Grapefruit USA Inc (GPFT)

[KingKong_Guru]

Lobcrab....

I agree with you. I have also spoken to people who worked/working on 3d reconstruction algorithm . 3d reconstruction is mathematically complex problem often computationally expensive.

Most of these algorithm uses multiple 2d images taken around the object to reconstruct 3d images. Each 2d picture is represented by NxN matrix of pixels . Each pixel can have different color , brightness etc. In very simple language , placing each pixel of each images in space (assigning the Z cordinate ) gives 3d images.

I have spoken to engineer who have written software code to implement Feldkamp algorithm , which is widely used algorithm to reconstruct 3d image in cone beam X ray systems including O ARM. Feldcamp algorithm also works on traditional approach , where a given point is matched from multiple angles to determine the location of pixel in space (Z cordinate).
People can get an idea.
http://www.umiacs.umd.edu/~ramani/cmsc828d/lecture28_6pp.pdf

Even though Feldcamp algorithm is approximate ....it still takes significant amount of time to reconstruct 3d images ..in order of multiple seconds.
Feldcamp algorithm require large number if projection (more than 200) , typically taken from every 1-2 degree in circular gantry.
Cone beam angle should also be very small to get better images quality.

So why DVIS can reconstruct 3d images fast.
As we discussed ..creating 3d images is basically assigning Z cordinate for each pixel of every picture taken around the object.
The axis connecting X ray source and detector is actually Z axis.
The perpendicular axis are X and Y with respect of one reference location.

Basically distance from X ray source to given pixel is actually location of pixel on Z axis.

Lets say in first DVIS rotation s/W determines the distance of each pixel of every picture from X ray source . Storing these distance for all pixel actually creates 3d base map.
Now as per DVIS patent ....2d picture are pasted on 3d base map....which is nothing but assigning Z cordinate from stored 3d map look up table.

This technique allows DVIS to construct 3d images very quickly ....although there are some open ended question like ....How does DVIS updates the 3d base map when there is movement in target position.

Measuring distance of pixel of 2d picture from the X ray source is soul of DVIS S/W algorithm .....how it is done ....I am not sure ....but I feel it something to with brightness of pixel appearance on the flat panel detector .....closure the pixel to X ray source ..brighter the appearance .....

Overall note:

DVIS is viable technology .....hardware is truly very close to CBCT systems .... Only difference is S/W algorithm to construct 3d image ....which is simple , elegant and brilliant ...


KKG