|
Monday, 04 April 2005 06:27 |
a brain dump of sorts...
Open Source + Open Data = Open Science
The Visible Dave Project is a research experiment in scientifically modeling my own physical body. My focus is on the general topics of MRI, DICOM, 3d modeling, 3d projections, data visualization, data fusion, photogrammetry, and fun tricks for personal pleasure. If anyone has access to time on an MRI/fMRI machine in testing, please let me know...I would love to co-register with more datasets. Ultrasound time would also be cool. CT is great, but I don't enjoy lots of radiation. ;)
Where is the data from?
The data is from an MRI I had done in 2002. I didn't have it for diagnostics, it was just a sample MRI so that I could do visualizations with it.
My head dataset is 256x256 which is pretty high resolution for MRI (low resolution when compared to things like CT). I have two datasets consisting of the sagital and transverse view. Slice thickness varies between the sagital(1.5mm) and transverse(1.7mm) views.
The head MRI datasets show my nose cut off because of phase-wrap. This is because the MRI tech didn't set the field of view correctly on the machine for my head. In the sagital view, the nose data actually appears behind the back of my head. The transverse view does not contain my nose data at all! Someday, I will create a program to correct for the phase-wrap so that I can have a nice complete surface model. (bummer, this was a real bummer man)
My aim in this project is to understand better the DICOM format, MRI physics, and voxel (3d pixel) processing. I've also come up with a bunch of silly uses for it like... it would be fun to create a 3d projection of my internal structures and allow for realtime surface exploration. Fun stuff!!! Plus, it is neat to see pretty pictures of the inside of my head. :-P
If you are interested in my DICOM data for your own research, please let me know. I have them available to download... I just enjoy hearing from people interested in this sort of thing.
I don't have problems with you using the data on presentations and papers. I enjoy collaboration. :)
I'll continue updating this page as my project progresses.
Here is some views of the MRI head1 dataset:
| Random 2d views | Transverse | Sagittal |  |  |  | These movies show the REAL contents of my own head, yup it even contains a brain!
| 3D perspectives
        
For larger images checkout the gallery page |
Visible Human
I plan to build means of correlating my models against other models. I think the NIH - Visible Human project provides a wonderful dataset to correlate against.
The National Library of Medicine's Visible Human Project
Using VHD to build a comprehensive human model
Marching Through the Visible Man
Project Update! July 14th, 2006
I'm starting to think about ways to print my head out in physical form. At this point I have no idea on the best or cheapest way to do it. Anyway, if it is relatively cheap...I may just do it. How cool is that? A 3d model of your head on the desk. Talk about personalized art! 3d printers like ZCorp may be an option.
Update: 07-29-2006
A cool side note about some cool crystal art
[Read more about this art] | I just recently bought some art from Bathsheba Grossman, she really inspired me when I saw her rendering of DNA with molecular surface! To the left you'll see a picture of the DNA model from Grossman's site but the picture really doesn't do it justice.
Grossman does all sorts of great things with 3d art (check out her metal work). She also does work with crystal proteins @ http://crystalprotein.com/
There is also another company doing work with crystal proteins luminorum but they don't look quite as pretty. | |
On another somewhat seemly un-related side note, all these thoughts on three dimensional data, visualization, and math has got me thinking about a kids toy. Zometool is a very neat toy, it allows you to teach a child geometry and math using sticks and balls. It feels a little like a mix between Legos and Tinkertoys, but zome is much more mathematical. With it you can create structures like hyper-cubes. Neato. Definitely going on my fun toy wishlist.
I'm thinking about buying the book Zome Geometry, which gives a nice tour through a lot of great math structures and theorems. My nephew just got a kit and I was pretty impressed. You can even buy professional kits for huge structures.
Zome web resources
The Zome system is made by Zometool, Inc.
Model of the month
ZOME PUZZLES
ZomeCad
Anyways, back to the thought at hand.... thoughts on MRI modeling(DICOM)....
More info on DICOM
DICOM is a image format for medical imaging. Hospitals have ultrasounds, CTs, MRIs, and other equipment that store and transmit data in DICOM. So access and processing of these formats is very handy to have.
DICOM is not just an image format. DICOM also does transport and routing (and I'm sure other things I don't yet know of...seems like a pretty big spec and I haven't had time for it all).
One nice thing about DICOM is the ability to route information to machines on your network. Each machine and interface have their own titles. I don't yet know how I feel about this transport stuff...
DICOM Resources:
The DICOM Standard
DCMTK - DICOM Toolkit
MIR DICOM Central Test Node Software
BET - Brain Extraction Tool
FSL & FreeSurfer Courses
FreeSurfer
FMRIB Software Library
Volume Rendering the Cortex
Bio-medical Imaging in Java
Project Update Jan-21-2007
I would like to get another head scan so that I can do comparative modeling of my internal structures. I would like to use these comparative models to build custom models which I can further study.
So I can with my MRI, label and 3d model my brain surfaces. I could then overlay additional information for the surfaces. It would be very interesting to model EEG data on my brain region models. This would allow me to model structures over time in 3d and apply higher dimensional information to the question at hand. For instance, model the REM cycles and provide a count of periods of activity within the stages. Interesting.
Knowing the surface of 3d things and how they interact in space can provide you with a wealth of information. In this model, I'm studying the real matter that gives rise to my thoughts and actions. The shape and interactions of my physical brain define the world as I perceive it.
It amazes me, when I think about the effects of morphology on our physical world. In biology, surface interactions indicate enough information to build medicines that change 3d protein confirmation and thereby control gene expression. Controlling gene expression can offer the ability to modify feedback loops that happen over time.
