TECHNIQUES FOR REDUCING ALIASING IN CT SYSTEMS
Publication Date: 2017-Sep-13
The IP.com Prior Art Database
A method to reduce aliasing comprises applying an anti-aliasing filter and down-sampling process to mitigate image quality (IQ) concerns associated with combining views. The method reduces aliasing in the sampled data, which flows through to an aliasing reduction in the final image. The gain in aliasing performance is further available for trade-off for boosting image resolution.
The present disclosure relates generally to Computed Tomography (CT), and more particularly to reducing aliasing in CT systems.
In CT systems, data transmission limitations are caused by, for example, limited bandwidth of the slipring transmission, and computational complexity limitations are caused by, for example, a high number of views that are prepared and back-projected. Such data transmission limitations or computational complexity limitations are overcome by combining two or more views. However, combining views usually reduces final image quality (IQ) and causes aliasing. On the other hand, if views are not combined, aliasing and resolution are worse, and therefore, combining views is imperative.
As such, aliasing obscures anatomy, and makes it more difficult to read an image than the noise in the image, especially on scans of extremities (e.g., ankle, knee, wrist, and the like), C-spine, facial bone, and inner auditory canal scans.
Therefore, there exists a need for reducing aliasing in CT systems.
BRIEF DESCRIPTION OF DRAWINGS
FIG.1 depicts a flow specific design in view direction.
FIG. 2 depicts conditions used in simulations.
FIGS. 3 and 4 shows images generated using the product reconstruction.
FIG. 5 shows noise improvement using the techniques described herein
The present disclosure combines CT image views by applying an anti-aliasing filter and down‑sampling process, instead of a simple image combination. Such a combination reduces aliasing in the sampled data, which flows through to an aliasing reduction in the final image.
The gain in aliasing performance is acceptable as is. However, such gain in aliasing performance can be traded-off for image resolution boosting that otherwise wouldn’t be possible due to generally unacceptable aliasing.
The proposed method operates by collecting views on the detector, applying an anti-aliasing digital filter across the views, and then down-sampling to a reduced number of views. This view reduction method offers an image quality improvement when reduced data transfer across the slipring or reduced computation for prep and back projection is needed.
FIG. 1 illustrates a flow diagram of the method and a specific design in view direction. First, the views are collected on the detector, using the standard view generation on the system. The view sampling rate can be increased in multiple ways, without any limitations due to possible IQ concerns. Then, a digital filter is applied across sub-views, for example, using direct convolution (repeated multiplication and summation across views), FFTs or IFFTs. The digital filter is a well-designed anti-aliasing filter, such as a Kaiser window. The views are then downsampled by reducing the number of views by 10 times. The digital filter from the previous step would only need to be applied on views that would be preserved and passed through the data chain. Interpo...