Description

This summer school covered theoretical and practical aspects regarding the analysis of immune cell migration and interaction, using intravital microscopy and new A.I.-based methods. We had 18 participants (4 from USI, 14 external), from 4 european countries including BSc, MSc, PhD students, postdocs and imaging specialists.

Organizers. Dr. Santiago Fernandez Gonzalez, Dr. Diego Ulisse Pizzagalli, dr. Pau Carrillo Barbera
Tutors. Diego Morone, Alain Pulfer, Kevin Ceni, Benedikt Thelen
Supported by. IRB PhD program    Cell Migration PhD program    USI - Biomedical PhD program    IMMUNEMAP consortia    Euler Institute

Program

DAY 1: June 14th. Introduction to intravital microscopy (IVM).

Learning objectives. Understanding the applications and advantages of intravital microscopy / Understanding the importance of Open Data research / Practice with the immunemap platform

Introduction and networking activity (Pizzagalli)

1

9:00 - 10:00

Physics of 2-photon intravital microscopy (Morone)

1

10:30 - 11:30

Intravital imaging of the immune system (Gonzalez)

1

11:45 - 12:45

Image formation / histograms / Properties of digital images (Barbera / Morone)

1

14:00 - 15:00

The role of Open Data and Open Source in biomedical research , FAIR principles and IMMUNEMAP project (Pizzagalli / Barbera)

1

15:10 - 15:45

FIJI introduction (Morone)

1

15:45 - 17:00

DAY 2: June 15th: Analysis with tracking. 

Learning objectives. Understanding which type of information can be extracted from IVM data using the classical image analysis pipeline, challenges and new perspectives

Exercise 1

  • Which histogram is saturated
  • Thresholding
  • Use pixel inspector

2

9:00 - 9:30

Image analysis pipeline, quantification of cell motility, dynamism, and interaction (Pizzagalli)

2

9:30-10:30

The possibilities of Machine Learning in IVM (Pizzagalli)

2

10:45-11:30

Excercise 2 (Morone / Pulfer / Barbera Pizzagalli)

  • cell detection (classical / Otsu etc)
  • Thresholding with TWS
  • stardist
  • trackmate
  • automatic tracking with trackmate
  • computation of measures: collective (i.e. preferential direction) and individual cell behavior (i.e. change of speed)

2

14:00 - 17:00

DAY 3: June 16th: Analysis without tracking.

Learning objectives. Understanding how image processing techniques working at pixel-level can be applied to analyze IVM data without the usage of cell tracking

Exercise: track-based measures on neutrophil chemotaxis

Exercise: migration differences between WT and KO cells in spleen

 

9:00 – 10:30

Discussion Exercise 1 (Pizzagalli)

Contact analysis in Exercise 1 (Pizzagalli)

Advanced analysis without tracking (of/recruitment) (Pizzagalli)

3

10:45 - 11:00

Image processing techniques (Pulfer)

3

11:15 - 12:15

Exercise 3

  • Quantify the recruitment in vessels
  • Analyze whole LN images with OF
  • Apply the bayes coloc plugin by Alain (cell-cell interaction)
  • Imaris workshop (tracking, coloc, heatmaps, and cell-cell interaction)

3

14:00 - 17:00

DAY 4: June 17th:  Beyond tracking.

Learning objectives. Understanding how computer vision methods for action recognition can be applied to analyze IVM data without the usage of cell tracking

Inside AI: neural networks and clustering (Pizzagalli)

4

9:00 – 10:00

Trends in computer vision: classic, deep learning, generative (Pulfer)

4

10:05 - 11:00

Action recognition applied to immune cells (Pizzagalli)

  • Review of motility patterns displayed by immune cells in vivo and their biological meaning
  • Application of action recognition to quantify neutrophil dynamics and migration morpho-phenotypes

4

11:15 - 12:15

Exercise 4 (Pulfer / Pizzagalli)

  • Usage of the apoptosis detection program (incl. challenge)
  • Discussion on the limits and potential usage of Deep Learning in IVM
  • Neutrophil swarm detection

4

14:00 - 16:00

Discussion on good practices for IVM analysis (Barbera + Pizzagalli)

4

16:15 - 17:00

DAY 5: June 18th: Workshop and conclusion

Workshop: analyze data by participants (Pizzagalli):

- Quantification of leukocyte migration within and outside blood vessels via track-based measures and pixel classification.

- Analysis of DCs – Macrophage interaction via supervised spectral unmixing, contact analysis, and bayesian colocalization.
- Quantification of neutrophil swarm dynamics via unsupervised machine learning.

5

9:00 - 12:00

Exam

5

14:00 - 15:00

Concluding remarks

5

15:30 - 16:30

 

A collection of intravital microscopy videos of immune cells along with their tracks (3d) manually annotated by multiple operators
RAW data available as Imaris IMS, TIF sequences, and Cell Tracking Challenge (CTC) format
Ground truth available as CSV files and Cell Tracking Challenge (CTC) format

link: http://www.nature.com/articles/sdata2018129

 More coming soon

General motility measures (track based and step based)

This tool reads Imaris exported xls files, as well as TrackMate xml files and computes basic motility patterns  useful in intravital imaging studies including:
track length, speed, directionality, displacement, arrest coefficient, density. Moroever, it creates plots of track with common origin.

Standalone program for Windows (download coming soon)
MATLAB source code (download coming soon)

Motility heatmap

This tool reads Imaris files (ims) and creates a motility heatmap based on Optical Flow as described in Pizzagalli et al. Frontiers in Immunology (2019)
It is useful to identify hotspots with high motility, or to analyze cell motility without requiring tracking.

Standalone program for Windows (download coming soon)
MATLAB source code (download coming soon)

Cell action recognition (track-based)

This tool to detects and count actions of neutrophils as described in Pizzagalli et al. Frontiers in Immunology (2019)
https://www.frontiersin.org/articles/10.3389/fimmu.2019.02621/abstract
Like a FACS gating, but for motility patterns.

  • Program for Windows (coming soon)
  • MATLAB script (download) - Requires Matlab to be executed
  • Imaris XTension (coming soon)

Semi-supervised Colocalization

This tool creates an additional imaging channel, specific for the cells of interest.

The user can draw a few lines on the cells of interest and other lines on the background.
The program will automatically separate them into 2 classes.

Useful when cells appear in more than one acquisition channel.

Please publish modules in offcanvas position.

Free Joomla! templates by Engine Templates