Neuromantic V1.5.2

Right, Neuromantic V1.5.2 is out, but only really fixes a couple of minor bugs related to 3D visualisation and stack rotation. The slice bar at the bottom right of the 3D window will allow you to define which section of the stack you see (in a simple way for the moment). If you can, I thoroughly recommend loading some MRA data or similar and trying that with some 3D glasses - it's pretty amazing.

The other minor thing it adds is a list of web links in the About menu, which mean that users can directly get to the Neuromantic user group and main page without having to remember the URLs.

The next version (out soon), will actually add some proper functionality - I'm currently finishing off the ability to drag the reconstruction segments around in the 3D window. This will have two major advantages:

1) You'll be able to more easily correct Z errors in this view from semi-automatic tracing.
2) When working with overlaid 3D data, you can drag the compartments in this view to achieve higher accuracy than may be possible with the 2D view.

At some point in the future it would be nice to be able to trace in this mode also (like Imaris FilamentTracer), but that's probably going to take quite a while. To some extent it probably depends on how bored I get over Christmas ;)

Labels: , ,


Neuromantic V1.5.1

V1.5.1 fixes the irritating 3D visualisation error, and should now run successfully on any passable computer. Again, it does not currently subsample the stacks images, so I wouldn't expect great results with stack resolutions of greater than 1024x1024 pixels, unless you have few slices.

This will be resolved in future releases.

The luminosity of the 3D display of the stack is generated with application of the current histogram stretching settings in the "Image processing" panel in the 2D window. So, if 3D volume isn't showing up well, boost the luminosity and regenerate the textures (by disabling then re-enabling the 3D overlay).

Also, the 3D volume currently doesn't look great from the top, due to the data not been pre-sliced that way. I may not bother fixing this directly and try to just replace the whole thing with a 3D OpenGL texture - I'm wasting quite a lot of memory with the current method.

However, to whet your appetite, here are some example renderings from the updated visualisation window. The data here is MRA (Magnetic Resonance Angiography) data from the Krasnow Institute, George Mason University, and the arterial reconstruction rendered was performed manually by Susan Wright using Neuron_Morpho. However, with any luck she'll be using Neuromantic from now on ;)

This is obviously a step away from the standard neuronal reconstructions that Neuromantic was originally designed to work with but, you have to admit, the data is exceptionally pretty! Unfortunately, when I visited Krasnow just before the Society for Neuroscience meeting I didn't have time to get my own head scanned in the end - I would have loved to have been able to put some renderings of my own brain on here.

Labels: , ,


Visualisation bug

Right, I've identified the bug with slow rendering - it appears that, through some monumentous coincidence, the laptop I developed the visualisation code on is possibly the only computer in the world that can correctly render non-power-of-two sized textures in OpenGL. Marvellous.

I'll release an update as soon as I've fixed it.

Neuromantic V1.5.0 Now Available!

Right, it has been a while, but there's an all new singing and dancing version of Neuromantic finally ready. For a full list of what's been changed, see the _README.txt file in the ZIP.

It should be noted that some of the new functions are incomplete and/or not thoroughly tested yet, so may or may not work for you. However, I wanted to get a version out asap after SfN, and the main functionality is all significantly improved. There will be a several more updates in the next few weeks, though, that fix current issues, so keep checking back.

Significant changes include:

* Huge amounts of general updating/bug fixes, including fixing the severe localisation error where SWC files could be generated with either commas or decimal points depending on Windows settings. Neuromantic now loads both regardless and always saves out using the standard decimal point notation (thanks to Padraig Gleeson of UCL for pointing that one out).

* Support for loading MorphML, as well as batch conversion from MorphML to SWC/

* A lot of functionality added to the toolbars, including visualisation options, connectivity editing and segment type options.

* Options to facilitate editing trees, including the ability to reverse sections of dendrites (on toolbar).

* Edit tracking options (added for Duncan Donohue of the Krasnow Institute for editing neurons for NeuroMorpho.org. Enable this in Options->Commenting->Record Changes to keep complete track of any changes made to a reconstruction.

* Additional visualisation modes in 2D with a filled option (like NeuroLucida) and a mode for showing the directionality of the neurites for error-spotting (the arrows point away from the soma).


Some of the functionality is not yet fully completed and thus should be considered as a Beta version.

Statistics calculations
This is the most complete and tested of the new functionality, and may be found on main menu under "statistics". Options exist to calculate a variety of morphometrics, as well as allowing statistical comparison of populations of reconstructions. The results on specific measures may be slightly different from that output by L-Measure due to differing ways in which outlying compartments are excluded from the main calculation.

Graphing options
Allows the plotting of properties such as branch order and euclidean distance, as well as a huge number of other properties. However, it can be a little unreliable at times and needs some more debugging.

3D visualisation
This is certainly the prettiest new addition but, alas, the most unreliable and incomplete for now. There is some (as yet unknown) issue with the OpenGL code, and on some computers it will work perfectly and on others it will freeze and kill the application, and this is independent of the power of the graphics card. Be careful to save any data before trying this, and I'm working hard to find out what on Earth is causing this issue.

Also, at the moment it doesn't downsample the slice images, so don't be surprised if it freezes when you try to display your 3000x3000x100 stack, OK? ;) It was primarily implemented to allow visualisation of MRA data with images resolutions <500 pixels squared or so. When it works, though, it looks beautiful - I'll post some screenshots soon.

Anyway, this mode allows overlaying the current slice data over the reconstruction in the 3D window, as well as masking by luminosity and various levels of translucency. Enable it in the 3D window menu with Overlay->Show Image Stack from the menu.