance the mitophagy process.

Neurons make up less than half of the brain. Yet when it comes to brain mapping, they get all the limelight. It's easy to see why: as shockingly powerful mini-processors, neurons and their connections--together dubbed the connectome--hold the secret to highly efficient and flexible computation. Nestled inside the brain's wiring diagrams are the keys to consciousness, memories, and emotion. To connectomics, mapping the brain isn't just an academic exercise to better understand ourselves--it could lead to more efficient AI that thinks like us.

Neurons communicate via synapses--tiny, button-like protrusions that sprout from one neuron and connect it to the next. These minuscule structures are thought to be the backbone of learning and memory, changing in strength and number as we learn. At about 1/5,000th the width of a human hair, synapses can be hard to visualize, and researchers are just beginning to develop the tools necessary to do so. In a study published in Cell Reports on August 2, researchers at the Chinese Academy of Sciences and Shanghai University used a combination of deep learning algorithms and high-resolution electron microscopy to map out how frightful experiences rearrange brain connections. They found that when mice learn to fear the sound of a buzzer, neurons in their hippocampus form more connections with other neurons downstream and shuttle more mitochondria to synaptic sites.