Fabio Priante 1 and Filippo Federici2,3
1Chemistry, Aalto University, Espoo, Finland
2Nanolayers Research Computing LTD
3Applied Physics, Aalto University, Espoo, Finland
This session offers a basic introduction to machine learning (ML) techniques for microscopy image analysis. First, essential ML concepts and model training procedures are outlined using PyTorch. The tutorial then moves to practical implementations, showcasing the use of a Convolutional Neural Network (CNN) for classification tasks, with high-resolution AFM images as an illustrative example. Subsequently, the common challenge of image segmentation is addressed. Considering the problem of identifying cell nuclei in optical microscopy images, a traditional computer vision approach is first attempted, and then compared with modern deep learning methods, particularly with the popular U-Net CNN architecture. The tutorial aims to equip participants with practical knowledge, that could be readily applied to their own microscopy problems.