I'm not sure I understand the connection between level sets and tissue density here. Would level sets be used to encode the change in density, for instance (we might expect various tissue types to exhibit roughly constant density), or the difference between the measured density and a literature reference density (if this difference is near zero we are likely observing the desired tissue type)?
keenan
@Kuragama Good question: a CT scan would simply provide you with something like density, rather than distance to a surface. So to visualize the surface you could (for instance) draw everything contained in a sublevel set, i.e., all the values below a given density value. Or, above a given density. Or, between a minimum and maximum density. For instance, if you know (a priori) the density of bone, you might "turn on" all voxels in some range of density values near bone density to visualize the skeleton. This representation is still implicit (since we don't know immediately which points are contained in the shape), just in a different way than a distance function. The use cases are similar, e.g., testing whether a point is inside/outside a bone region would still be cheap to compute.
I'm not sure I understand the connection between level sets and tissue density here. Would level sets be used to encode the change in density, for instance (we might expect various tissue types to exhibit roughly constant density), or the difference between the measured density and a literature reference density (if this difference is near zero we are likely observing the desired tissue type)?
@Kuragama Good question: a CT scan would simply provide you with something like density, rather than distance to a surface. So to visualize the surface you could (for instance) draw everything contained in a sublevel set, i.e., all the values below a given density value. Or, above a given density. Or, between a minimum and maximum density. For instance, if you know (a priori) the density of bone, you might "turn on" all voxels in some range of density values near bone density to visualize the skeleton. This representation is still implicit (since we don't know immediately which points are contained in the shape), just in a different way than a distance function. The use cases are similar, e.g., testing whether a point is inside/outside a bone region would still be cheap to compute.