Bioimage Data Analysis Workflows ‒ Advanced Components and Methods (Learning Materials in Biosciences)

✍ Scribed by Kota Miura (editor), Nataša Sladoje (editor)

• Publisher: Springer
• Year: 2022
• Tongue: English
• Leaves: 218
• Edition: 1st ed. 2022
• Category: Library

✦ Synopsis

This open access textbook aims at providing detailed explanations on how to design and construct image analysis workflows to successfully conduct bioimage analysis.

Addressing the main challenges in image data analysis, where acquisition by powerful imaging devices results in very large amounts of collected image data, the book discusses techniques relying on batch and GPU programming, as well as on powerful deep learning-based algorithms. In addition, downstream data processing techniques are introduced, such as Python libraries for data organization, plotting, and visualizations. Finally, by studying the way individual unique ideas are implemented in the workflows, readers are carefully guided through how the parameters driving biological systems are revealed by analyzing image data. These studies include segmentation of plant tissue epidermis, analysis of the spatial pattern of the eye development in fruit flies, and the analysis of collective cell migration dynamics.

The presented content extends the Bioimage Data Analysis Workflows textbook (Miura, Sladoje, 2020), published in this same series, with new contributions and advanced material, while preserving the well-appreciated pedagogical approach adopted and promoted during the training schools for bioimage analysis organized within NEUBIAS – the Network of European Bioimage Analysts.

This textbook is intended for advanced students in various fields of the life sciences and biomedicine, as well as staff scientists and faculty members who conduct regular quantitative analyses of microscopy images.