Project Update Feb-11-2007
FSL Evaluation and Example Data Suite
FreeBSD and MRI
FSL & FreeSurfer Courses - Lecture Slides and Practical Instructions
Update: Feb-15-2007
It would be very nice to integrate my brain models with something like what is output from: Photorealistic models with MakeHuman
Using photo-realistic models which are also physically-realistic could be very interesting. Then add to that the speech synths which will reproduce speech based on personalized vocal cord models...and 3d animations of the human body based on the real physical joints. Integration of multiple models provides for very scientifically accurate immersive environments for research and collaboration.
Update: March-22-2007
I'm finding more of the programs and libraries I need to start getting results from my data. Thanks so much for everyone who has provided me with comments and questions.
I also found two new web resources
AMIDE: Amide's a Medical Imaging Data Examiner - competely free tool for viewing, analyzing, and registering volumetric medical imaging data sets.
CT Sim - The Open Source Computed Tomography Simulator
Update: May, 2007
OsiriX - DICOM Viewer
NMR research (MRIUtil,vtkCNMRRLibrary,itkCNMRRLibrary)
vtkCNMRRLibrary 1.0.3 Documentation - VTK based MRI tools
dicom2 - convert medical images and DICOM files to various other formats, while optionally performing some rudimentary image processing tasks
Medical Imaging Interaction Toolkit(MITK) - combines VTK and ITK
Slicer
NA-MIC Wiki
BIRN - Biomedical Informatics Research Network
The VolPack Volume Rendering Library
ITK-SNAP Home Page
The Papyrus toolkit 3
NeuroSlice brain model - MRI teaching model
Interactive course about MRI physics
dcm4che - Open Source Clinical Image and Object Management
MacOS Notes
On a mac, first of all, I would suggest installing OsiriX. It is quite good.
Apple - Seminars Online - Getting Started with OsiriX
Installing FSL on Apple Mac
Dcmtk 3.5.4 on Mac OS X (fink packages)
Making Your Own Visible Woman
VoluMedic a add on for LightWave3D with DICOM support.
Philips Free DICOM viewer
DesAcc : DICOM Medical Imaging Software
itk-SnAP Home Page
itk SNAP Tutorial and User's Manual
When building itksnap.trunk on the Mac I had trouble finding the ITK_FLTK_RESOURCE I would always get a ITK_FLTK_RESOURCE-NOTFOUND error from cmake. This is the resource file that FLTK provides for mac. It is included in your FLTK src tree under FL. (fltk.trunk/FL). The actual resource file is named mac.r inside this directory.
Data Format Working Group â Neuroimaging Informatics Technology Initiative
NIfTI-1 Data Format â Neuroimaging Informatics Technology Initiative
NIfTI-1 Support â Neuroimaging Informatics Technology Initiative
ImageJ Plugin - NIfTi Input/Output
ImageJ Plugin - Import Dicom Sequence
dcm2nii DICOM to NIfTI image conversion
PyNIfTI - Python bindings to NIfTI
MRIcron Index Page
OSIRIS Presentation EN
medisoft.gr - Sante DICOM Software
ReviseMRI.com: Study MRI physics
Update: I've gotten the latest ITK-SNAP and it has produced me a great 3d mesh from the results of the automatic segmentation engine within ITK-SNAP. I've uploaded a few screen shots in my gallery. I've found getting the parameters right for segmentation is hard. There is a real art to the proper selection.
What I'd like to be able to do is load a reference set of data with existing segmentation results and try to match my real data against it. I would like to get a huge set of labels for the brain and do this. It should be possible, right?
PlanUNC - Welcome
MRIcro Anatomical Templates
Working with meshes
A common format used today in fabrication shops is STL.
MeshLab - is an open source, portable, and extensible system for the processing and editing of unstructured 3D triangular meshes.
In the future, I'd like to load these meshes into a 3d silverlight application so people can select and explore my structures right on the website.
I like this picture because it is symmetric...and a bit freaky. ;) It is of me playing with the Mac PhotoBooth mirror effect. I plan to build a model of the above view.
I could take my mesh structures and build a model which would represent this scene using the eyes, nose, and mouth to orient the models together. I should be able to do photo realistic lighting and shading...
(A plastic replica of a life sized human brain that shows multiple views. )
It would be great to produce custom educational diagrams/models based on my own data...
I thought Brain 3D Chart - Raised Relief Chart was neat too.
NIH's Clinical Image Processing Department - Software - etdips
Data visualization in crystal - some very cool crystals of the human heart and CT scans shown.
Tactile Visualisation: Feel your data! - has an awesome pdf with information on taking data and creating physical 3d models for exploring data tactilely.
NITRC: The Neuroimaging Informatics Tools and Resources Clearinghouse - a large database of fMRI related software
SourceForge.net: Grassroots DICOM - Grassroots DiCoM is a C++ library for DICOM medical files. It is automatically wrapped to python/C#/Java (using swig). It supports RAW,JPEG (lossy/lossless),J2K,JPEG-LS,RLE and deflated. It also comes with DICOM Part 3,6 & 7 of the standard as XML files.
idoimaging.com - dicom and related software index.
Dicom3tools Software - command line utilities for creating, modifying, dumping and validating files of DICOM attributes, and conversion of proprietary image formats to DICOM.
DVTk (DICOM) - Wiwi - Downloads
Update: Nov 13, 2008
OsiriX 1.0 for the iPhone!! Haha, I love to see stuff like this running mobile.
|
|
Last Updated ( Thursday, 13 November 2008 08:19 )
|
|
|
Random Quote
| With sufficient thrust, pigs fly just fine. However, this is not
necessarily a good idea. -- RFC 1925 |
|
|
|
