Workshop Program

What is machine learning:

Life sciences research relies heavily on microcopy imaging coupled with computational methods for visualizing and quantifying information from the imaging data. Many open source and commercial software packages are available and actively developed for these purposes.
As bioimage analysts in a core facility we provide image analysis support for a wide variety of applications. To scientifically address specific biological problems, one needs to define the set of relevant biological objects and the set of desired measurements. The next step is to define a sequence of required image processing steps, and then select a specific algorithm implementation for each of those steps. Such a sequence of image processing algorithms with a specified parameter set is what we call a “workflow”. The implementations of the algorithms that are used in the workflows are the “components” constituting that workflow.
Components are usually available within software packages and libraries which we term “collections”.
In this talk we will give a short overview of the current bioimage analysis platforms, tool collections and components that we often use. We will cover components that implement classical algorithms as well as available machine learning and deep learning methods and will give guidance on aspects to consider when choosing the platform and components that best suit the user’s needs.

An introduction to the use of current workflows for cryo-EM structure determination. This tutorial will cover single-particle analysis workflow including: motion-correction, CTF estimation; automated particle picking; 2D classification; initial model generation; 3D classification and 3D refinement. The processing of cryo-EM movies typically requires a simple GPU workstation, we will briefly go over the required specs.

A typical image analysis workflow includes segmentation of regions and objects (eg. nuclei or cells), measuring multiple features for objects and regions, and quantifying relations between objects. Fiji is an open-source software platform based on ImageJ and a collection of compatible plugins focusing on general purpose image analysis for life-sciences. It is scriptable and enables fast prototyping of image analysis workflows. Ilastik, StarDist and other Fiji plugins provide an easy way to exploits recent advances in machine-learning and deep-learning based algorithms as handy components for use within image analysis workflows.
In this hands-on workshop we will introduce you to building image analysis workflows using Fiji, Ilastik and StarDist.
We will NOT cover the basics of image analysis, and we highly recommend all attendants to go through the following Mooc on Image Processing and Analysis for Life Scientists by EPFL image analysis team, prior to the workshop.
All attendants are expected to bring their own laptops with proper Fiji and Ilastik
installations (instructions will be given in advance, and we will not handle installation during the workshop).

Dragonfly is the ideal framework for integrating data from multiple sources into a single environment, whether your images come from a leading microscope and imaging hardware vendors or RAW files produced by academic software.
It is a commercial software, however the company provide non-commercial license for academy use. The software provides an extensive set of tools for the multi-dimensional display, transformation, segmentation, registration, and measurement of multi-scale multi-modality image data. Dragonfly delivers qualitative and quantitative results for material characterization, structure properties, surface analysis, process evaluation, quality control testing, or any analysis function that requires a high-degree of accuracy.
Image segmentation and intuitive masking operations in Dragonfly lets you quickly identify and label regions of interest to gain detailed insights into structures and properties. Whether analyzing pores, fibers, grains, phases, or anything else, Dragonfly’s quantification and analysis tools give you powerful options for counting, measuring, and characterizing image features.
In this hands-on workshop. We will go over basic tools in Dragonfly for data processing and filtering, data segmentation and quantification of selected ROI.
We will use two sets of data to demonstrate software capabilities:
1. FIB scan of Drosophila sperm storage
2. microCT scan of pig hyoid bone using bone analysis plugin
All attendants are expected to bring their own laptops with Dragonfly software:
https://www.theobjects.com/dragonfly/get-non-commercial-licensing-program.html
(Installation instructions will be given in advance, and we will not handle installation during the workshop).

Many materials science studies use scanning transmission electron microscopy (STEM) to characterize atomic-scale structure. Conventional STEM imaging experiments produce only a few intensity values at each probe position. However, modern high-speed detectors allow us to measure a full 2D diffraction pattern, over a grid of 2D probe positions, forming a four dimensional (4D)-STEM dataset. These 4D-STEM datasets record information about the local phase, orientation, deformation, and other parameters, for both crystalline and amorphous materials. However, these datasets can contain millions of images and therefore require highly automated and robust software codes in order to extract the target properties. In this workshop, I will introduce our open source py4DSTEM analysis toolkit, and teach tutorials for phase contrast imaging and crystal orientation mapping.
Software required:

All attendants are expected to bring their own laptops with:

• A python and conda installation –
I recommend Miniconda which is a small and lightweight version of conda: https://docs.conda.io/en/latest/miniconda.html
Alternatively some users might prefer to install Anaconda: https://www.anaconda.com/products/distribution

• Create a conda environment for py4DSTEM and install it.
Instructions:
1 – launch the terminal (conda prompt in windows, terminal in OSX or linux)
2 – run these command lines:
conda update conda
conda create -n py4dstem
conda activate py4dstem
conda install -c conda-forge py4dstem pymatgen jupyterlab
3 – with the py4DSTEM conda environment activated, type: “jupyter lab”

• More information on the installation is given here: https://github.com/py4dstem/py4DSTEM

• The tutorials I plan to run: (depending on time I could cut this down to 1 or expand to more than 2):
https://github.com/py4dstem/py4DSTEM_tutorials/blob/main/notebooks/ptycho02_MoS2.ipynb
o https://github.com/py4dstem/py4DSTEM_tutorials/blob/main/notebooks/orientation_01_AuAgPd_wire.ipynb
All datasets etc. required are linked in these notebooks.