Objective:To evaluate the performance of deep learning reconstruction (DLR) in visualizing lenticulostriate arteries (LSAs) on cerebral CT angiography (CTA).Methods:This cross-sectional study retrospectively analyzed cerebral CTA from 38 patients who underwent cerebral CTA at the Fourth Affiliated Hospital of Harbin Medical University between January and December 2023. Images were reconstructed using filtered back projection (FBP), three-dimensional adaptive iterative dose reduction (AIDR), and DLR-advanced inteuigent clear-IQ engine(AiCE) algorithms (FBP group, AIDR group, DLR-AiCE group). On axial images, the mean CT values, standard deviation (SD), signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were measured at the origin of LSAs, cerebrospinal fluid in lateral ventricles, temporal muscle, and head of the caudate nucleus. Subjective evaluations were performed for overall vascular visualization and LSAs delineation. Comparisons of subjective and objective evaluation indexes among the 3 groups were performed using the complex measurement ANOVA, Friedman test, or χ2 test. Results:The CT, SD, SNR and CNR values at the origin of LSAs, cerebrospinal fluid in lateral ventricles, temporal muscle, and head of the caudate nucleus demonstrated statistically significance among DLR-AiCE group, AIDR group and FBP group ( P<0.001), in which, except for the difference between the FBP group and the AIDR group in the CT value of the head of the caudate nucleus and the CT value of the cerebrospinal fluid of the lateral ventricle which was not statistically significant ( P>0.05), the remaining pairwise comparisons between the groups for each site measurements were statistically significant ( P<0.001). The difference in the overall comparison of the subjective scores of the overall vessels and LSAs in the images of the DLR-AiCE group, the AIDR group, and the FBP group was statistically significant ( P<0.001), and the two-by-two comparisons showed a statistically significant difference ( P<0.001) except for the difference in the subjective scores of LSAs between the FBP group and the AIDR group. Conclusion:The DLR-AiCE algorithm significantly reduces image noise and improves image quality, enabling superior visualization of LSAs, thereby enhancing diagnostic confidence.

Objective:To evaluate the performance of deep learning reconstruction (DLR) in visualizing lenticulostriate arteries (LSAs) on cerebral CT angiography (CTA).Methods:This cross-sectional study retrospectively analyzed cerebral CTA from 38 patients who underwent cerebral CTA at the Fourth Affiliated Hospital of Harbin Medical University between January and December 2023. Images were reconstructed using filtered back projection (FBP), three-dimensional adaptive iterative dose reduction (AIDR), and DLR-advanced inteuigent clear-IQ engine(AiCE) algorithms (FBP group, AIDR group, DLR-AiCE group). On axial images, the mean CT values, standard deviation (SD), signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were measured at the origin of LSAs, cerebrospinal fluid in lateral ventricles, temporal muscle, and head of the caudate nucleus. Subjective evaluations were performed for overall vascular visualization and LSAs delineation. Comparisons of subjective and objective evaluation indexes among the 3 groups were performed using the complex measurement ANOVA, Friedman test, or χ2 test. Results:The CT, SD, SNR and CNR values at the origin of LSAs, cerebrospinal fluid in lateral ventricles, temporal muscle, and head of the caudate nucleus demonstrated statistically significance among DLR-AiCE group, AIDR group and FBP group ( P<0.001), in which, except for the difference between the FBP group and the AIDR group in the CT value of the head of the caudate nucleus and the CT value of the cerebrospinal fluid of the lateral ventricle which was not statistically significant ( P>0.05), the remaining pairwise comparisons between the groups for each site measurements were statistically significant ( P<0.001). The difference in the overall comparison of the subjective scores of the overall vessels and LSAs in the images of the DLR-AiCE group, the AIDR group, and the FBP group was statistically significant ( P<0.001), and the two-by-two comparisons showed a statistically significant difference ( P<0.001) except for the difference in the subjective scores of LSAs between the FBP group and the AIDR group. Conclusion:The DLR-AiCE algorithm significantly reduces image noise and improves image quality, enabling superior visualization of LSAs, thereby enhancing diagnostic confidence.