Morpheus can address some of your issues. Sorry for the length. See below...
morphmet wrote: > Dear Morphemetricians- > > > > I am working on developing a project involving 3D landmark data of > primate crania with which I plan to run a series of geometric > morphometric analyses. Additionally, in order to make my analysis > comparable to previous analyses, I would like to extract some > traditional morphometric data from my landmark dataset. I am also > planning to designate semilandmark curves using .ply files constructed > from 3D point clouds. But my problem is that many of my measurements and > the placement of my curves require that I establish a sagittal plane for > my entire point configuration. So I have two questions: Morpheus does not directly support .ply files, but assuming you have the means to extract the point coordinates and curves... > > > > 1) How would I go about rotating my entire landmark configuration so > that the sagittal plane (as designated by a number of midline landmarks) > lies along a single axis within the coordinate system? I understand from Technically, the "midline" landmarks lie around a mid-sagittal plane. Therefore, they require two dimensions. Projection onto either of these dimensions (or any others within that plane) lie close to a single dimension, but this is probably not interesting. > previous postings how to manually rotate my configuration by 90 degrees, > but is there any way to designate that particular points are in the > sagittal plane and that those points are along the x, y, or z axis? Are > there any programs that this can be done with and then the rotated > landmarks coordinates exported? I have been successful using Geomagic to > rotate configurations, but then the exported coordinates are completely > unordered, which makes them essentially useless. This is what I have done in Morpheus. First, Morpheus distinguishes between two classes of landmarks - primary and secondary. Only the primary landmarks are used for Procrustes computations. The secondary landmarks are just carried along by the application of parameters (translation, rotation, and scale) estimated from the primaries. So, if you specify only the midsagittal landmarks as primary points, you can align the whole skull data set (secondary landmarks, curves, etc.) to these midplane points. Now, if these points cover the lateral outline of the skull (e.g., some are on neurocranium, some on the basicranium, some on the face), the first two principal components of the scatter of these points in physical space are approximately: PC1 = antero-posterior axis, PC2 - supero-inferior axis, PC3 - deviations from the midsagittal plane. Morpheus has an option to align the results of the fitting to the PCs of the consensus configuration (again using only the primary points in the consensus). So... 1) Put all landmarks for all specimens in morpheus format file (importing from an NTSYS format file is easiest). 2) Change all points EXCEPT the mid-sagittal ones into secondary points using the "demote" feature of morpheus. Something like: "Command: demote p 3" Means demote point (you can also demote curves) number 3. In the current version of Morpheus, this is rather clumsy since the old primary point 3 will become the next numbered secondary point and all primary points numbered higher than three will have their number reduces by 1. This is virtually impossible to keep track of in your head, so I use the "Options" button to turn on point "number"ing in the plot, and read the numbers I need off the plot. (The new version under development treats "primary" vs. "secondary" as point attributes, not separate lists of points. This is much easier to follow from the user's (and programmer's) perspective.) After doing the above, you have primary points on the mid-sagittal plane and all other points are secondary. Next, turn on PCA alignment for GPA: "Command: list superimposition options" This will list the settable options for the superimposition procedures. Find the one about "finalOrientation." The default is "none," meaning no reorientation of the data after superimposition. Setting this to "PCA" will tell the program to align the results to the PCs of the primary point scatter. "Command: set superimposition finalorientation pca" Now do the GPA. Viola, the data will appear with the skull facing along the x-axis, top-to-bottom will be the y-axis, and the z-axis (assuming a favorable distribution of landmarks) will be distance from the mid-sagittal plane. You then use the promote command to bring the secondary landmarks back into the analytical fold: e.g., "Command: promote p 2" Again, look at the plot to see the current numbering of the landmarks in the two sets. Note that the above illustrates the advantage of knowing what various programs are doing with the data and something about the structure of the data, itself. With this, you can do useful things that wouldn't necessarily have an entry in a morphometrics table of contents. That is, nothing described above "knows" anything about skulls or particular landmarks like bregma or basion or which points are on the "face." It involves only the application of generic morphometric procedures to sets of coordinates about which *you* know something. The procedure might not work, but it has for all the human craniometric data I have processed. > > > > 2) Does anyone know of any programs that can be used with 3D data > to extract traditional morphometric data other than simple distances > between landmarks? Specifically, I?m looking to measure the angulation > of structures against particular planes and also to measure distances as > projected onto a particular plane. I?m sure that this could > alternatively be done in Excel using multiple complicated algorithms and > formulae, but I?m afraid I?m not that mathematically savvy. Morpheus can compute a number of things from your data by putting measurement commands in the data file or importing them from another morpheus data file containing only measurement commands (allows reuse of particular measurements across multiple data sets). Mainly, the program can compute distances and angles: DIST i j The distance between points i and j. ANGL i j k Then angle formed by points i,j,k at point j (I think). Other useful measurements could be added in future versions. "Command: list measurements" will show the numbers, which can be saved by exporting all of the data to matrix format - the measurements will have their own columns. If the above alignment procedure works for your data, the z-coordinate will be the distance of a point to the midsagittal plane. Hope this helps, dslice > > > > Thanks so much for your assistance. > > > > Claire E. Terhune > > Ph.D. Student > Institute of Human Origins > > School of Human Evolution and Social Change > Box 872402 > Arizona State University > Tempe, AZ 85287-2402 > [EMAIL PROTECTED] > > -- Dennis E. Slice Department of Anthropology University of Vienna -- Replies will be sent to the list. For more information visit http://www.morphometrics.org
