Sie sind auf Seite 1von 5

Image reconstruction is a mathematical process that generates images for example from X-

ray projection data acquired at many different angles around the patient. Image
reconstruction has fundamental impacts on image quality.

Two major categories of reconstruction methods exist

1. Analytical reconstruction

The most commonly used analytical reconstruction methods on commercial CT scanners are
the filtered backprojection (FBP), which uses a filter on the projection data before
backprojecting the data onto the image space.

FBP involves two phases a filtering and a Back projection phase which is the inverse
operation of the forward projection of an image but instead of each detector getting the
integral line of the object function now each point of the object domain receives the value of
the detector point where it projects.

Generally,

A straight line in cartesian coordinates can be described either by its slope intercept form,
y = ax + b Or x cos + y sin =

The projection of a parallel ray beam may be modeled by a set of such lines as,

If all the values of & are considered the above equation becomes

The approach is to back-project each projection and then sum all the backprojections to
generate one image
The Final image is formed by integrating all backprojected images

In general, the Image formed from a single backprojection obtained at an angle is given
by:

Final image is formed by integrating all backprojected images

Obtaining backprojections yields blurred results.


Another reason that explains why the back projections appear so blurry is because the
Fourier domain of an object is sampled in such a way that the low frequencies that take care
of the smooth surfaces are sampled much more densely than the high ones that take care of
the details and the sharp edges . As the High frequencies are under-presented the resulted
back projection is not sharp.

Straightforward solution to this problem is filtering the projections before computing the
back projections. Supplying a high pass filter on the detector function, the low frequencies
will be suppressed such that they will not result in an overly blurred back projected image.

As it is usually happens in image processing, the filter is applied in its Fourrier domain so the
function is fully transformed using the Fast Fourrier transform method and the resulting
spectrum is then filtered by multiplying it with a filter.
The result is that the low pass frequencies are somewhat suppressed and the high ones gain
importance. Also at some point there is a cut off and all frequencies higher than that are set
to zero. After filtering the spectrum is inverse fully transformed resulting in a new detector
function which is called Q and is back projected.

Using 2-D Inverse Fourier Transform of F(u, v) is:

Taking, u = cos & v = sin & dudv = d d, we can express above equation in polar
coordinates:

Then, using Fourier-Slice theorem

Splitting the integral in 2 expressions, for in the range 0 to 180 & 180 to 360 and using
the fact G(, + 180) = G(-,) we get

or

The term inside the bracket is inverse Fourier Transform of the product of two frequency
domain functions.

For normal back projection we get the function Fbp (x,y) over the lines of x-rays but for FBP
we do this back projection on the filtered detector function q which is the inverse Fourier
Transform function of the Fourier Transform Q which is the result of the original Fourier
Transform function P multiplied by the absolute value of W which is the filter.
https://www.youtube.com/watch?v=pZ7JlXagT0w&t=77s

https://www.youtube.com/watch?v=Sued23642Js&t=1158s

DIGITAL IMAGE PROCESSING IMAGE RECONSTRUCTION by Dr. K. M. Bhurchandi

2. iterative reconstruction (IR).

Das könnte Ihnen auch gefallen