----- Forwarded message from Murat Maga <[email protected]> ----- Date: Thu, 14 Mar 2013 00:04:24 -0400 From: Murat Maga <[email protected]> Reply-To: Murat Maga <[email protected]> Subject: RE: Software for landmarks To: [email protected]
For your issue #2, I would suggest 3D Slicer (http://www.slicer.org) with couple caveats 1. Spend considerable amount of time making sure the reference coordinate systems are correct (so that you don't end up with mirror images, upside down stacks etc) and the resolutions are read correctly. This should be done for any 3D software anyways. 2. Make sure you have plenty of RAM available for Slicer (my rule of thumb: 8 times the size of your dataset. More if you plan to do anything more than landmarking. 3. Making a surface model out of your Ct scan and using that to place the landmarks will avoid the nuisance of perspective problems inherent to volume rendering. This might be computationally unfeasible for a large dataset without resampling. 4. Have a high-end GPU. 5. Save often. M -----Original Message----- From: [email protected] [mailto:[email protected]] Sent: Wednesday, March 13, 2013 8:44 PM To: [email protected] Subject: Software for landmarks ----- Forwarded message from William Ary <[email protected]> ----- Date: Wed, 13 Mar 2013 17:25:31 -0400 From: William Ary <[email protected]> Reply-To: William Ary <[email protected]> Subject: Software for landmarks To: [email protected] Hello, everyone- My lab is using segmented high-resolution CT images to generate models for geometric morphometrics. I am having problems doing the following: 1. Finding homologous landmarks for smooth bones that have no intersections with other bones. This forces us to use less repeatable things like ridges and prominences that are harder to place landmarks on in a consistent manner. I plan to use a sliding semilandmark method to solve this problem, possibly a 3-D version as I have heard that the Geomorph package in R handles 3D semilandmarks. Other packages like SPHARM, MorphoJ, Morphologika etc. are in the running if they can solve my landmark issue. 2. Picking a software for placing landmarks that works with our large high-resolution models with millions of elements. I have sampled the vast array of programs to some degree, having tried Paraview, Meshlab, MiniMagics, and our segmentation software, Analyze. I have encountered problems with all of them as follows: Paraview- Doesn't record points, cannot find way to place multiple points, doesn't record Cartesian 3-space coordinates. Meshlab- Looked great but bogs down or crashes when placing points, likely due to model complexity MiniMagics- Could not figure out how to place multiple landmarks/model size issues. Investigating if MIMICS is better. Analyze- Points disappear on the topology of the model and are hard to see once the model is moved. After all this, I welcome any suggestions you may have on which programs allow picking landmarks/semilandmarks on high resolution models and how to generate landmarks on our smooth bones. We are hoping there is an alternative to coarsening our models. I would be especially grateful to know how to go about placing a series of sliding semilandmarks. I have read some papers on this, but suggested reading is always appreciated! Many thanks! Will Ary Master's candidate San Diego State University [email protected] ----- End forwarded message ----- ----- End forwarded message -----
