MICCAI 2019

MICCAI 2019 DAILY Workshop Poster Kerstin Kläser is a third-year PhD student at UCL in London under the supervision of Sébastien Ourselin and Jorge Cardoso. She speaks to us following her poster session yesterday at the SASHIMI workshop. The work she presented explores MR to CT synthesis for PET/ MR attenuation correction. It hypothesises that the main aim of PET for CT synthesis when used for PET attenuation correction is that we shouldn’t minimise the error in the CT image, but we should actually make sure that the error in the resulting PET image is as low as possible. She explains that they developed a network that generates pseudo CTs in three stages. The first stage is a feedforward network that minimises an L2-loss. This is a well-known method. The second stage has a second network that predicts the PET residuals that would result when this initial pseudo CT would be used for Improved MR to CT Synthesis for PET/MR Attenuation Correction Using Imitation Learning attenuation correction for the PET. Basically, the second network can be described as a function that can estimate the residual in the resulting PET. In the third stage, they train a combined network of the two. "MR to CT synthesis for PET/ MR attenuation correction" "we shouldn’t minimise the error in the CT image, but we should actually make sure that the error in the resulting PET image is as low as possible." 12