© L. Fisch

ML in Medicine

AI methods have facilitated work in many areas of medicine. Tasks such as tumor detection in CT scans or melanoma classification in dermatology can nowadays be automated to a large extent by machine learning. In addition, machine learning can assist medical staff in diagnosis by aggregating medical data into biomarkers. For example, machine learning is being used to estimate the biological age of individuals based on MRI images of the brain. The difference between the estimated and actual age is the used as a biomarker that can indicate neurological disease.

Courses

In the module "Machine Learning in Medicine", the methods of machine learning and their application to medical data are introduced in two consecutive block seminars "Introduction to Medical Machine Learning" and "Advanced Course in Medical Machine Learning". The first course provides a beginner-friendly introduction to machine learning for students with little prior knowledge on programming. Students with prior knowledge in computer science and completers of the first course will be given the opportunity to deepen their skills in the second course.

Both courses can be attended online or in person. Following the examination regulations of the medical faculty, a certificate of credit can be issued after active participation (attendance for at least 85% of the hours).
Furthermore, it is possible to use the course independent of the semester, i.e. to study the course content (Jupyter notebooks) at your own time and convenience.

Introduction to Medical Machine Learning

AI methods have greatly facilitated work in many areas of medicine. Tasks such as tumor detection in CT scans or melanoma classification in dermatology can nowadays be automated to a large extent by Machine Learning. This seminar not only covers the principles of Machine Learning, but also introduces the basics of Python as well as the most important packages like NumPy and Matplotlib. This enables participants without previous programming experience to attend the seminar.

Contents:

  •     basics of programming in Python,
  •     arrays and tables (NumPy, pandas),
  •     visualization of data (Matplotlib, Seaborn),
  •     fundamentals of machine learning (fastai, sklearn, photonai).

No previous experience is necessary to participate in the course.

Course materials are freely available as Google Colab notebooks here.

Advanced Course in Medical Machine Learning

The second course focuses on the application of neural networks to different data modalities (e.g. 3D MRI images, audio files). For this purpose, the Python packages PyTorch and fastai are introduced, which enable the efficient application of neural networks. Different architectures of neural networks and their fields of application will be investigated and the necessary data preprocessing will be applied. Concepts of machine learning and programming from the first course will be taken up and extended, e.g. with methods of object-oriented programming and data augmentation.

This course will take place for the first time in the winter semester 2023/24.