Chaotic Sensing

Area of Science:

• Medical Imaging
• Signal Processing
• Computational Science

Background:

• Sparse imaging techniques are crucial for efficient data acquisition in fields like magnetic resonance (MR) imaging.
• Existing methods like compressed sensing often rely on transform domains, which can be computationally intensive.
• Limited measurements in MR imaging necessitate innovative approaches to maintain image quality and reduce scan times.

Purpose of the Study:

• To introduce Chaotic Sensing (ChaoS), a novel sparse imaging methodology.
• To demonstrate the efficacy of fractal sampling for deterministic linear measurements.
• To enable artifact removal and high-fidelity image reconstruction from limited data.

Main Methods:

• Developed Chaotic Sensing (ChaoS) utilizing fractal sampling within the discrete Fourier transform.
• Introduced a novel fractal that generates image-independent, turbulent artifacts.
• Employed image denoising for artifact dampening and maximum likelihood estimation for image recovery.
• Established finite iterative reconstruction schemes based on digital periodic lines for discrete tomography.

Main Results:

• ChaoS enables the use of limited, deterministic linear measurements through fractal sampling.
• Chaotic artifacts are image-independent, facilitating their removal via denoising.
• The method supports linear measurement and optimization strategies for image recovery.
• Achieved theoretically exact image representation recovery from fractal sampling.

Conclusions:

• Chaotic Sensing (ChaoS) offers a robust sparse imaging solution with efficient artifact management.
• The fractal sampling approach bypasses the need for additional transform domains, simplifying reconstruction.
• ChaoS is particularly advantageous for limited data acquisition scenarios, such as in magnetic resonance imaging, by supporting linear measurements to reduce scan time.