MICCAI 2016 Daily - Wednesday

MICCAI Daily: Stéphanie Marchesseau, what is your current position? Stéphanie: I currently work as a research fellow at the Clinical Imaging Research Centre in Singapore. MICCAI Daily: What is the work that you are presenting here? Stéphanie: Here, I present a work that combines registration and segmentation in order to be unbiased by the segmentation and by the registration. In other words, usually you have two methods to do local deformation or local tracking. You have either t he 4D segm entation which can be done manually, or (what is commonly used) registration algorithms, which is only image-based. You can apply this registration algorithm to one frame of the segmentation. Usually the end-diastole. The main problem is that it’s very biased that you use this frame while using the end-diastole, and not the end-systole. What I try to show here is that there is a big discrepancy, and we should not discard all the other frames in the segmentation. Instead, we should use all of the framework segmentation and combine with the registration to be much more robust. MICCAI Daily: What is the medical value of what you are doing? Stéphanie: There is what I show in this poster, but there are also other applications. We will only be able to tell where is the scar from just seeing the MRI images, which is not a sequence for tissue characterization. When you have dead muscle, and of course the behavior is different for this. So we want to be able to measure it. The usual techniques to do it are not that robust yet. They are biased toward which frame you are going to use. MICCAI Daily: Is it difficult to do that? Stéphanie: It’s not difficult. What I suggest is actually very simple. You use exist ing things because you Segmentation And Registration Coupling From Short- Axis Cine MRI: Application To Infarct Diagnosis Stéphanie Marchesseau Presentation 14 MICCAI Daily: Wednesday