ISBI Daily - Thursday

soundwaves and you get an echo off whatever is there which forms the image. The strength of the signal that you have to create the image depends on the angle of incidence of the ultrasound wave. If you have a spherical object, like the skull of a fetus, any area that is tangent to the ultrasound wave doesn’t appear, because there’s no echo to go back. They have to include that information in a graphical way to the network to complete the segmentation of the skull. They also include information about the shadow-casting effect . That means that you have a structure which is very thick closer to the probe that is throwing the rays, and then you have another area that you want to acquire, so because that first object is very thick, they project a big shadow on the area behind it, so that will affect the quality of the image behind it. They include that information in the network. Juan tells us they have proved the clinical effectiveness of their work, based on assessing malformation in the head. The state of the art in fetal imaging is working with 2D images, so they investigated the effectiveness of diagnosis using 3D ultrasound and proved that they could identify with almost 100% accuracy the anatomy and formation of the head, by comparison to a 50% success rate in traditional 2D biometrics. Juan says there are two things they are looking at next: “ Not everybody has access to a 3D probe, in developing countries or remote areas , for instance. What we’re trying to answer is, can we reconstruct the whole 3D scalp from only 2D images ? We are also looking to not only focus on the head, but on the rest of the fetal anatomy. This will help to improve the diagnosis of the development of the fetus .” Juan Cerrolaza ISBI DAILY Thursday 9 Want to find out more? Come along to Juan’s presentation today at 17:15-17:30 in the Empire Room. “ …identify with almost 100% accuracy the anatomy and formation of the head!