Comparison of three reconstruction algorithms for low-dose phase-contrast computed tomography of the breast with synchrotron radiation

Background

Phase-contrast breast CT imaging holds promise for improved diagnostic accuracy, but an optimal reconstruction algorithm must balance objective image quality metrics with subjective radiologist preferences.

Purpose

This study systematically compares three reconstruction algorithms—filtered back projection (FBP), unified tomographic reconstruction (UTR), and customized simultaneous algebraic reconstruction technique (cSART)—to identify the most suitable approach for phase-contrast breast CT imaging.

Methods

Fresh mastectomy samples were scanned at the Australian synchrotron using monochromatic 32 keV X-rays, a mean glandular dose of 2 mGy, flat-panel detectors with 0.1 mm pixels, and 6-m distance between the rotation stage and the detector. Paganin's phase retrieval method was used in conjunction with all three CT reconstruction algorithms. Objective metrics, including spatial resolution, contrast, signal-to-noise, and contrast-to-noise, were evaluated alongside subjective assessments by seven experienced radiologists. Ratings included perceptible contrast, sharpness, noise, calcification visibility, and overall quality.

Results

cSART excelled in objective metrics, outperforming UTR and FBP. However, subjective evaluations favored FBP due to its higher image contrast, revealing a discrepancy between objective and subjective assessments.

Conclusions

The findings highlight the contrast-focused nature of radiologists’ subjective assessments and the potential of cSART for delivering superior objective image quality. These insights inform the development of hybrid evaluation tools and guide clinical translation for future live patient imaging studies.