Medical Visualization/Simulation with Blender

Greetings,

(I’m not so sure if i can ask for advice on this forum, forgive me if i’m wrong)

I’ve had severe headaches, and i 'm posting this after a month of considering(this project). Actually, as a 36 month blender user, I’m nothing but ‘friendly’ with blender - before now, i have used blender only as a tool for making cool images and doing physical simulations. But (as a to-be medical scientist), i feel that the power and functionality of blender can be a great foundation to my project involving medical imaging. So, the following is a rough sketch of what i have in mind:

< Medblend dev. plans >

  1. Visualization
    (1). representation of CT/MRI/PET/SPECT/… 3d voxel data/image sequence in 3d navigatable space (in blender)
    (2). create cross sectional images out of voxel data/image sequence
    - voxel data : seems straightforward enough
    - image sequence : interpolation and fitting to the prototype meshes
    (3). volume texture rendering

  2. Create the anatomical structures
    (1). mesh modeling of the basic human anatomy (prototype creation)
    (2). make an algorithm to automatically create patient organ mesh with reference to CT/MRI scan data - reform the prototype

  3. Simulating the body
    (1). body fluids (blood, lymph, or digestive fluid) - determine blood pressure, intracranial pressure(ICP), etc.
    - something more than softbody physics is needed to simulate tissues, organs, and especitally the brain
    - could be extremely useful in predicting brain vessel’s ischemia or rupture by clogging/tumor
    (2). skeletal/muscular forces - determine the forces exerted on specific body parts to predict rupture, etc.

  4. Graphical/Intelligent user interfaces - make it easy to learn, navigate, and use

  5. Optimization / Miscellaneous
    - coding with c++/python/blender script
    - a new daughter-program
    (1). leaving only the essential features of blender for time efficiency
    (2). adding all of the above to the reduced blender

It seems to be a huge project, and i’m not sure i can do this only with my two or three colleagues. Will i need more team? Also, i do wish to know if blender is indeed the right basis program for this kind of project. Maybe coding with openGL directly will be much better? (Still, blender’s great features are too good to overlook.) Also, (to be honest) the physical simulations take too long to calculate and is also not precise. Can there be a solution to this? It would be of great help if anyone could advise on this project.

Thanks in advance

To be honest I don’t know much about medical simulations, so take my opinion with a large grain of salt. It seems like your idea would be a large project, but one that would be extremely beneficial to the medical community.

Accurate visualization and reconstruction should not be a problem, provided you are extremely careful with how you model.

Here’s a video of a CT scan rendered in Blender, so that part must be possible:

Simulating the body’s functions, however, might be a much harder thing to tackle. While Blender’s simulation algorithms could be a potentially good basis on which to build a such a system, I don’t think they are nearly accurate (perhaps flexible would be a better word) enough to be used from inside Blender for hard core medical data . I don’t know exactly how it works though, so maybe someone who knows the code more can better answer this part.

Greetings Firnengreen,

I think your idea is excellent. I am myself a scientist (structural bioinformatics), spent years on creating tools of this kind (although more generic than just proteins), now finally integrating them into Blender. Here is some advice:

  • Be very very serious about this. If you are not willing to spend at least 1000 hours, just give it up.
  • Don’t count on too much help. Forums are great for specific questions, but don’t try to recruit a team or something, it will fail epically. Better to post your progress from time to time, perhaps someone will volunteer but 90% chance that such a person will be useless to you (lacking skill, dedication, or disagreement about the direction of the project).
  • On the other hand, don’t try to reinvent the wheel! Don’t design your own file format for MRI/SPECT, write instead an importer for existing file formats. There are plenty of scientists interested in MRI visualization/physical modeling, there should be several existing projects. Try to attach yourself to one of them as “the guy who uses our stuff with Blender” but at the same time, keep what you write flexible and adaptable to other formats and libraries (so you won’t suffer too much if the other project dies).

EDIT: “not reinventing the wheel” also goes for drawing and GUI. It is a huge and wasteful sink of time to write your own OpenGL code for voxels or GUI, Blender is more than capable to do what you need. On the other hand, if you ever get to the simulation stage, I would recommend a dedicated library (published in a medical/scientific journal) over the use of Blender’s simulation code.

that would be a mghty project
are you very good with blender 3D soft.python and C++?
For meshs simplifcation you might have to use Meshlab also and other external soft!

there is a group in South America that has done some work with blender to re construct Faces from Skulls data!
http://arc-team-open-research.blogspot.com.br/2012/05/forensic-facial-reconstruction-with.html

i think your talking like 10 000 hours here!
CAT MRI scripts are already available and work with voxels in blender
for other simulations not certain if it alread exist in blender !

if you still thinking seriously about it
i can give a few links for voxels and CAT/MRI scans
PM me and i’ll get a list ready!

happy bl

May be of some use, medical visualization at the protein level:
http://bioblender.eu/

No, that program is totally useless. I work with protein structures all day, and you can’t do anything with Bioblender beyond the “pretty picture” level.