ICube MSII Seminar: Evolution of Image Analysis for Microscopy: From Classical Methods to Artificial Intelligence

Sidnei PACIORNIK: Full Professor at Pontifícia Universidade Católica do Rio de Janeiro

This seminar will take place on Monday, 15th December at 2:00 PM in the A302 amphitheater at the ICube Illkirch site (300 boulevard Sebastien Brant, 67400 Illkirch).

Title: Evolution of Image Analysis for Microscopy: From Classical Methods to Artificial Intelligence

Abstract: In recent decades, digital image processing and analysis has been instrumental in automating the quantification of microstructures obtained by different types of microscopy, eliminating the uncertainties associated with human-operator-dependent techniques. The classic methodology (pre-processing => segmentation => post-processing => quantification => classification) allows you to solve a wide variety of problems, but has several limitations in more complex images.
With the accelerated development of Artificial Intelligence, new methods based on Deep Neural Networks (Deep Learning), in particular Convolutional Neural Networks (CNNs), have brought about a real revolution.
In this seminar we will introduce the classical methodology and compare it with AI-based methods (e.g. Segment Anything and YOLO), showing how new methods are able to solve previously unsolvable problems. In addition, the use of AI brings great generalization capacity, allowing us to apply the same algorithmic flow to different quantitative microscopy problems.