Impact of a silver filter with deep learning reconstruction on noncontrast abdominopelvic MDCT in terms of image quality, density values and radiation dose reduction.

INTRODUCTION

To assess how the use of a silver beam filtration with deep learning reconstruction (DLR) impacts the unenhanced abdominopelvic CT in terms of soft tissues density, image quality and radiation dose.

METHODS

This study evaluated two unenhanced abdominopelvic CT acquisitions of 110 patients performed with a conventional filtration (group A) and a silver filtration (group B). The HU values of several soft tissues and extra corporeal air were recorded for each acquisition using an identical circular ROI. Interobserver agreement and subjective qualitative analysis of image were assessed. The student's paired t-test was used to evaluate differences between the two examinations in terms of HU values, SNR, CNR, DLP and CTDIvol. Statistical difference was considered as significant at p ≤ 0.05.

RESULTS

The subjective image quality and interobserver agreement were judged excellent without significant difference between both acquisitions. We found lower HU values for the silver filter than for the conventional filter with a percentage decrease between 0.03 and 0.09% except for urine which was found at 38%. However, there was no statistically significant difference except for the subcutaneous fatty tissue and urine. The radiation dose was found to be significantly lower by 36% in the examinations done with silver filter when compared to the conventional filter.

CONCLUSION

Implementing the silver filtration with DLR on non-contrast abdominopelvic CT acquisition achieves a significant dose reduction while preserving image quality, albeit a reduction of the HU value of soft tissues, more intense for positive low-density structure (urine/water).

IMPLICATIONS FOR PRACTICE: This fact must be taken into account in clinical routine especially in the context of indeterminate cyst or adrenal mass on non enhanced acquisition.