Hi Giuseppe, On Apr 25, 2013, at 5:52 PM, Giuseppe Isu wrote:
> Hi everebody, > > I'm facing a geometry reconstruction task on vmtk. I need to reconstruct a > carotid bifurcation, but I'm experiencing some difficulties at the bulb > region. Generally I have an amazing image quality, I can easily segment level > sets for all branches (high meta levels within the lumen over 1000 and low > levels otherwise), but due to blood recirculation I have low levels and kind > of disturbed uneven image at the bulb. I tried to initialize the segmentation > in different ways but the result is always a staired surface which follows > the pixel level fluctuations in that region. I have seen similar issues in the presence of complex flow patterns in the past. > I thought to use vmtkimagevoiselector to divide my voi in three domains in > order to isolate the bulb volume, and then smooth the image only in the > bifurcation region. > > The idea is to perform level set segmentation separately for the three > domains and then use vmtkimagecompose to reassemble my segmented volume. > > A problem occurs when I pipe vmtkimagevoiselector. To obtain a precise > subdivision of the original volume I want to use -boxbounds insead of > interactive voi selection. I pipe this code: > > vmtkimagevoiselector -ifile myname.vti -interactive 0 -boundbox xmin xmax > ymin ymax zmin zmax -ofile myoutput.vti --pipe vmtkimageviewer > > but the output image results empty. what's wrong? it seems to me to have > used the right pipe synthax. > > Then if I omit the part "-interactive 0" I get the interactive option to be > enabled typing "i" and, again, empty image. Setting aside the issue with vmtkimagevoiselector for a second, I suggest you don't split the image in three domains, but just keep your original image around, generate a (full-sized) smooth version and create three (or two, as a matter of facts) different segmentations for the different regions, using the original or smoothed images as you see fit. Then you can compose the resulting level set images using vmtkimagecompose, which will effectively merge the shapes you segmented in a single shape. Now, on to the vmtkimagevoiselector. BoxBounds are specified in physical units (mm?), not in indices - this might be the problem. To have vmtkimageviewer show physical units, specify the -continuouscursor 1 option at the command line for vmtkimageviewer itself. By clicking on the image planes you'll see the position in physical units appearing on screen. Hope this helps. Luca > > Can you give me any tip? > > Thank you all > > -- > Giuseppe > ------------------------------------------------------------------------------ > Try New Relic Now & We'll Send You this Cool Shirt > New Relic is the only SaaS-based application performance monitoring service > that delivers powerful full stack analytics. Optimize and monitor your > browser, app, & servers with just a few lines of code. Try New Relic > and get this awesome Nerd Life shirt! > http://p.sf.net/sfu/newrelic_d2d_apr_______________________________________________ > vmtk-users mailing list > vmtk-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/vmtk-users ------------------------------------------------------------------------------ Introducing AppDynamics Lite, a free troubleshooting tool for Java/.NET Get 100% visibility into your production application - at no cost. Code-level diagnostics for performance bottlenecks with <2% overhead Download for free and get started troubleshooting in minutes. http://p.sf.net/sfu/appdyn_d2d_ap1 _______________________________________________ vmtk-users mailing list vmtk-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/vmtk-users
