Assuming you import each slice as an image, you can bring them each in as a field (ReadImage), group them (CollectSeries) and Stack them into a volume. You could isolate this operation in a macro to simplify the networks you write. In that case you could create a ForEach loop to do each Import and then the CollectSeries->Stack at the end. That macro could also handle the appropriate number of slices/images as well as the pixel size of each at Import. What would be returned in each case would be a single field as a 3d volume.
BTW, I think someone has written a DICOM filter (e.g., perhaps Ed Farrell, who retired from IBM Research a few years ago). I recall references to it in the past. 1. Rectilinear regions could be specified by Slab. the results of which could be passed to Statistics to get mean values. Alternatively, coordinates could be specified interactively on a rendered image by ProbeList or Pick and the points could be used to define a polygon to then create a more abitrarily-shaped ROI. 2. If you keep the slices as a series rather than stack them as a volume, when you apply whatever operation it will be done on the individual images separately, putting the results into a similar series. You could then animate, for example, through the slices to see the changes per time. 3. The Filter module supports a number of different kernels, and you can add your own. Reduce uses weighted average interpolation to smooth data to a lower-resolution grid, maintaining the same bounding box. Alexander Kluge <[EMAIL PROTECTED]> @opendx.watson.ibm.com on 07/14/2001 09:22:25 AM Please respond to [email protected] Sent by: [EMAIL PROTECTED] To: [email protected] cc: Subject: [opendx-users] First advice please First, I apologize for a question one could probably answer by r<ing>tfm thoroughly. But this takes really a lot of time and I would like to have a little success first as a motivation for the start. Could you be so kind to draft in very short terms possible ways to achieve the results desired. Maybe some of the sample files has a similar structure? Given are 1 to 3 stacks of 40 to 60 time slices/repetitions of an image of 196 to 256 x 256 pixel with 16bith depth. For a given Data set, all images have the same pixel size. Time axis is known, intervals are constant. Data are stored in one dicom series, but can be converted to tiff. 1. I want to specify one/several ROI in one image of the series and get the mean value of the pixel values of all images as y, the time beeing the x axis. 2.1 I want to perform a similar computation on every single pixel for the whole series, e.g. computing the slope of intensity changes over time, transit times and so on and get the result e.g. as a color coded parameter map as an overlay over the original grid. 2.2 I want to filter all images with the same algorithm (smoothing, grouping e.g. 3x3 pixel into one) to be able to perform the calculation of 2.1. in a realistic amount of time (maybe due to the web- and x-server running on my W98 machine, opendx is really slow). Thank you for any help Alexander
