ISMRM23 - Advances in Image Analysis: How to Get the Most from Your Images

ISMRM & ISMRT Annual Meeting & Exhibition • 03-08 June 2023 • Toronto, ON, Canada

Advances in Image Analysis: How to Get the Most from Your Images

Weekend Course

ORGANIZERS: HyungJoon Cho, Janine Lupo, Andrew Scott

Sunday, 04 June 2023

716A/B

07:45 - 11:30

Moderators:
Core principles in image analysis: Bettina Baeßler & Ilwoo Park
Advanced image analysis methods: Suchandrima Banerjee & Jing Liu

Skill Level: Basic to Advanced

Session Number: WE-21

CME Credit

Session Number: WE-21

Overview
This 4-hour course will cover a range of image processing-related topics, starting with methods for improving image quality and ensuring that your images are well aligned. The first half of the course finishes with talks discussing strategies for fitting models to our MRI data to obtain quantitative parameters and different approaches for analyzing our resulting parametric maps. The second part of the course covers more advanced techniques including traditional machine learning analysis methods, network-based analysis of structural and functional brain MRI data, and a tutorial on radiomics.

Target Audience
Clinicians and engineers with an interest in image analysis and scientists focused on image processing.

Educational Objectives
As a result of attending this course, participants should be able to:
- Describe methods available for pre-processing MRI data and evaluating data quality;
- Identify the most appropriate technique(s) for analyzing different types of MRI data; and
- Prepare initial studies in connectomic and radiomic analyses.

		Core principles in image analysis
07:45		Methods for Improving Image Quality: A Diffusion MRI Perspective View Presentation Video M. Okan Irfanoglu Keywords: Image acquisition: Image processing, Image acquisition: Artefacts, Image acquisition: Motion correction MR images suffer from several imperfections including (but not limited to) low SNR, artifacts and image distortions. In this seminar, we will first discuss whether manipulating images with image processing or machine learning techniques is beneficial or damaging, and then we will go over techniques (and tools) to improve image quality for each of these issues. Our examples will mostly focus on diffusion MRI but what can be done for other modalities such as anatomical or functional images will also be discussed and illustrated
08:15		Image Registration View Presentation Video Julio Carballido-Gamio Keywords: Image acquisition: Image processing The applications of *Image Registration* in medical imaging are numerous, ranging from the alignment of longitudinal scans to segmentation of anatomical structures. In this session, I will cover material that will help you understand the different components of the intensity-based image registration framework.
08:45		Basic Principles of Model-Based Fitting for MRI View Presentation Video Dong-Hyun Kim Keywords: Image acquisition: Modelling Model-based fitting is a powerful technique for extracting quantitative information from MRI data. We will discuss the challenges associated with model-based fitting, including model selection, noise and artifacts, parameter estimation, validation and reproducibility, and computation time.
09:15		Voxel vs. ROI-Based Statistical Analyses: From Histograms to Patients View Presentation Video Kyrre Emblem Keywords: Image acquisition: Image processing, Transferable skills: Reproducible research, Image acquisition: Visualization Making good use of MRI data from a clinical study can be a challenge, especially when faced with the task of analyzing data from advanced imaging techniques in small patient cohorts. This talk will address some of the current challenges with image analyses in a clinical setting. Using neuroimaging and cancer as examples, the talk will discuss potential strategies to help produce and evaluate robust, repeatable, and clinically meaningful image parameters in patient studies with the typical low sample size. Different approaches for assessing and analyzing resulting parametric maps will be presented, including use of dynamic and longitudinal imaging data.
09:45		Break & Meet the Teachers
		Advanced image analysis methods
10:00		Machine Learning: What to Use & When View Presentation Video Peter LaViolette Keywords: Image acquisition: Machine learning In this seminar I will discuss many ways that machine learning (ML) can be used for image processing, from model fitting to image segmentation. Many approaches vary depending on the application. A broad overview using radio-pathomics as an example will be discussed.
10:30		Network-Based Analyses: Graph Theory for Evaluating Brain Connectivity View Presentation Video Olga Tymofiyeva Keywords: Neuro: Brain connectivity I will start by sharing my dream of a clinical application of network-based analyses: a short MRI scan for adolescents in primary care that would enable 1) brain network-based diagnosis of psychiatric disorders, 2) brain network-based treatment of psychiatric disorders, and 3) brain network-based prevention of psychiatric disorders. I will then talk about the challenges of the two main steps in constructing a brain network: i) choosing connectivity measures that will serve as the network "edges" and ii) dividing the brain into regions that will serve as the network "nodes." Various local and global network properties will also be discussed.
11:00		Tutorial: Practical Considerations When Performing Radiomics View Presentation Video Masoom Haider Keywords: Body: Body, Cross-organ: Cancer, Image acquisition: Machine learning Radiomics has not yet made it into clinical care. For this to happen high quality biomarker technical validation is required. In this presentation, we go through key items to consider for meaningful radiomics research. This includes the availability of high-quality source data, proper technical validation methodology, good-quality analysis, external validation, a clear understanding of the use case.