Sparse-view, short-scan, dedicated cone-beam breast computed tomography: image quality assessment

Area of Science:

• Medical Imaging
• Radiology
• Image Reconstruction

Background:

• Dedicated cone-beam breast computed tomography (BCT) utilizes full-scan acquisition (360 degrees).
• Sparse-view acquisition and shorter scan trajectories are being explored to reduce scan time and radiation dose.
• Image quality assessment is crucial for evaluating the efficacy of novel BCT acquisition techniques.

Purpose of the Study:

• To quantify the impact of sparse-view acquisition in short-scan trajectories on image quality in dedicated cone-beam breast CT.
• To compare short-scan BCT image quality with 360-degree full-scan acquisition.
• To evaluate the feasibility of iterative reconstruction techniques for short-scan BCT.

Main Methods:

• Utilized projection data from 30 full-scan BCT exams with calcified lesions.
• Reconstructed images using Feldkamp-Davis-Kress (FDK) for full-scan data (reference) and a fast, iterative, total variation-regularized, statistical reconstruction technique (FIRST) for short-scan data (204 and 270 degrees).
• Quantified image quality using variance, signal-difference to noise ratio (SDNR), full-width at half-maximum (FWHM) of calcifications, bias, and root-mean-squared-error (RMSE).

Main Results:

• Short-scan reconstructions showed bias and RMSE close to, but not zero, compared to the FDK reference.
• FWHM of calcifications in short-scan reconstructions did not significantly differ from the reference, except in one specific direction.
• Variance and SDNR were significantly improved in short-scan reconstructions compared to the full-scan FDK reconstruction.

Conclusions:

• Demonstrates the feasibility of short-scan, sparse-view, compressed sensing-based iterative reconstruction in BCT.
• Shorter scan times and reduced radiation dose are potentially achievable without sacrificing image quality.
• Iterative reconstruction techniques like FIRST show promise for optimizing BCT imaging protocols.