A Data-Centric Strategy for Developing CT Dose and Noise Reference Levels from Clinical Patient Populations
To develop a data-centric strategy solution for developing CT dose and noise reference levels across large clinical patient populations and in CT scanners.
METHOD AND MATERIALS
This IRB-exempt study evaluated CT abdominopelvic (AP)-related examinations performed in 2017 by 22 scanners from two vendors with 11 models in 3 site hospitals. An in-house developed informatics system automatically extracted protocol information, patient size (cross-sectional diameter), dose, and in vivo noise magnitude within images. Protocol nomenclature categorization was performed using a decision tree machine learning algorithm. Four reference patient size intervals were identified: 13-20, 20-30, 30-40, and 40-50 cm. Noise Reference Level (NRL), Noise Reference Range (NRR), Dose Reference Level (DoRL), and Dose Reference Range (DoRR) were defined for each size range as the median and interquartile interval of noise and dose, respectively.
60,000 CT AP studies with 64 different convolution kernels for patients ages 0-70 and sizes 13-48 cm were identified. NRLs ranged between 15.8 to 18.4 HU with NRRs for the four size ranges were the following: 13.2-24.7,12.6-22.5,12.5-23.2, and 12.1-22.8 HU. DoRLs ranged within 11.9-16.1 mGy. The four DoRRs were 9.5-21.4,7.9-17.3,9.9-21.3, and 10.8-23.2mGy.
This study offers the first even data-crunching solution for developing CT dose and noise reference levels using clinical patient data. New reference levels and ranges simultaneously consider image noise and radiation dose information across patient populations. The new metrics enables prospective optimization of clinical practice to maximize the imaging benefit and patient safety.
A new solution is introduced for simultaneously defining image quality and dose reference levels across different patient body habitus. The methodology enables prospective optimization of clinical practice to maximize the imaging benefit and patient safety.
Ding, A; Ria, F; Zhang, Y; Solomon, J; Samei, E
RSNA 104th Scientific Assembly and Annual Meeting
Conference Start Date
Conference End Date