dzsyang / 3dtracking

3D Tracking Based on Machine Learning

A 3D real-time tracking microscopy based on machine learning algorithm (T370PRP45001)

Introduction

• Goal: Individual identification and tracking with high density.
• Milestones: 2D with low density ---> 2D with high density ---> 3D with high density.
• Languages: Python for machine learning; C++ for controlling.

Group Members

Progress Log

References (Literature Review)

all-in-one

• BTrack
  • Paper:
    • Automated deep lineage tree analysis using a Bayesian single cell tracking approach
    • Local cellular neighbourhood controls proliferation in cell competition
  • Code:
    • Github repo
    • Form: Python package
  • Using:
  • Methods:
  • Test:
• Deeptrack
  • Paper:
    • Quantitative Digital Microscopy with Deep Learning
  • Code:
    • Github repo
    • Form: Python package
  • Using:
    • U-Net
    • lodeSTAR
    • MAGIK
  • Methods:
  • Test:
    • in ``
• ULTRACK
  • Paper:
    • Large-Scale Multi-Hypotheses Cell Tracking Using Ultrametric Contours Maps
  • Code:
    • Github repo
    • Form: Python package
  • Using:
    • Watershed
    • Cellpose
    • Stardist
    • U-Net
    • ILP(integer linear programming)
  • Methods:
  • Test:
    • in ``
• EmbedTrack
  • Paper:
    • EmbedTrack—Simultaneous Cell Segmentation and Tracking Through Learning Offsets and Clustering Bandwidths
  • Code:
    • Github repo
    • Form: Python project (no package)
  • Using:
  • Methods:
  • Test:
• LIM tracker
  • Paper
    • LIM Tracker: a software package for cell tracking and analysis with advanced interactivity
  • Code:
    • Github repo
    • Form: Fiji plugin
  • Using:
    • Cellpose
    • Stardist
  • Methods:
  • Test:

cell segmentation

• Cellpose
  • Paper
    • Cellpose3: one-click image restoration for improved cellular segmentation
  • Code:
    • Github repo
    • Form: Python package
  • Using:
    • Pytorch
  • Methods:
  • Test:
    • in tests/cell_segment/cellpose/
• Stardist
  • Paper:
    • Cell Detection with Star-Convex Polygons
    • Star-convex Polyhedra for 3D Object Detection and Segmentation in Microscopy
  • Code:
    • Github repo
    • Form: Python package / Desktop app / Fiji/ImageJ/... plugin
  • Using:
    • Tensorflow
  • Methods:
  • Test:
    • in tests/cell_segment/stardist/