Abstract:Objective To explore the feasibility and application value of compressed sensing technique in fetal brain magnetic resonance imaging (MRI) and evaluate the effect of different noise reduction levels on the image.Methods A retrospective analysis was performed on 50 fetal cases of craniocerebral magnetic resonance imaging from December 2018 to April 2019 to explore the differences of imaging data and clinical related information between using conventional 2D fast sequence (BTFE-TFE,as A group) and compressed sensing three-dimensional T2-weighted imaging (CS-3D-T2W).Results According to different noise reduction levels on CS-3D-T2W,CS-DSW,CS-DSM,CS-DSS imagings were reconstructed and obtained (as B,C,D groups),and there were significant differences in signal-to-noise ratio (SNR) among four groups ［126±70.73 in A group,171.72±65.22 in B group,165.66±22.11 in C group,185.22±90.64 in D group,P<0.05 ］.There were no statistical differences in contrast to noise radio (CNR) and artifact of fetal movement(P>0.05).In qualitative evaluation,the consistency of image quality scores by blind image quality assessment in B,C and D groups was higher than that in A group (κ=0.97).Ultrasound diagnosis was consistent with MRI results.Conclusion Compressed sensing can improve the image quality of fetal brain MRI and the accuracy of diagnosing fetal central nervous system diseases,reduce scanning time and enhance the safety for both the pregnant woman and fetus.