Dr. Olivier Debeir directs the LISA/image unit at ULB, Belgium, and is a specialist in biomedical image processing. During his 1 month secondment at INEN, Lima, he gave a 2 day workshop for INEN staff, described below.
During day I we introduced the two approaches developed at the Lisa for the MRI in vivo liver assessment. The first approach studies the specific contrast agent Primovist, known to be metabolized by the healthy hepatic cells. The relative and absolute liver volume is evaluated using a 3D volume registration before and after contrast agent uptake.The second approach aims to enable artefact free diffusion weighted imaging (DWE) which is usually impaired by the organ motion (respiration and heart). While respiration motion can be controlled using breath-hold, the heart beats are constant. The proposed method exploits a motion amplification framework adapted to real time CINE MRI sequences. This allows to better detect the moment (80ms long) during the cardiac cycle when the liver is mostly still. This approach also enable to build a displacement map of the organ of interest, here the liver, in order to detect possible inhomogeneities. This approach will be tested at the INEN on unusual liver pathologies observed in Peruvian patients.
On day II, the workshop presented several techniques, recently developed in the machine learning domain, that made a breakthrough in the image processing domain, in particular in the medical imaging domain: the deep neural networks. After a general introduction, several application specific examples are shown involving different medical imaging modalities.