The INM endeavors to pursue captivating research projects within a rich and highly heterogeneous environment. The tutorials offered during our retreat have traditionally presented an exceptional opportunity to embrace this diversity and expand one’s understanding of scientific concepts, methodological approaches, and future challenges that extend beyond our daily work routine. In the INM Retreat 2023, we are pleased to announce that researchers from various institutes once again extend their offer to share their knowledge and expertise on selected topics.
Overview
Session | Tutorial | Organizer(s) | Building | Room |
9:00 – 10:45 | Anatomy | Svenja Caspers | 15.9v | 4001b |
Lightweight data publishing on Jülich Data with DataLad | Michał Szczepanik | 04.7u | 338 | |
Imaging | Ravichandran Rajkumar | 16.15 | 2009 | |
9:00 – 13:00 | Deep Learning for Neuroscience | Timo Dickscheid | Part I: 14.6 | Part I: 241 |
Part II: 16.3 | Part II: 211 | |||
11:15 – 13:00 | Imaging | Ravichandran Rajkumar | 15.2 | E1 |
The siibra toolsuit for accessing the EBRAINS human brain atlas | Timo Dickscheid | 15.9v | 4001b | |
Research Data Management and Research Software Engineering | Michael Denker | 04.7u | Lecture hall |
Anatomy
9:00 - 10:45
Svenja Caspers
The tutorial will cover the basic principles of neuroanatomy, including major landmarks for orientation and understanding of basic functions. It will combine insights about the brain’s structure across scales and organizational levels. Accompanying the theoretical foundations, the tutorial will provide hands-on training on virtual and live dissection of brain sections.
Lightweight data publishing on Jülich Data with DataLad
9:00 - 10:45
Michał Szczepanik
Jülich DATA (data.fz-juelich.de) is the central institutional repository for research data of the research center Jülich and supports sharing, preserving, citing, exploring, and analyzing research data with descriptive metadata, without hosting large files. In this tutorial, participants will discover how DataLad (datalad.org) can integrate with Dataverse and have the best of both worlds: Discoverability and metadata with Jülich DATA, and actionable data tracking with DataLad. With conceptual and hands-on elements, we will learn how to publish or clone lightweight DataLad datasets to and from Jülich DATA.
Deep Learning for Neuroscience
9:00 - 13:00
Timo Dickscheid, Christian Schiffer, Alexandre Strube, Sabrina Benassou
Machine Learning - in particular deep learning - has become an indispensable tool for analyzing large neuroscience datasets. The Helmholtz AI team at Jülich is closely connected to these developments and supports research activities at the intersection of AI, high-performance computing (HPC) and neuroscience. Many of the methods and solutions are not limited to neuroscience and medical applications, but can be transferred to different tasks and scientific domains.
This tutorial we will give an overview of state-of-the-art deep learning methods in the context of biomedical image analysis and show concrete examples in INM where deep learning already supports neuroscientists in analyzing their data.
The second part of this tutorial will offer a hands-on course on how to bring deep learning pipelines on JSC’s HPC systems. If you plan to join the hands-on part, please read the prerequisites here carefully.
Schedule
| 9:00 - 9:20 | Deep Learning for image analysis - An overview Timo Dickscheid |
| 9:20 - 9:50 | How Deep Learning helps us decoding the microstructural organisation of the brain Christian Schiffer |
| 9:50 - 10:00 | Coffee Break |
| 10:00 - 13:00 | Hands-on How to run your Deep Learning pipeline on HPC Alexandre Strube, Sabrina Benassou (Coffee Break as required) |
Imaging
9:00 - 13:00
The tutorial on Imaging will provide insights into the basics of Magnetic Resonance Imaging (MRI), functional Magnetic Resonance Imaging (fMRI), Positron Emission Tomography (PET), and Magnetic Resonance Spectroscopy (MRS) and their applications in neuroscience research and clinical practice. The Imaging tutorial will begin with the beginner level introduction to MRI, followed by an introduction to fMRI and current state-of-the-art layer-specific fMRI techniques. The second part of the tutorial begins with a lab tour to provide a firsthand experience of MRI in action. This will be followed by tutorials on introduction to metabolic imaging techniques including PET and MRS.
*Detailed information regarding the lab tour of MR systems will be promptly communicated to registered participants via email.
