Qionghai Dai

The paper introduces DeepWonder3D, a pipeline for rapid and robust 3D neuronal extraction from multi-view projections, achieving high fidelity and efficiency in calcium dynamics analysis.

This paper introduces isoST, a deep learning framework that infers isotropic 3D gene expression from sparse 2D spatial transcriptomics data, enabling accurate 3D reconstruction of tissues and uncovering spatial gradients missed in 2D analyses.

This paper introduces a novel optical segmentation method for compressed readout of neuronal voltage dynamics, achieving high-speed recording of subcellular activity with improved data efficiency.

The paper identifies a TRPM7-Annexin A1 pathway that enables circulating tumor cells to adapt to fluid shear stress in capillaries, facilitating metastasis.

This paper introduces Meta-rLLS-VSIM, a super-resolution microscopy technique that achieves near-isotropic resolution with reduced training time and data requirements.

This paper introduces a time-lapse image super-resolution neural network with confidence quantification for optical microscopy, improving temporal consistency and reliability.

This paper introduces an unsupervised deep-learning method for multi-animal tracking that performs well in challenging conditions without requiring human annotations.

This paper introduces BRIEF, a method for compressing biomedical data with high fidelity and low bandwidth, enabling efficient sharing and downstream analysis.

This paper introduces a miniaturized mesoscope (SOMM) for high-resolution calcium imaging in freely moving mice, achieving widefield, cellular resolution with robustness against motion artifacts.

This study investigates the role of acetylcholine in shaping stable olfactory representations across the Drosophila brain using advanced imaging techniques and network analysis.

This paper introduces DeepCAD, a self-supervised learning method to enhance calcium imaging by reducing noise and improving SNR, leading to better neuron extraction and spike inference.

This paper introduces an unsupervised image transformation method for optical microscopy that avoids data annotation and achieves high-fidelity image transformations across modalities.

This paper introduces a high-sensitivity DCS system with 1024 parallel detection channels, enabling fast and sensitive measurement of blood flow and physiological variations.

This paper introduces MUSE, a method that combines single-cell morphology and transcriptional data to better characterize tissue heterogeneity and identify subpopulations missed by individual modalities.

This paper introduces a deep learning model to improve GWAS by detecting complex genetic effects, revealing potential causal pathways in disease development.

This paper introduces ED-LTFM to overcome tissue scattering in wide-field deep imaging, demonstrating improved imaging depth in mouse brain in vivo.

The paper introduces scScope, a deep learning method for cell-type composition identification from large-scale single-cell RNA-seq data.