Author: Matthew R. Hoerner, Maryam Naseri, Mena Shenouda 👨🔬
Affiliation: Yale University School of Medicine, Yale University 🌍
Purpose: To evaluate acquisition parameters of computed tomography (CT) scanner protocols across different machines to provide patients and clinicians consistent care and image quality, respectively.
Methods: CT protocols were extracted from three (General Electric) scanners, Scanners A, B, and C. Protocols of Scanner A were updated recently and used as the reference standard with which to compare the other protocols using an in-house automated method. Two metrics were introduced to combine parameters of importance that considered patient dose and image noise: the “dose index” and the “image noise index.” For a given acquisition, the former metric was quantified as the product of the tube current modulation noise parameter (noise index) and the square root of the native slice thickness. The latter metric was quantified by dividing the dose index by a statistical iterative reconstruction noise reduction factor and the square root of the reconstructed slice thickness.
Results: There were differences in 232 out of 333 (~70%) protocols that shared the same name between Scanners A and B. There were differences in 228 out of 254 (~90%) protocols for Scanners A and C. The average percent error for the dose index between Scanners A and B was 25.8% and 42.69% between Scanners A and C. The average percent error for the image noise index between Scanners A and B was 38.3% and 44.67% between Scanners A and C. The slice thickness profile and the noise reduction factor setting often led to differences in the metrics. The type of filter used for reconstructions was also occasionally different.
Conclusion: Through the derived metrics, this work quantified differences in CT protocols that impacted patient dose and image noise, factors that should not vary for the same protocols across different scanners of similar make and model. Future work will incorporate image spatial resolution and contrast.