Computational Biology and Machine Intelligence

IsoVEM: Accurate and Robust Isotropic Reconstruction for Volume Electron Microscopy Using a Video Transformer

This repository is the official implementation of ProtLocNet: Morphology-Aware Self-Supervised Representation Learning of Protein Localization in Single Cells.

It includes six basic models (UNet, ResUNet, AttUNet, UNet++, U^2Net, Swin-UNETR), along with prediction code and metric evaluation code. Furthermore, the evaluation module can detect inaccurately annotated images in the dataset by identifying excessively large HD95 metrics.

Implementation for the paper "Diffusion Prior-based Zero-Shot Denoising for Real-World Fluorescence Microscopy Images via Encoder-Decoder Guided Sampling".

A foundation model designed for learning the representation of scRNA-seq and scATAC-seq data.

The official code repository for the ECCV 2024 accepted paper "Representing Topological Self-Similarity Using Fractal Feature Maps for Accurate Segmentation of Tubular Structures".