Doctoral defence: Daniel Majoral Lopez "Deep neural networks for microscopy images“

On January 23 at 11:00 Daniel Majoral Lopez will defend his doctoral thesis "Deep neural networks for microscopy images" to obtain the degree of Doctor of Philosophy (in Computer Science).

Supervisors:
Dr. Leopold Parts, Wellcome Sanger Institute (UK), University of Tartu
Professor Raul Vicente Zafra, University of Tartu

Opponents:
Professor Nataša Sladoje, Uppsala University (Sweden)
Dr. Craig Glastonbury, Human Technopole (Italy)

Summary
Progress in artificial intelligence has increased the range of things a computer can do. The thesis explores how to apply artificial intelligence to images of cells seen through a microscope. The thesis includes three different scientific articles. The first article discusses how to detect cells in microscopy images obtained by illuminating with white light. This illumination is known as Brightfield microscopy and is simple and cheap, but it is difficult to see cells in the images obtained. The authors found how to use artificial intelligence to see the cells in these images and provide practical advice, helping other people do it too. The second article develops an artificial intelligence method to detect the cells that transform into sperm. A cell has a complicated process of becoming sperm undergoing a lot of changes, the artificial intelligence can tell in which state of the process it is. Infertility affects 10-15% of couples and hopefully, artificial intelligence analysis of images can help to find out the problem affecting the couple. The third article describes a new algorithm called Kaizen, which uses artificial intelligence to detect cells in varied microscopy images. Instead of detecting cells, Kaizen generates images of individual cells from a microscopy image. The individual cell images are merged into a composite image. This composite image is then compared to the microscopy image allowing detection of errors and non-predicted objects. These three articles show how artificial intelligence can improve the analysis of microscopy images. Hopefully, in the coming years, artificial intelligence will spearhead breakthroughs in understanding cellular structures and disease mechanisms, to improve the human condition.

The defence will be held also in Zoom (Meeting ID: 959 0941 7702, passcode: ati).