20-10-2012, 01:37 PM
Ultrasound Image Enhancement using Super Resolution
Ultrasound Image.ppt (Size: 4.05 MB / Downloads: 72)
Spatial Resolution
Spacing between the pixels in an image, measured in pixels per inch. It is divided in to two categories
Axial resolution: It is the spatial resolution in the direction of the propagation of the ultrasound wave.
Ultrasound frequency ↑ Axial resolution↑ Depth of penetration ↓
Lateral Resolution: It is the spatial resolution in the direction perpendicular to the propagation of the ultrasound wave.
Number of array elements in Transducer ↑ Beam width ↓ Lateral Resolution ↑ Cost ↑.
Why Super Resolution ?
For enhancement of the resolution of an image.
Increasing resolution by sensor modification may not be an option.
Better than Interpolation.
It will be very useful in medical image reconstruction where multiple frames of same scene is used as raw data.
The LR image pixels are aligned by registration. So, possibility of alias is lessened.
Cost, less as existing LR imaging systems can be utilized with minimum modification of hardware.
Different approaches of Super Resolution Techniques
Non-uniform interpolation technique.
POCS technique.
Gerchberg-Papoulis technique.
Iterative back-projection (IBP) technique.
Frequency domain technique.
Regularized SR reconstruction approach, etc.
Non-uniform interpolation approach
Registration Interpolation Restoration
Advantages
Takes relatively low computational load.
Makes real-time application possible.
Disadvantages
Degradation models are limited ( applicable when the blur and noise characteristics are same for all LR images).
Optimality is not guaranteed as restoration stage ignores the interpolation errors.
conclusion
A new Super-resolution reconstruction algorithm in polar domain is developed which is applicable directly on image data available in polar format.
The new technique can be employed to the existing ultrasound imaging systems with minimal change of existing hardwire of the systems.
The new modified technique improves the image quality with respect to conventional SR and LR images.
The new technique provides a way to model non-rigid type of random motions between multiple lower resolution data (obtained in polar co-ordinate).
Noise suppression capability is also better in the case of modified RSR technique compare to modified CSR technique.