Application of deep learning reconstruction algorithms in head and neck CT imaging
SU Lei
WEI Kaitong
HU Lili
LIANG Xiaoxue
MA Yaqing
SUN Qiang
Abstract：Objective This study aims to explore the application of deep learning reconstruction algorithms in head and neck CT imaging.Methods A total of 56 patients undergoing facial CT imaging due to trauma,infectious lesions,or tumors were randomly divided into a standard dose group(Group A,28 patients)and a low dose group(Group B,28 patients).The venous phase scan data for both groups were reconstructed using the 50%iterative reconstruction algorithm(ASIR-V 50%),DLIR-M and DLIR-H,with subgroups named AAS-50%,ADL-M,ADL-H,BAS-50%,BDL-M,and BDL-H.Statistical differences in image noise,signal-to-noise ratio(SNR),contrast-to-noise ratio(CNR),and subjective image quality scores between the subgroups were com-pared.Results The BAS-50%group exhibited the highest standard deviation(SD),while,the ADL-H group had the lowest SD value.The SNR and CNR of the DL-M and DL-H groups in both Group A and Group B were higher than those of the ASIR-V 50%group,and their SD values were lower,with all differences being statistically significant.Comparisons of the ADL-H group with the BDL-M and BDL-H groups showed statistically significant differences,as did the comparison between the BDL-M and BDL-H groups.Even with a 33%reduction in radiation dose,the subjective evaluations of the DL-M and DL-H groups remained higher than those of the ASIR-V 50%group.Conclusion This study demonstrates that deep learning reconstruction algorithms significantly im-prove image quality in head and neck CT imaging,while effectively reducing radiation doses.
Keywords：Deep learningTomographyX-ray computedMaxillofacial regionImage reconstructionImage qualityRa-diation dose
Publication Date：2025-07-30
Online Publishing Date：2025-08-26(First online date of this platform, not the publication date of the document)
Pages：4( 23-26 )