Better medical prognoses and treatments thanks to AI in medical scanners

(19-11-2021) In his PhD, Milan Decuyper investigates how the resolution of detectors and medical scanners can be improved so that better prognoses and medical treatments become possible in the future.

Driven by the ever-increasing amount of computing power available and digital data generated, artificial intelligence (AI) systems are increasingly finding their way into our daily lives.

"The need and potential of AI is also emerging in medical care," Milan Decuyper explains, "Electronic medical records contain a wealth of information that can be used for personalized and precision medicine. Due to the immense amount and complexity of this data, especially within medical imaging, it is not possible to fully utilize all this information."

For this reason, AI algorithms are being developed to enhance the radiological workflow. In his doctoral research, Milan focuses on two applications within medical imaging.

"The first is at the beginning of the imaging process (the acquisition phase) where we use neural networks to improve the spatial resolution (location accuracy) of positron emission tomography (PET) detectors and thus scanners," Milan explains, "We optimized the complexity and training procedure of the networks and achieved better performance compared to an established algorithm."

The second application is at the very end of the imaging process. "We developed convolutional neural networks (layer in neural networks that can also detect more abstract things) that can non-invasively, automatically and accurately segment and diagnose primary brain tumors based on pre-therapy magnetic resonance (MR) scans. A computer-aided diagnosis is important for determining prognosis and optimal treatment strategy," Milan concludes.

Read the entire PhD

-

PhD Title: Advancing Medical Imaging with Artificial Intelligence: PET Acquisition Enhancement and MRI-Based Brain Tumour Diagnosis

-

Milan Decuyper

Milan Decuyper, born in Bruges on May 23, 1994, started studying civil engineering at Ghent University in 2012. In July 2017, he obtained the degree of Master of science in electrical engineering: communication and information technology with high distinction. In August of that year, he started as a PhD student on a BOF research project at the MEDISIP (Medical Imaging and Signal Processing) research group within the Faculty of Engineering and Architecture. Under the supervision of the supervisors, Roel Van Holen and Stefaan Vandenberghe, he conducted research on the use of artificial intelligence to enhance medical imaging. Milan is first author of two publications and co-author of one publication in international scientific journals. He has also presented his work at eight national and international conferences through proceedings and presentations.

Contact: Milan Decuyper, Roel Van Holen, Stefaan Vandenberghe 

-

Editor: Jeroen Ongenae - Final editing: Ilse Vercruysse - Illustrator: Roger Van Hecke