Application of deep learning reconstruction in optimizing head DWI scanning time,signal-to-noise ratio and structural similarity
SUN Qi'an
HE Ling
LI Zheng
JIAO Zhiyun
LU Lu
Abstract：Objective To evaluate the value of deep learning reconstruction(SupMR)in optimizing scan time,signal-to-noise ratio(SNR),and structural similarity for head diffusion-weighted imaging(DWI).Methods A total of 40 healthy volunteers were prospectively and consecutively enrolled from December 2024 to January 2025 in Affiliated Hospital of Yangzhou University.All participants underwent head MRI scans using both low number of excitations diffusion-weighted imaging(LN-DWI)and conventional DWI(C-DWI)sequences.The LN-DWI images were post-processed using SupMR to generate Sup-DWI images,with the scan time recorded for each sequence.Three sets of images(LN-DWI,C-DWI,and Sup-DWI)were evaluated by subjective and objective methods(region-of-interest signal-to-noise ratio,contrast-to-noise ratio(CNR),apparent diffusion coefficient(ADC),peak signal-to-noise ratio(PSNR),and structural similarity index(SSIM).Intraclass correlation coefficient(ICC)was used to analyze the consistency of ADC values among the three groups,as well as the intra-and inter-observer agreement of subjective scores between two radiologists.Results Compared with the C-DWI group,the scanning time of Sup-DWI group was reduced by 52%.The results of consistency analysis showed good intra-and inter-observer consistency between the two radiologists(ICC>0.75),and good consistency of ADC values in all brain parenchyma regions between each pair of groups(ICC>0.75).The results of subjective score analysis showed a statistically significant difference among the three groups(P<0.05);pairwise comparison showed no significant difference between Sup-DWI group and C-DWI group(P>0.05).Objective evaluation and analysis showed that the differences in SNR and CNR between the three groups were statistically significant(P<0.05);pairwise comparison showed that the differences in SNR and CNR were statistically significant(P<0.05).The PSNR and SSIM values of the Sup-DWI group were greater than those of the LN-DWI group,and the differences were statistically significant(P<0.05).Conclusion SupMR technology can significantly shorten the DWI scanning time,optimize image quality,and help improve the efficiency of examination.
Keywords：deep learning reconstructiondiffusion weighted imagingimage reconstructionimage quality
Publication Date：2025-11-20
Online Publishing Date：2025-12-16(First online date of this platform, not the publication date of the document)
Pages：6( 1398-1403 )