Grant STW Perspectief for deep learning for medical image analysis

Prof. Josien Pluim and Dr. Mitko Veta are part of the research program Deep learning for medical image analysis that aims to support medical professionals in diagnoses and treatment by means of medical image analysis systems. The consortium wants computers to learn how to better analyze the large numbers of images made in hospitals, for example x-rays, CT scans and histological slides. The program will receive 3.4 million from STW, the partners will contribute 1.4 million euros. The research group Medical Image Analysis of Prof. Josien Pluim will appoint two PhD candidates on the project.

STW Perspective

Technology institute STW approved six large research programs that have to lead to innovative technologies. These programs aims at a more energy friendly Internet of Things, curing arthrosis, more efficient protection against floods, better imaging support for clinicians, new ways to tame light and a brain prosthesis that will make blind people see again. STW provides over 17 million euros. Social organizations, start-ups, small companies and big players like NXP, Philips, ASML and Thales invest over 8 million euros.

The TU/e is involved in three programs: a more energy friendly Internet of Things, better imaging support for clinicians and new ways to tame light. Together they receive 9 million euros. The research programs have a duration of four to six years.

Research description DLMedIA: Deep Learning for Medical Image Analysis (in Dutch)

Programmaleider: Prof.dr. B. van Ginneken, Radboudumc

Deelnemers: Canadian Institute for Advanced Research, COSMONiO, Delft Imaging Systems, Erasmus MC, Philips Electronics, Philips Healthcare, Pie Medical, Quantib, Radboudumc, ScreenPoint Medical, Technische Universiteit Eindhoven, Thirona, UMC Utrecht, Universiteit van Amsterdam

Röntgenfoto’s, CT-scans, mammografieën: beeldverwerking is een hoeksteen van de geneeskunde. Maar tegenwoordig komen er zoveel beelden op artsen en analisten af dat ze hulp moeten krijgen van computers om de beelden te interpreteren. Het DLMedlA-consortium wil computers zelf laten leren hoe ze betere analyses kunnen maken. Dat gaat met behulp van nieuwe, veelbelovende ‘convolutional networks’. Dat zijn speciale kunstmatige, neurale netwerken die, zo denken de onderzoekers, op termijn de huidige computeroplossingen kunnen verslaan. Het consortium wil uiteindelijk de volgende generatie medische beeldverwerkingssystemen ontwikkelen die medici kunnen helpen bij hun diagnoses en behandelingen. Dit programma valt binnen het Innovative Medical Devices Initiative (IMDI.nl).