摘要
目的探讨基于深度学习重建算法(deep learning reconstruction,DLR)的扩散加权成像(diffusion weighted imaging,DWI)在颅脑MRI检查中的应用价值。材料与方法回顾性分析40例颅内占位性病变患者的MRI影像学资料,比较激励次数(number of excitations,NEX)为2和1时常规重建(c2-DWI,c1-DWI)与DLR(DL2-DWI,DL1-DWI)4组图像的质量差异,比较灰质、白质的信噪比(signal-to-noise ratio,SNR)、对比噪声比(contrast-to-noise ratio,CNR),病灶区域及对侧正常区域的表观扩散系数(apparent diffusion coefficient,ADC)。由两位医师采用双盲法对整体图像质量、噪声水平和磁敏感伪影分别使用5分法评分。结果DLR序列灰质和白质的SNR、CNR均高于常规重建序列,差异有统计学意义(P<0.001);ADC值在病变区域及对侧正常区域差异无统计学意义(P>0.05);DLR的整体图像质量和噪声水平评分均高于常规重建,差异有统计学意义(P<0.001);磁敏感伪影差异无统计学意义(P>0.05)。结论DLR可显著提升DWI图像SNR、CNR及主观评分,有效降低图像噪声,在NEX减半,缩短扫描时间的同时,虽对磁敏感伪影改善有限,但不影响ADC值的准确性。
Objective:To explore the application value of diffusion weighted imaging(DWI)based on deep learning reconstruction algorithm(DLR)in cranial MRI examination.Materials and Methods:A retrospective analysis was conducted on the MRI imaging data of 40 patients with intracranial space occupying lesions.Four sets of image quality differences were compared between conventional reconstruction(c2-DWI,c1-DWI)and DLR(DL2-DWI,DL1-DWI)with a number of excitations(NEX)of 2 and 1.The signal-to-noise ratio(SNR)and contrast-to-noise ratio(CNR)of gray and white matter were compared,as well as the apparent diffusion coefficient(ADC)of the lesion area and the contralateral normal area.Two physicians used a double-blind method to score the overall image quality,noise level,and magnetic susceptibility artifacts using a 5-point scale.Results:The SNR and CNR of DLR sequence gray matter and white matter were higher than those of conventional reconstructed sequence,and the difference was statistically significant(P<0.001).There was no statistically significant difference in ADC values between the lesion area and the contralateral normal area(P>0.05).The overall image quality and noise level scores of DLR are higher than those of conventional reconstruction,and the difference is statistically significant(P<0.001).There was no statistically significant difference in magnetic sensitivity artifacts(P>0.05).Conclusions:DLR can significantly improve the SNR,CNR,and subjective score of DWI images,effectively reducing image noise.While NEX is halved and scanning time is shortened,although there is limited improvement in magnetic sensitivity artifacts,it does not affect the accuracy of ADC values.
作者
张晏华
郁仁强
郁斌
吴治伟
赵春刚
万露
万承鑫
张志伟
ZHANG Yanhua;YU Renqiang;YU Bin;WU Zhiwei;ZHAO Chungang;WAN Lu;WAN Chengxin;ZHANG Zhiwei(Department of Radiology,Dazhou Central Hospital,Dazhou 635000,China;Department of Radiology,the First Affiliated Hospital of Chongqing Medical University,Chongqing 400016,China;Department of Radiology,Chongqing Red Cross Hospital(Jiangbei District People's Hospital),Chongqing 400020,China)
出处
《磁共振成像》
北大核心
2025年第7期65-71,共7页
Chinese Journal of Magnetic Resonance Imaging
关键词
扩散加权成像
表观扩散系数
激励次数
深度学习重建
颅内占位性病变
磁共振成像
diffusion weighted imaging
apparent diffusion coefficient
number of excitations
deep learning reconstruction
intracranial space-occupying lesions
magnetic resonance imaging
