Application Value Of Compressed Sensing Combined With 3D-TSE Sequence In The Diagnosis Of Supraspinatus Tendon Injury

Part Ⅰ The feasibility of CS-3D-TSE sequence in shoulder MR examinationObjective:To evaluate the feasibility of 3D turbo spin echo(TSE)imaging with compressed sensing(CS)for the assessment of shoulder.Methods:24 healthy volunteers were recruited for unilateral shoulder MR examination,including axial 3D-T1WI-TSE sequence,axial CS-3D-T1WI-TSE sequence and axial 2D-T1WI-TSE sequence,in which CS factor was 5.Image quality evaluation included qualitative and quantitative evaluation.Qualitative evaluation image quality was assessed by two physicians using Likert scale.For quantitative evaluation,two doctors take the mean value by measuring the image SNR and CNR.The four regions of interest were respectively placed at the greater tubercle of humerus,Infraspinatus muscle,long head tendon of biceps brachii and the background air 1cm away from the edge of greater tubercle of humerus.SNR and CNR are calculated according to the formula SNR=SI tissue/SI background and CNR=|SI tissue 1-SI tissue 2|/SI background.CNR were divided into three categories.CNR1 was the SNR difference between the humeral head and the long head tendon of biceps brachii/SI background,CNR2 was the SNR difference between the greater tubercle of humerus and the infraspinatus muscle/SI background,and CNR3 was the SNR difference between the infraspinatus muscle and the long head tendon of biceps brachii/SI background.The consistency between the two doctors was evaluated by kappa test,and the qualitative and quantitative evaluation was carried out by one-way ANOVA.Results:The scanning time of the CS-3D-T1WI-TSE sequence was 78.9%and 54.9%shorter than the time of the 3D-T1WI-TSE sequence and the total 2D-T1WI-TSE sequence,respectively.(1)Qualitative analysis:there was good agreement between the 2 physicians regarding the clarity scores of the 3 kinds of sequence anatomical landmarks(kappa values were 0.684,0.711,0.644,respectively);There was good agreement between 2D-T1WI-TSE and 3D-T1WI-TSE sequences in scoring of image artifacts(kappa values were 0.636,0.692,respectively),while there was general agreement between CS-3D-T1WI-TSE sequences in scoring of image artifacts(kappa=0.564).(2)Quantitative analysis:in terms of assessing SNR of the humeral greater tuberosity,there was no statistical difference among the 3 sequences(P=0.565;In terms of assessing SNR in the infraspinatus,there was no statistical difference between 3D-T1WI-TSE and CS-3D-T1WI-TSE sequences(P=0.979),and 2D-T1WI-TSE sequences were superior to 3D-T1WI-TSE and CS-3D-T1WI-TSE sequences with statistically significant differences(P<0.001);In the evaluation of the biceps longus tendon SNR,there was no statistical difference among the 3 sequences(P=0.150).There were no statistical differences in CNR1 and CNR2 among the three sequences(P=0.249,P=0.586),CNR3 between CS-3D-T1WI-TSE sequence and 3D-T1WI-TSE sequence(P=0.676),and CNR3 between 2D-T1WI-TSE sequence and CS-3D-T1WI-TSE sequence and 3D-T1WI-TSE sequence(P<0.001).Conclusion:CS combined with 3D-TSE sequence is feasible for MRI imaging of shoulder joint.While improving the scanning speed,the image quality does not decline significantly.Part Ⅱ Comparative study of oblique coronal CS-3D-PDW-SPAIR sequence and oblique coronal 2D-PDW-SPAIR sequence on the injury of supraspinatus tendon.Objective:To evaluate the diagnostic value of CS-3D-PDW-SPAIR sequence for supraspinatus tendon injury.Methods:Data were collected on patients with a primary clinical diagnosis of rotator cuff tear at our institution,including 2D-TSE sequence image sets,oblique coronal CS-3D-PDW-SPAIR sequence images with a CS factor of 6.5;They were divided into arthroscopic data group and non arthroscopic data group.Two sets of radiologists independently reviewed and recorded 2D-PDW-SPAIR versus CS-3D-PDW-SPAIR sequences for supraspinatus tendon injuries,respectively.Regarding the sensitivity,specificity,accuracy of 2D-PDW-SPAIR versus CS-3D-PDW-SPAIR sequences for supraspinatus tendon injury in the arthroscopic data group,kappa test was performed to compare the two sequences separately for arthroscopic agreement.In the no arthroscopic data group,the kappa test evaluated the agreement between the two groups in physician diagnosis of supraspinatus tendon injury.Results:the total scanning time of the CS-3D-PDW-SPAIR sequence was 36%shorter than that of the 2D-TSE sequence image set(274 s vs 430 s).In the group of 70 patients without arthroscopic data,there was good agreement between the two groups in diagnosing supraspinatus tendon injuries(kappa=0.711).In a cohort of 32 patients with available arthroscopic data,2D-PDW-SPAIR sequences had a sensitivity of 88.5%,specificity of 50.0%,accuracy of 81.3%,and a kappa value for concordance with arthroscopy=0.633;The CS-3D-PDW-SPAIR sequence had a sensitivity of 92.3%,a specificity of 33.3%,an accuracy of 87.5%,and a kappa value of agreement with arthroscopy=0.681.Conclusion:Under the premise of no obvious decrease in image quality,CS-3D-PDW-SPAIR sequence can shorten the total scanning time of 2D-TSE sequence,and has considerable diagnostic performance compared with 2D-PDW-SPAIR sequence in the diagnosis of supraspinatus tendon injury.It is a reliable imaging method for the diagnosis of supraspinatus tendon injury.