Hi, 
I am trying to use APP2 to do neuron reconstruction, and encounter 2 questions:

1. How to tune APP2 parameters automatically?

---- Most of time, you don't need to do change anything. But of course, based on your understanding of the image content, you may choose some image threshold to remove background, or set to be automatically generated (set the threshold to be -1). Many times GSDT will make the results better, but it will take a little bit longer to run than the no-gsdt case. Normally we don't change the connection type or short-branch pruning parameters. *** It is important to note that your data may contain a lot of noise of other problems, which should be handled by your special pre-processing programs, but not APP2, which only generates a neuron tree based on an input image. For instance, if your data contains multiple touching neurons, you need to specify which one to trace. You could do that by defining several markers to roughly define the spatial range of tracing.

Since I want to apply this method to around 100 neurons, I tried default parameters first.
However, in most of the cases, the number of branches are less than expected when compared with image data.
In a few cases, the reconstruction results are quite different from their original images.
I think this problem should be able to solved by adjusting the parameters, but so far I can't find a suitable set of parameters.
Any suggestion about how to solve this problem?

---- what are your data?? If you trace mouse neurons and fly neurons at the same time, how can you imagine the parameters would be the same??

Here are the examples I tested (.lsm) and their corresponding results (.swc): download here
(because the image data is really huge, forgive me for only uploading some of them)
Except for bkg_thresh was changed to -1 (auto-thresholding), the rest of the parameters were set by default.
These neurons could be found here: BRC
(For example, the neuron name of the file "Cha-F-100065.lsm_x95_y103_z56_app2_-1.swc" is Cha-F-100065.)

2. If I already have neuron skeleton data, is that possible to use APP2 to re-estimate neurite radius? 
(I guess it can be done by inserting neuron skeleton from the parameter "inmarker_file" but failed.)
 ---- just load the image data first into vaa3d, then open its 3D viewer, then drag and drop the neuron skeleton into this 3d viewer, then right click the neuron  and enter the editing mode, then right -click again to choose "change nearest neuron segment radius" , and then choose the parameter -1, which means to re-estimate automatically based on the image content.

Thanks!