✦ Table of Contents

Reviewers
Acknowledgement
Contents
Editors and Contributors
1 Introduction
1.1 Introduction
References
2 Batch Processing Methods in ImageJ
2.1 Introduction
2.2 Types of Batch Processing Methods in ImageJ
2.3 Tools
2.4 Dataset
2.5 Core Workflow for Processing a Single Image
2.6 GUI-Based Methods
2.7 Scripting-Based Methods
2.7.1 Preparing the Code for Batch Processing
2.7.2 ImageJ Macro, IJ1
2.7.3 Two Different Methods to Get User Input
2.7.4 ImageJ Macro, Scijava
2.7.5 Command-Line Headless Methods
2.8 Collecting Measurement Results During Batch Processing
2.8.1 Collecting Measurements Within an Array
2.8.2 Collecting Measurements Within a Table
2.8.3 Collecting Measurements When Using SciJava
2.9 Application to Bioimage Analysis
Solutions to the Exercises
References
3 Python: Data Handling, Analysis and Plotting
3.1 Tools to Follow the Chapter
3.2 Why Python?
3.2.1 Python Versions
3.2.2 Python Packages and Environments
3.2.3 Anaconda
3.2.4 Jupyter Notebook
3.3 pandas: Python Data Analysis Library
3.3.1 Syntax: Creating a DataFrame
3.3.2 Basic Numeric Operations
3.3.3 Import Data Using pandas
3.3.4 Reshape the Data: How to Create Tidy Data
3.3.5 Split-Apply-Combine
3.4 Python Visualization Landscape
3.4.1 JavaScript
References
4 Building a Bioimage Analysis Workflow Using Deep Learning 9pt Estibaliz Gómez-de-Mariscal, Daniel Franco-Barranco,[-15.8pt] Arrate Muñoz-Barrutia and Ignacio Arganda-Carreras-4mm
4.1 Why You Should Know About Deep Learning
4.2 Dataset
4.3 Tools
4.4 Workflow
4.4.1 Step 1: Setting up a Google Colaboratory Notebook
4.4.2 Step 2: Download and Split the Data into Training, Validation and Test
4.4.3 Step 3: Train a Deep Learning Model for Binary Segmentation
4.4.4 Step 4: Evaluating the Trained Model
4.4.5 Step 5: Building a DeepImageJ Bundled Model to Process New Data
4.4.6 Step 6: Process All Images in Fiji Using DeepImageJ and MorpholibJ
Appendix
Training Hyper-Parameters
Optimizer
Halo and Receptive Field of a Network
Data Augmentation
Solutions to the Exercises
References
5 GPU-Accelerating ImageJ Macro Image Processing Workflows Using CLIJ
5.1 Introduction
5.2 The Dataset
5.2.1 Imaging Data
5.2.2 The Predefined Processing Workflow
5.3 Tools: CLIJ
5.3.1 Basics of GPU-Accelerated Image Processing with CLIJ
5.3.2 Where CLIJ Is Conceptually Different and Why
5.3.3 Hardware Suitable for CLIJ
5.4 The Workflow
5.4.1 Macro Translation
5.4.2 The New Workflow Routine
5.4.3 Good Scientific Practice in Method Comparison Studies
5.4.4 Benchmarking
5.5 Summary
Solutions to the Exercises
References
6 How to Do the Deconstruction of Bioimage Analysis Workflows: A Case Study with SurfCut
6.1 Introduction
6.1.1 A Workflow and Its Components
6.1.2 What Is Deconstruction?
6.1.3 A Case of Study of Workflow Deconstruction: SurfCut
6.1.4 What Is SurfCut?
6.1.5 What Was SurfCut Developed for?
6.1.6 Other Similar Tools
6.2 Dataset
6.3 Tools
6.4 Workflow
6.4.1 Step 1. Identification of Components in the Textual Description
6.4.2 Step 2. Drawing a Workflow Scheme
6.4.3 Step 3. Assessment of Prerequisites and Limitations
6.4.4 Step 4. Identification of Components in the Code
6.4.5 Step 5. Code Refactoring
6.4.6 Step 6. Replacing a Component: Shift Mask in the Z-Axis Direction
6.4.7 Step 7. Benchmarking: Comparison of Two Alternative Components
6.4.8 Step 8. Linking to Another Workflow: FibrilTool
6.5 Analysis of the Results: Presentation and Discussion
6.6 Concluding Remarks
References
7 i.2.i. with the (Fruit) Fly: Quantifying Position Effect Variegation in Drosophila Melanogaster
7.1 Introduction
7.1.1 What Is the Big Deal with Color Images and Fly Eyes?
7.1.2 How Is PEV Quantified Now and Potential Issues
7.1.3 The Fallacy of Human Perception and Why Automated Analysis of Images Is King
7.2 Dataset
7.2.1 Imaging Conditions
7.2.2 About Image Acquisition, Preprocessing, and Color Normalization
7.2.3 Dataset Download
7.3 Tools
7.3.1 ilastik Configuration in ImageJ
7.4 Workflow Overview
7.5 Step 1. Workflow Preparation
7.5.1 Selecting the Working Directory
7.6 Step 2. Cropping Left and Right Eye Areas
7.7 Step 3. Segmentation by Using Ilastik
7.8 Step 4. Extracting Measurements from the Segmented Objects
7.8.1 Part A: Simple Metrics Using [Analyze Particles...]
7.8.2 Part B: Crowdedness and Organizedness
7.9 Deriving Crowdedness
7.9.1 First compute the Max Triangular Packing Value
7.10 Assessing Organizedness
7.10.1 Computing Pairwise Distances
7.10.2 Ratio of Maximum and Minimum Distances (r)
7.11 Step 5. Exporting the Calculated Metrics into Tables
7.12 Step 6. Batch Processing and Further Considerations
7.12.1 A Fly Head Has Two Eyes
7.12.2 Eyes That Have Zero or One Patch
7.12.3 Step 6.3 Batch Processing into Multiple Folders
7.13 Visualizing Results: Presentation and Discussion
Solutions to the Exercises
References
8 A MATLAB Pipeline for Spatiotemporal Quantification of Monolayer Cell Migration
8.1 Introduction
8.2 Dataset
8.2.1 Experimental Considerations
8.3 Tools
8.3.1 Setting up the MATLAB Environment and Executing the Analysis Pipeline
8.4 Workflow
8.4.1 Pipeline Overview
Parameter Initialization
Output Directory Structure
8.4.2 Part 1: Estimation of Velocity Fields, Semantic Segmentation, and Calculation of Wound Healing Measurements
Estimating Velocity Fields
Segmenting the Cellular Foreground
Calculating the Wound Healing Over Time
Part 1: Outputs
Part 1: Parameter Sensitivity and Trade-Offs
Part 1: Practical Usage of the Outputs
8.4.3 Part 2: Kymographs
Part 2: Outputs
Part 2: Parameter Sensitivity and Trade-Offs
Part 2: Practical Usage of the Outputs
8.4.4 Part 3: Feature Extraction
Part 3: Outputs (See sofullcolor46Table 8.6)
Part 3: Parameter Sensitivity and Trade-Offs
Part 3: Practical Usage of the Outputs
8.4.5 Part 4: Principal Component Analysis: PCA
Part 4: Outputs (See sofullcolor46Table 8.8)
Part 4: Practical Usage of the Outputs
8.4.6 Tips and Troubleshooting for Advanced Users
Solutions to the Exercises
References
A Supplementary Information
Index