02985 Summer school on biomedical image analysis – from acquisition to fairness and bias

2023/2024

In order to participate in the course, you need to register on the homepage and pay the participation fee:
https:/​/​biomed.compute.dtu.dk/​
Kursusinformation
Summer school on biomedical image analysis – from acquisition to fairness and bias
Engelsk
3
Ph.d., Fagligt fokuseret kursus
Kurset udbydes som enkeltfag
August
August 14. - 16. 2024
Hotel Christiansminde in Svendborg, Denmark
The course is a combination of lectures and practical programming exercises.
[Kurset følger ikke DTUs normale skemastruktur]
The evaluation is on August 15, 2024
Mundtlig eksamen og bedømmelse af øvelser
To pass the course you need to bring a poster and present it at a poster session, participate in group based exercises in form of a larger programming challenge, present their results of the exercises, be evaluated as pass/not passed by the summer school organizers (university faculty).
15-20 minutes per team
Alle hjælpemidler er tilladt
bestået/ikke bestået , intern bedømmelse
Practical experience with biomedical image analysis, machine learning and deep learning.
Minimum 30 Maksimum: 100
Rasmus Reinhold Paulsen , Lyngby Campus, Bygning 324, Tlf. (+45) 4525 3423 , rapa@dtu.dk
Anders Nymark Christensen , Lyngby Campus, Bygning 324 , anym@dtu.dk
01 Institut for Matematik og Computer Science
Department of Computer Science, University of Copenhagen
Aalborg University
https://biomed.compute.dtu.dk/
I studieplanlæggeren
June 2024
Overordnede kursusmål
Biomedical image analysis is a broad topic that covers a variety of disciplines in modern computer science, AI Research, medical sciences and biology. In this summer school, we aim at given the participant a good overview of the current state-of-the-art in several of the fundamental topics used in modern biomedical image analysis.
Læringsmål
En studerende, der fuldt ud har opfyldt kursets mål, vil kunne:
  • Describe the main ways of biomedical image acquisition
  • Import and visualize large 3D data sets using Python
  • Describe and evaluate state-of-the-art 3D high-energy scanning approaches
  • Do basic analysis of human motion data
  • Describe and evaluate the use of motion tracking systems for 3D image corrections
  • Implement and use implicit methods for 3D data representations
  • Describe and use different parameterizations of complex 3D shapes
  • Describe the concept of fairness and bias in relation to medical image analysis
  • Implement and use a selection of methods for 2D and 3D structural analysis
Kursusindhold
Biomedical image analysis typically starts with the acquisition of data. This can for example be large 3D volume data from high-energy scanners, human motion data from optical or infrared sensors, cell images from microscopes or standard historical photos of species. We have invited a group of speakers that are specialist in several of these techniques.

A common issue with complex data is how to represent the data in a compact way but where the information is still preserved. Currently, there is a large research interest in implicit representation of 3D shapes for machine learning. These representation and other suitable ways of handling complex biomedical data is also presented at the summer school.

While much modern image analysis is based on large deep neural networks, there is still a need for knowing about core methods like advanced 3D morphological analysis and statistics of shapes. There will be expert speakers in this field.

With the modern data driven approaches it can be very hard to judge the fairness and biases of the models. We will have a session on these complex topic.

The course will include group based project work, where the participants make a programming project relating their research to the summer school’s topics. It will be in the form of a challenge.
Litteraturhenvisninger
Please see the homepage:
https:/​/​biomed.compute.dtu.dk/​talks-materials-and-exercises/​
Bemærkninger
In order to participate in the course, you need to register on the homepage and pay the participation fee:
https:/​/​biomed.compute.dtu.dk/​
Sidst opdateret
19. marts, 2024