A high-resolution photon-counting breast CT system with tensor-framelet based iterative image reconstruction for...

Area of Science:

• Medical Physics
• Radiology
• Image Reconstruction

Background:

• High-resolution spectral breast CT utilizes silicon strip photon-counting detectors.
• Traditional filtered-back-projection (FBP) may not optimize radiation dose reduction and image quality.
• Iterative reconstruction methods offer potential for dose reduction and enhanced image quality.

Purpose of the Study:

• To investigate radiation dose reduction using tensor framelet based iterative image reconstruction (TFIR) in spectral breast CT.
• To compare image quality metrics (CNR, spatial resolution) between TFIR and FBP.
• To evaluate the feasibility of TFIR for clinical implementation.

Main Methods:

• Computer simulations with a 10 cm water phantom and experimental phantom studies with a 1.3 cm PMMA phantom.
• Inclusion of contrast materials (iodine, polyethylene, bone-equivalent plastic) at varying concentrations.
• Image reconstruction using both FBP and TFIR techniques, followed by quantitative analysis of contrast-to-noise ratio (CNR) and task-based modulation transfer function (MTF).

Main Results:

• TFIR achieved comparable spatial resolution to FBP at one-third the radiation dose for low contrast targets.
• TFIR demonstrated substantially superior spatial resolution for high contrast targets compared to FBP.
• TFIR resulted in a 1.6-1.8 times increase in CNR, depending on contrast levels.

Conclusions:

• TFIR enables significant radiation dose reduction (up to two-thirds) in spectral breast CT while maintaining or improving image quality.
• TFIR offers superior CNR and spatial resolution for high contrast targets, making it advantageous for specific diagnostic tasks.
• GPU-accelerated TFIR reconstruction is rapid (approx. 10s/slice), supporting its potential integration into future CT systems.