Part 1: MR Imaging
Basics of MR Imaging:
In this tutorial the fundamental principles and techniques of Magnetic Resonance Imaging (MRI) will be introduced. The MR imaging process will be explained using simple spin physics and fundamental imaging protocols will be introduced. Building on this knowledge, we will briefly explore the advantages and disadvantages of ultra-high field MRI systems (7 Tesla and above).
Basics of fMRI with focus towards layer fMRI with EPIK:
This tutorial begins by explaining the fundamental principles behind fMRI and how it measures changes in blood oxygenation to infer neural activity. A special focus will be placed on Layer fMRI using inhouse developed EPIK sequence, an emerging technique enabling investigation of activity within specific layers of the cerebral cortex. In addition to exploring the principles and applications of Layer fMRI, this tutorial will also delve into the crucial aspect of data preprocessing in Layer fMRI studies. Participants will gain insights into the essential preprocessing steps, including motion correction, spatial normalization, and noise reduction techniques, ensuring the generation of accurate and reliable brain activity maps at the layer-specific level.
Part 2: Metabolic Imaging
PET Imaging Basics and Amino acid PET in brain Tumor diagnostics:
This tutorial will start with a solid foundation on Positron Emission Tomography (PET) imaging basics, explaining how it works, and its significance in modern medicine. A significant portion of this tutorial will be dedicated to exploring the specialized use of amino acid PET in brain tumor diagnostics. Brain tumors present unique challenges in their detection and characterization, making traditional imaging techniques often insufficient. However, with the advent of amino acid PET tracers, such as Fluoroethyltyrosine (FET), a new dimension of accuracy and precision has been added to brain tumor imaging. Throughout this tutorial, the advantages and limitations of amino acid PET imaging will be discussed in comparison to other imaging modalities and showcase real-life case studies that demonstrate its clinical efficacy.
MR Spectroscopy:
MR Spectroscopy is a powerful non-invasive technique that provides valuable insights into the biochemical composition of tissues. By harnessing the principles of Magnetic Resonance Imaging (MRI), MRS offers a unique window into the metabolites present in living organisms. In this tutorial, the fundamentals, and applications of MRS will be introduced, uncovering its pivotal role in research and clinical settings.
Schedule
| Session I – MR Imaging | |
| 9:00 - 9:45 | Basics of MRI Dr. Felder |
| 9:45 - 10:00 | Discussion |
| 10:00 - 10:45 | Intermediate level topic (fMRI with focus on Layer fMRI) Dr. Küppers and Dr. Pais-Roldán |
| 10:45 - 11:15 | Coffee Break (with possibility to interact with Speakers) |
| Session II – Metabolic Imaging | |
| 11:15 - 11:45 | Lab Tour (MRI systems) |
| 11:45 - 12:15 | PET Imaging Basics and Amino acid PET in brain Tumor diagnostics Prof. Langen |
| 12:15 - 12:45 | MR Spectroscopy Dr. Ali Gordjinejad |
The siibra toolsuit for accessing the EBRAINS human brain atlas
11:15 - 13:00
Timo Dickscheid, Kim Lothmann
siibra is a software tool suite implementing an openly accessible brain atlas framework which connects multimodal datasets from different resources to anatomical structures in reference spaces at different spatial scales. The tool suite is designed to address both interactive exploration through an interactive 3D web viewer (siibra-explorer) as well as integration into data analysis and simulation workflows with a comprehensive Python library (siibra-python). In this session, we first introduce the multidimensional concept of the atlas framework and explore some key features such as the BigBrain interactively. We then turn to concrete programming tutorials in Python. These include fetching brain region maps, accessing the BigBrain dataset, and extracting multimodal regional features such as cortical thicknesses, cell and neurotransmitter densities, gene expressions and connectivity data. We will finish with some concrete data analysis examples.
Research Data Management and Research Software Engineering
11:15 - 13:00
In this tutorial, we will cover two introductory topics in the context of managing research data and creating research software. In the first part of the tutorial, we will introduce the NIX file format and accompanying software to demonstrate hands-on the use of data models to produce self-documenting descriptions of research data that are richly annotated with metadata. To this end, we will focus on representing an electrophysiological dataset and analysis results in a way suitable for sharing with a collaboration partner. In a second part, we will focus on practical aspects of implementing measures to increase the quality of research software and a primer on how to contribute to open software, including aspects of version control, versioning, creating and reviewing code contributions, and software testing.
