Image processing for correlated and multimodal imaging techniques
Image processing for correlated and multimodal imaging techniques
General context
Within the context of the COST action COMULIS (Correlated Multimodal Imaging in Life Sciences, CA17121) we are organising a training event on ‘Image processing for correlated and multimodal imaging techniques’.
COMULIS (Correlated Multimodal Imaging in Life Sciences) is an EU-funded COST Action that aims at fuelling urgently needed collaborations in the field of correlated multimodal imaging (CMI), promoting and disseminating its benefits through showcase pipelines, and paving the way for its technological advancement and implementation as a versatile tool in biological and preclinical research.
The purpose of this event is to dedicate three days of intense training to an audience with little or basic expertise. The program will first cover a full day of online lectures (19 October 2020), followed by two days of hands-on training (8 & 9 March 2021, Ghent).
Target audience has little or basic expertise in the field.
Saskia Lippens, VIB Bioimaging Core, Ghent
Sebastian Munck, VIB Bioimaging Core, Leuven
Raphaël Marée, Univeristy of Liège
- Participation in day 1 (lectures) is free for everyone and will be organized online.
- Registration fee for the hands-on sessions is € 100 (21% VAT included).
- To participate in the hands-on sessions in 2021, you will be able to upload your CV and abstract later this year. The number of participants will be limited to 20. If you only want to attend the lectures, it is not necessary to submit your motivation and CV.
Trainers
Marie-Charlotte Domart
Christian Van Hove
Ilya Belevich
Noeska Smit
Joris Roels
Perrine Paul-Gilloteaux
Raphaël Marée
Program
Lectures - Online
Opening of the training
Registration
Practical info on the online workshop
Welcome by the scientific committee
CLEM
Marie-Charlotte Domart, Francis Crick Institute, London, UK
Correlative light and electron microscopy (CLEM) combines the benefits of fluorescence and electron imaging, revealing protein localisation against the backdrop of cellular architecture. The correlative imaging field is expanding rapidly, and encompasses workflows that link many different imaging modalities. We link fluorescence microscopes (widefield, confocal, super-resolution and light-sheet) with electron microscopes (scanning, transmission, serial block face and focused ion beam) [1] and X-ray microscopes (microCT and soft X-ray) to analyse a range of biological samples, from single cells to whole model organisms. Depending on the biological question, correlative workflows can be tailored to incorporate almost any imaging modality.For example, in collaboration with Eva Frickel at the Crick, we combined super-resolution confocal microscopy with serial section TEM to image Toxoplasma gondii-infected human macrophages, and confirmed that GBP1 plays a key role in parasitophorous vacuole rupture upon infection [2]. To study the innate immune response to Toxoplasma in vivo,we developed a new correlative multimodal imaging approach, using a combination of light, X-ray and 3D electron microscopy techniques [3]. Since no single imaging technique can reveal all these details, the new approach we developed, combining five different imaging modalities, played a key role in the gathering of multidimensional information about our zebrafish model and its microscopic structure and function. This targeted approach revealed that Toxoplasma invades and replicates inside a parasitophorous vacuole, and recruit host cell mitochondria. Strikingly, macrophages are recruited to the infection site and play a key role in Toxoplasma control. These results highlight the benefits of multimodal approaches, and open the possibility to exploit zebrafish for discoveries within the field of parasite immunity.
References:
[1] Peddie, C.J. & Collinson, L.M. Exploring the third dimension: Volume electron microscopy comes of age. Micron. 61, 9-19. (2014).
[2] Fisch D, Clough B, Domart MC, Encheva V, Bando H, Snijders AP, Collinson LM, Yamamoto M, Shenoy AR & Frickel E. Human GBP1 differentially targets Salmonella and Toxoplasma to drive recognition of microbial ligands and caspase-mediated death. Cell Rep. 32(6):108008. (2020).
[3] Yoshida N, Domart MC, Peddie CJ, Yakimovich A, Mazon-Moya MJ, Howkins T, Collinson LM, Mercer J, Frickel E & Mostowy S. In vivo control of Toxoplasma gondii by zebrafish macrophages. Dis Model Mech. 13(7):dmm043091. (2020).
Multimodal imaging
Christian Van Hove, UGent Institute for Biomedical Technology, BE
In 1895 the German physicist Wilhelm Conrad Roentgen discovered the X-rays, an achievement that earned him the first Nobel Prize in Physics in 1901. These X-rays produced the first medical images in the beginning of the previous century. Nowadays, a wide range of medical imaging devices are available that can be roughly sub-divided into two main categories: structural and functional imaging devices. The most commonly used structural imaging technologies are computed tomography (CT), magnetic resonance imaging (MRI) and ultrasound imaging (US). These techniques were developed to non-invasively visualize macroscopic anatomical and physiological changes in tissues. Functional imaging modalities, such as single photon emission tomography (SPECT) and positron emission tomography (PET) focus on the visualization of molecular/cellular targets in living subjects, such as transporters, cell surface receptors and intracellular enzymes. Important advantages of these functional imaging techniques include early disease detection and early assessment of treatment response, and both can improve survival and reduce costs. However, these functional imaging devices are hampered by low spatial resolution and inability to provide anatomic detail. Therefore, at the beginning of this century, a functional and an anatomical imaging device were integrated into single hybrid device (e.g. SPECT/CT, PET/CT and PET/MRI), maximizing their separate strengths and providing anatomic localization of functional processes in a single examination. The availability of multimodal (hybrid) medical imaging improves our ability to characterize lesions and to affect treatment decisions and patient management. We have just begun to exploit the truly synergistic capabilities of multimodal medical imaging. Continued advances in instrumentation and functional imaging probes development will improve our ability to noninvasively characterize disease processes. In recent years, powerful imaging tools have been developed to detect diseases at an early stage of their development, to assess patients prognosis, stratification and monitoring treatments, and for image-guided therapy. These applications may be improved by using functional imaging probes that allow for multimodal molecular imaging, in particular, by combining different types of imaging modalities that simultaneously furnish information on multiple physiological and anatomical aspects of diseases. Analysis of these high-content image data would assist physicians to diagnose better and treat diseases.
