Congress:
EuroSafe Imaging 2020
Keywords:
Performed at one institution, Not applicable, Retrospective, Image verification, Dosimetric comparison, Radiation safety, Dosimetry, CT-Angiography, CT, Radioprotection / Radiation dose, Paediatric, Head and neck, Action 7 - Radiation protection of children
Authors:
B. Habib Geryes, R. LEVY, N. Boddaert, C.-J. ROUX
DOI:
10.26044/esi2020/ESI-03962
Description of activity and work performed
Examinations were performed, in our paediatric department, on a Revolution HD CT scanner (General Electric Healthcare, installed in 2017) with an adaptive statistical iterative reconstruction (ASIR-V). The local median value of CTDIvol, calculated on 0-1 y patients who underwent head CT scanner for trauma indication using the initial routine scan protocol, exceeded the updated national DRL. Thus, a multidisciplinary working group, composed of two paediatric neuro-radiologists, medical physicist, referent technologist and vendor application specialist, was involved to perform an iterative optimisation process. In order to reduce the CTDIvol values, the CT scan protocols were modified (kV, noise index, mA range or level of the reconstruction algorithm for enhanced contrast) and evaluated in terms of radiation exposure and image quality [7]. Median values of CTDIvol were calculated and compared with the DRL after protocol modifications. Image quality was evaluated on CATPHAN® 600 phantom and on patients by two experienced paediatric neuro-radiologists. For each scan protocol, radiologists evaluated whether or not image quality was sufficient for diagnostic purposes, especially with regard to the required contrast for the differentiation between white and grey matters. When image quality was deemed insufficient, the working group modified again the CT scan protocol accordingly and carried out another evaluation of CTDIvol and image quality. The iterative optimisation process was stopped, after three iterations of protocol modifications, when the compromise was reached between lowest median CTDIvol with sufficient image quality for diagnostic purposes.
The figure 1 shows the evolution along study periods of median values of CTDIvol. Figure 2 shows patient images along the four study periods. A total of 131 patients (0-1 y) who underwent head CT for trauma indication were included in this retrospective study. Thirty-eight, 11, 28 and 54 patients had respectively undergone the initial CT scan protocol and the three modified protocols through optimisation iterations. During the first period (between January and February 2019), the image quality was deemed sufficient (Figure 2A) but the median CTDIvol value (22 mGy) exceeded the new DRL (20 mGy) with the initial protocol. A first iteration of protocol modifications was carried out to reduce the CTDIvol mainly by reducing the kV. During a second period (March 2019), the median CTDIvol value (18 mGy) was under the DRL but the image quality was judged to be generally insufficient for diagnostic due to the limited contrast in certain areas of the brain (Figure 2B). A second iteration of protocol modifications was carried out mainly on the min-max mA range in order to be close to the DRL. During the third period (between April and May 2019), the median CTDIvol value was at the same level of the DRL (20 mGy) and the image quality was judged sufficient for diagnostic (Figure 2C). Following the application specialist advice, a third iteration of protocol modifications was carried out by using an enhanced contrast reconstruction algorithm and increasing a little the noise index. During the fourth period (between June and August 2019), the median CTDIvol value was 19.5 mGy but the image quality was judged insufficient for diagnostic (Figure 2D). Even if the image contrast was better than in the previous study period, there is indeed more blurring which limits the interpretation in the critical anterior areas of the brain (Figure 2D’). The two readers gave similar results in their evaluation. Therefore, the working group decided to validate the third period protocol that guaranteed the lowest CTDIvol level in accordance with the needed image quality.
The results of image quality assessment on CATPHAN® 600 phantom with the four studied CT acquisition protocols are presented in table 1. The four different studied protocols give comparable image qualities. The 3rd validated protocol seems to have given a compromise.
Table 1: Signal to Noise Ratio (SNR), spatial resolution and low contrast resolution on CATPHAN® 600 phantom with the four different CT acquisition protocols using ArtiscanTM software version 4.6.1.0 (Aquilab, France).
Study Period
|
1
|
2
|
3
|
4
|
Signal to Noise Ratio (SNR)
|
352,66
|
341,49
|
346,86
|
348,01
|
Spatial Resolution (pl/cm)
|
10,16
|
10,28
|
10,29
|
10,12
|
Low contrast Resolution
|
2,0
|
3,0
|
3,0
|
2,0
|