Spectral extrapolation of spatially bounded images [MRI application]

Area of Science:

• Image processing
• Signal processing
• Medical imaging

Background:

• Spatially bounded images are confined to a region with a zero background.
• Extending the image spectrum requires prior knowledge of spatial zeros and spectral components.

Purpose of the Study:

• To present a spectral extrapolation algorithm for spatially bounded images.
• To analyze the resolution of the resulting finite support solution.
• To compare the finite support solution with existing methods.

Main Methods:

• Developed a spectral extrapolation algorithm utilizing prior knowledge of spatial zeros.
• Introduced a regularized version for noisy spectral data.
• Evaluated resolution using impulse response characteristics.

Main Results:

• The finite support solution demonstrates space-variant resolution, with better edge feature detail.
• It outperforms the zerofilled solution in both noisy and noiseless scenarios.
• It is comparable or preferable to the Nyquist solution in noisy data cases.

Conclusions:

• The spectral extrapolation algorithm effectively enhances resolution for spatially bounded images.
• The finite support solution provides a valuable alternative, particularly for noisy medical imaging data.
• Algorithm performance is dependent on known spatial zeros and spectral components.