Diagnostic Performance of Different Simulated Low-Dose Levels in Patients with Suspected Cervical Abscess Using a Third-Generation Dual-Source CT Scanner.

Moritz T Winkelmann,Sven S Walter,Vincent Schwarze,Ahmed E Othman,Christoph P Artzner,Manuel Kolb,Eva Stock,Saif Afat,Andreas Brendlin

doi:10.3390/diagnostics10121072

Moritz T Winkelmann, Sven S Walter + Show 7 more

Open Access

https://doi.org/10.3390/diagnostics10121072

Copy DOI

Abstract

The aim of this study was to investigate the effects of dose reduction on diagnostic accuracy and image quality of cervical computed tomography (CT) in patients with suspected cervical abscess. Forty-eight patients (mean age 45.5 years) received a CT for suspected cervical abscess. Low-dose CT (LDCT) datasets with 25%, 50%, and 75% of the original dose were generated with a realistic simulation. The image data were reconstructed with filtered back projection (FBP) and with advanced modeled iterative reconstruction (ADMIRE) (strengths 3 and 5). A five-point Likert scale was used to assess subjective image quality and diagnostic confidence. The signal-to-noise ratio (SNR) of the sternocleidomastoid muscle and submandibular gland and the contrast-to-noise ratio (CNR) of the sternocleidomastoid muscle and submandibular glandular fat were calculated to assess the objective image quality. Diagnostic accuracy was calculated for LDCT using the original dose as the reference standard. The prevalence of cervical abscesses was high (72.9%) in the cohort; the mean effective dose for all 48 scans was 1.8 ± 0.8 mSv. Sternocleidomastoid and submandibular SNR and sternocleidomastoid muscle fat and submandibular gland fat CNR increased with higher doses and were significantly higher for ADMIRE compared to FBP, with the best results in ADMIRE 5 (all p < 0.001). Subjective image quality was highest for ADMIRE 5 at 75% and lowest for FBP at 25% of the original dose (p < 0.001). Diagnostic confidence was highest for ADMIRE 5 at 75% and lowest for FBP at 25% (p < 0.001). Patient-based diagnostic accuracy was high for all LDCT datasets, down to 25% for ADMIRE 3 and 5 (sensitivity: 100%; specificity: 100%) and lower for FBP at 25% dose reduction (sensitivity: 88.6–94.3%; specificity: 92.3–100%). The use of a modern dual-source CT of the third generation and iterative reconstruction allows a reduction in the radiation dose to 25% (0.5 mSv) of the original dose with the same diagnostic accuracy for the assessment of neck abscesses.

Highlights

Cervical abscesses often occur as a complication of infection due to different pathogens such as Streptococcus viridans, Staphylococcus epidermidis, and Staphylococcus aureus
We evaluated the effect of simulated dose reduction on image quality and diagnostic performance in patients with suspected neck abscesses
By means of advanced modeled iterative reconstruction (ADMIRE), a dose reduction to up to 25% of the original dose is possible while maintaining diagnostic accuracy and acceptable image quality

Summary

Introduction

Cervical abscesses often occur as a complication of infection due to different pathogens such as Streptococcus viridans, Staphylococcus epidermidis, and Staphylococcus aureus. To localize the primary extent of infection and to exclude possible complications, contrast-enhanced multi-slice computed tomography is used as the primary diagnostic tool in the clinical work-up [3,4]. To diagnose a cervical abscess and potential complications, the CT scan has to include the region between the base of the skull and the aortic arch [6]. Lowering the radiation dose with standard filter back projection (FBP) will result in increased noise levels of the images with decreasing diagnostic confidence [13]. Iterative reconstruction has been shown to compensate for a lower dose by significantly reducing image noise while maintaining diagnostic confidence [11,13]

Methods

Results

Conclusion