Ultra-low dose CT for suspected physical abuse

Ultra-low dose CT (ULDCT) of the whole-body is a promising technique to replace radiography for the diagnosis of suspected physical abuse (SPA). This study aims to compare the image quality and radiation dose of ULDCT to standard dose CT (STDCT) in cases of SPA.

In this phantom-based study, two sets of newborn, whole-body, anthropomorphic phantom images were acquired using STDCT and ULDCT protocols on a GE Revolution Apex scanner with Deep Learning Iterative Reconstruction (DLIR). The effective dose (ED) of both protocols was calculated using Monte Carlo simulation. Observers then assessed image quality (IQ) at an international radiology congress. The evaluation captured data on demographics, visualisation of the bony anatomy of all body parts with STDCT and ULDCT, and confidence in diagnosis using either protocol. Visual grading analysis (VGA) was used on an absolute scale for IQ rating and comparison.

Forty-six observers were included in this study, 38-radiographers and eight-radiologists. The percentage ED difference between protocols was 93.5% (STDCT=0.56mSv vs ULDCT=0.04mSv) and 97.8% of the observers underestimated the dose reduction when questioned. For the bony anatomy of all body parts, STDCT showed significantly higher IQ than ULDCT (AUC_VGA=0.75, asymmetric 95% CI 0.69-0.8).

This study demonstrated that 41% of observers were confident using ULDCT protocol for SPA diagnosis and that ULDCT is a promising technique which may compete with projection radiography. Further work is needed to improve the ULDCT protocol and increase confidence while maintaining an ultra-low dose. ULDCT should be considered as a potential addition to radiographic skeletal survey (SS) in the investigation of SPA.

ULDCT provides substantially lower radiation dose with acceptable image quality in a phantom model. While not directly compared with radiographic SS, ULDCT shows potential as a complementary tool for SPA imaging, and further protocol optimisation and evaluation are required before clinical implementation.