Bragging for a day...

[martin]

Ok, I figure I can brag for a day.....

Here's a screen cap, of an application I am working on, it's not for a job per say, it's something I'll use on my resume....

It takes the images out of a DICOM directory, which are medical scans, and builds up a 3D image from the 2D data....it's just for "funseez".  The actual images are 2D slices of the head, I stack them on top of each other to create the illusion of 3D, that's the short explanation.

[Mike (DDTFA)]

What are you using as a viewer?  

Does your app have other capabilities, such as generating wire mesh models or reslicing at different angles?

[Timothy]

evil Ronnie's ?

[martin]

I was going use SharpGL to write the viewer, but that would have been a lot of time consuming work. I decided to go with Unity3D as the viewer. You can rotate the skull 360 degrees.

I added slider to enhance blue channel. Another to increase/decrease the intensity as DICOM images can vary in intensity...usually too low, and hard to see.

[martin]

Here is a mostly grayscale version at a different angle

[martin]

I've seen wire mesh generations of DICOM data, and they all looked horrible, and did not look worth doing/implementing...which is a pain.

You can "slice" the view into segments....I'll post what that looks like....

[martin]

[martin]

Mike you seem to know about this?!

Wire mesh generation?! heh-heh.....only somebody familiar with this would ask that question.

[Mike (DDTFA)]

I imagine the challenge would be finding the necessary contrast needed to find the subsurfaces in order to generate the wire mesh.

[Mike (DDTFA)]

I've looked at Blender, haven't used it but I have an idea of what it could do.  If you haven't heard of it (which I doubt), it's an open source program you can get for free.

[Mike (DDTFA)]

Also, if a body part could be imaged in 3D and then 3D-printed, a surgeon could practice a procedure on the model before doing it on a patient.   But I'd guess that someone has already thought of that. 

[martin]

Blender is an excellent open source program....

I have yet to see a wire mesh generated model of DICOM images that looked worth the time of day. They do the job, but it does not look useful, initially I thought it would be worth doing, but the end results look terrible. Perhaps adjusting contrast might help, I dunno, one application that I saw with this feature, and the app is top notch, the resulting mesh generated looked terrible.

[martin]

A better slice...slicing back into the jawline area....

[Mike (DDTFA)]

I'm guessing that the circular artifacts are a result of some variation of the exposure of the original slices.  That could be a variation in the scanner or perhaps due to intervening material between the exposure source and the point being imaged.  This may have something to do with the difficulty in generating wire mesh models.

[martin]

Circular artifacts are a common issue I see....

When I do thin slices, it will be more pronounced, when I don't do thin slices it is less noticeable,  when I do a full view with no slicing, it is usually not noticeable.

[martin]

You sound like you've had experience in medical imaging??????

[Mike (DDTFA)]

My only experience with medical imaging is from looking at scans of my own body parts for procedures done on me.

I have an interest in graphic arts expanding from a photography hobby with more of a focus on the technical side.  I know enough about Adobe Lightroom to be dangerous.

[martin]

Finding decent DICOM samples is kind of hard....there are a lot out there, but for demo purposes they were not very good samples available.

My skull is a 256x256x482 3D volume.

I sourced them from a 512x512x482 data set, but the resulting volume took up way too much memory....around 167MB or around there....my old PC couldn't handle it.

[martin]

Mike "I'm guessing that the circular artifacts are a result of some variation of the exposure of the original slices."

I would have to look at the originals and see how they look. I suspect it is the process of generating the 3D volume, that introduces this side effect. In some apps, this issue is pronounced. In others it is less pronounced.

[Dean]

Great work, Martin. Awesome, as our favorite false prophet would say.

[martin]

thx...

I'm like Scotty in Star Trek....I do this kind of stuff even on vacation.

[martin]

Here's another screen cap...

The first one is of a torso, but you can't see inside, the DCIM has too much high contrast so that the pixel "cloud" is too thick to see through.....

It took a while to fix, because of the nature of 3D volumes, I finally got it to work so that you could see "inside", that's what the second picture is...