Q&A and break
Segmentation
Ilya Belevich, University of Helsinki, FI
Image segmentation is an important step required for quantitative analysis and visualization of objects of interest. In my talk, I will explain the basics of image segmentation, why it is beneficial for a research project, what are the types of segmentation models and discuss various segmentation strategies. I will discuss semi-automatic tools, such as local thresholding, morphological operations and graphcuts - that significantly ease and speed-up the segmentation process. The modern, completely automatic machine learning techniques, such as image classifiers or convolutional neural networks (CNNs), are capable to improve the segmentation process even further. To illustrate that, I will show practical examples on how to train CNNs on own data and apply the trained network to segment new datasets. Finally, I will introduce Microscopy Image Browser (MIB) as a full workflow software package suitable for image processing, segmentation, quantitation and visualization of results.
Visualization
Noeska Smit, University of Bergen, NO
In the now present information age, data is acquired at an increasing rate which is often faster than our ability to analyze it thoroughly with conventional methods. Computer-based visualization systems provide (usually interactive) visual representations of datasets intended to help users carry out their tasks more effectively and efficiently. In many disciplines, we see the need for visualization in order to explore, analyze, and present information. In this lecture, basic visualization principles and their practical applications will be discussed. You will learn basic visualization theory illustrated by practical examples in order to make your own effective visualizations.
Q&A and break
Big data and registration
Joris Roels, VIB-UGent Center for Inflammation Research, BE
Data is becoming more and more important in life science research. Given the highly automated nature of current image analysis pipelines, it is crucial to maximize data quality. In this lecture, we illustrate how noise sources such as underexposure, blur, stitching artifacts, etc. in high throughput data acquisition and processing can affect data analysis. We study the complete imaging pipeline from acquisition to analysis and show where and how these sources intervene with the data. Moreover, we discuss image processing techniques from domains such as denoising, deconvolution, registration, super-resolution, etc. that can significantly improve data quality. Additionally, we provide guidelines on how to apply these methods correctly, and discuss their limitations, by means of illustrative examples.
Quantification
Perrine Paul-Gilloteaux, University of Nantes, FR
In this lecture, we will focus on the extraction of quantitative information from correlated multimodal datasets. We will review the different categories of questions relying on the quantification of correlated multimodal datasets, from the validation of a new imaging modality to the integration of heterogeneous information, based on examples. A particular focus will be done on the confidence estimation on these quantification, and the different possibilities to quantify this confidence in the correlation, in the medical and biological fields.
Q&A and break
Future challenges
Raphaël Marée, University of Liège, BE
Q&A and wrap-up
Hands-on session on MIB
Hands-on session on eC-CLEM in ICY and Elastix
Practical info
07 March 2022 - 08 March 2022
Ghent - VIB/UGent FSVM I
Technologiepark 71
9052 Zwijnaarde
Belgium
07 March 2022 - 08 March 2022
Ghent - VIB/UGent FSVM I
From Ghent Sint-Pieters station, you can take bus 49, 50 or 70 to Technologiepark. Please check Routeplanner De Lijn for schedules.
07 March 2022 - 08 March 2022
Ghent - VIB/UGent FSVM I
There is only one entrance to Technologiepark. You need to take a ticket at the entrance that can be validated at the reception desk of the FSVM building. Parking will only be allowed in regular parking spots (for instance in front of the building) and in the new parking tower. Parking alongside the roads or in other places where there is no regular parking is prohibited and after initial warnings, fines will be issued (50 euros).
07 March 2022 - 08 March 2022
Ghent - VIB/UGent FSVM I
+32 9 331 36 00
Participants will receive more information on how to access the online platform to attend the training a few days before the start of the training.
We will use Hopin as online platform:
- For more information on the use of this platform, check this video.
- Please close other apps to maximize band width and use Chrome, Firefox or Safari (iOS) as browsers.
- Here you can find some additional trouble shooting tips.
Next to the talks of the speakers that will take place on the stage, make sure you don't miss out on the Q&A and interactive discussions with the speakers in the sessions! Recordings will be available and send together with evaluations.
We hope you will take the opportunity to network with the other attendees and with the speakers. Go to networking (open at all time) and meet new people or invite people you want to have a private chat with or a small group chat of up to five persons (click on the 'people' tab and invite whoever you want to meet).