基于压缩感知技术三维MRI用于半月板损伤

目的 探讨压缩感知（CS）技术三维MRI（3D-MRI）用于膝关节成像的可行性及其评估半月板损伤的价值。方法 对26例疑诊膝关节损伤患者（损伤组）及30名健康志愿者（对照组）行膝关节MRI，采集常规脂肪抑制质子加权图像（fsPDWI）和CS-3D-MRI。记录损伤组患者关节镜检查结果。损伤组患者于关节镜检查前接受MRI，以单一DS方式（CS-DS_medium）重建图像，分析CS-3D-MRI诊断结果与关节镜结果的一致性。对照组均行右膝扫描，以3种降噪（DS）水平（CS-DS_weak、CS-DS_medium、CS-DS_strong）重建CS-3D-MRI，采用主、客观方法评价矢状位图像质量。结果 对照组不同方法重建CS图像和fsPDWI图像质量评分差异无统计学意义（Z=0.35，P=0.32）；4个序列图像信噪比（SNR）和对比信噪比（CNR）差异均有统计学意义（F=36.01、9.62，P均<0.05），CS-DS_strong图像SNR最高（P<0.05），fsPDWI与CS-DS_medium图像之间CNR值差异无统计学意义（P>0.05）。损伤组CS-3D-MRI均显示半月板损伤，与关节镜诊断结果的一致性强（Kappa=0.94，P<0.01）。结论 基于CS技术的3D-MRI可用于膝关节成像，能在保证图像质量的前提下缩短扫描时间，评估膝关节半月板损伤有较高价值。

Three dimensional MRI based on compressed sensing technology in diagnosis of meniscal injuries

Objective To investigate the feasibility of three dimensional MRI (3D-MRI) based on compressed sensing (CS) in knee joint imaging and its value in assessing meniscal injuries. Methods Knee MRI were performed on 26 patients with suspected knee injury (injured group) and 30 healthy volunteers (control group). Conventional fat saturation proton density-weighted imaging (fsPDWI) and CS-3D-MRI were collected. The results of arthroscopy of injured group were recorded. MRI of the right knee were obtained in control group, then CS-3D-MRI were reconstructed with 3 different denosing (DS) levels (CS-DS_weak, CS-DS_medium, CS-DS_strong), and the sagittal image quality was evaluated subjectively and objectively. Patients in injured group received MRI before arthroscopy. CS-3D-MRI of injured group were reconstructed with CS-DS_medium, and the consistency of CS-3D-MRI diagnostic results with those of arthroscopy was analyzed. Results For images of control group, there was no statistical difference of the quality scores of reconstructed CS images of different DS levels and fsPDWI (Z=0.35, P=0.32), while statistically significant differences of signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were found in images obtained with 4 different sequences (F=36.01, 9.62, both P<0.05), and SNR of CS-DS_strong images was the highest (P<0.05). No significant difference of CNR in CS-DS_medium image and fsPDWI was found (P>0.05). CS-3D-MRI diagnosed meniscus injuries in all patients in injury group, highly consistent with results of arthroscopy (Kappa=0.94, P<0.01). Conclusion Based on CS technique, 3D-MRI could be used for knee joint imaging, which was able to shorten scanning time on the premise of ensuring image quality, therefore had good value for evaluation on meniscus injuries.