Enhanced hyperspectral tomography for bioimaging by spatiospectral reconstruction

Area of Science:

• Medical Imaging
• Biophysics
• Computational Imaging

Background:

• Conventional energy-dispersive computed tomography (CT) suffers from low count-rates and poor signal-to-noise ratios.
• This necessitates high X-ray doses and long scan times for imaging spectral markers in biological samples.
• Accurate visualization of low-concentration, inhomogeneous stains in soft tissues remains challenging.

Purpose of the Study:

• To develop and validate a novel hyperspectral bright field imaging technique for computed tomographic (CT) imaging.
• To enable high-resolution mapping of stain distribution in fixed biological samples using K-edge imaging.
• To improve image quality and reduce scan times and radiation dose in spectral CT.

Main Methods:

• Application of hyperspectral bright field imaging to collect computed tomographic (CT) images with ~1 keV energy resolution.
• Development of a dedicated iterative reconstruction algorithm incorporating spatial smoothness and inter-channel correlation for low-dose, noisy datasets.
• Evaluation of spectral analysis methods, including K-edge subtraction and absorption step-size fitting.

Main Results:

• Achieved high-quality energy-dispersive tomograms from low-dose, noisy datasets.
• Demonstrated a 36-fold reduction in scan time for a multi-phase phantom.
• Successfully evaluated spectral analysis for an ex vivo iodine-stained biological sample.

Conclusions:

• The developed iterative reconstruction algorithm significantly enhances image quality in spectral CT from low-dose data.
• This approach offers new capabilities for visualization and elemental mapping in biological specimens.
• The open-source availability of reconstruction algorithms facilitates broader application in biomedical research.