Comparison Between MRI 3D Cube And Conventional 2D Sequence In Shoulder Joint

Objective To compare the sensitivity,specificity and accuracy of 3.0T MRI conventional 2D sequence and 3D cube sequence in the diagnosis of rotator cuff injury;to optimize the imaging parameters of 3D cube sequence,and to explore the clinical value of 3D cube sequence in patients with rotator cuff injury.Methods Twenty healthy volunteers and 30 patients with clinically diagnosed with rotator cuff injury were included,and all of them signed informed consent prior to scanning.Including 10 male and 10 female volunteers(average age:26.32±3.73years),and 14 male and16 female patients(average age:53.67±7.49years).Conventional 2D sequences and 3D Cube sequences of shoulder joint were collected in 3.0T MRI scanner.GE AW4.6 post-processing workstation uses sagittal 3D Cube raw data to obtain reconstructed 3D Cube images of each plane.Two magnetic resonance attending physician using 5 points evaluation method from five aspects(anatomic structure(the supraspinatus tendon,labrum,biceps long head tendon and bone marrow)display,image artifacts,image blur,fat suppression,overall quality)respectively of shoulder joint conventional 2D and 3D Cube sequences with subjective evaluation.A physician measured the signal strength(SI)and standard deviation(SD)of suprapathonic tendons,bone marrow,and fluids,and calculated the signal-to-noise ratio(SNR)and the contrast-to-noise ratio(CNR)of each anatomical structure.All patients with rotator cuff injury were confirmed by arthroscopy.The results of the arthroscopy were taken as the gold standard to calculate the sensitivity,specificity and accuracy of the conventional 2D sequence and 3D Cube sequence of the two physicians respectively.With α=0.05 as the test standard,Interclass correlation coefficient(ICC)was used to calculate the consistency of the two physicians in the subjective evaluation of image quality and the sequential diagnosis of rotator cuff injury in the two groups.Wilcoxon symbolic rank sum test was used to compare the differences between the subjective scores of 2D and 3D Cube sequences.One-way ANOVA was used to compare SNR and CNR of 3D Cube sequences with conventional 2D sequences.LSD test was used for homogeneity of variance,Dunnett’s test was used for heterogeneity of variance.Compare the sensitivity and specificity of rotator cuff injury between the two groups.Results1.Inter-observer agreements: anatomical structure(the supraspinatus tendon,labrum,biceps long head tendon and bone marrow)display,image artifacts,image blur,fat suppression,the overall quality of all aspects both observer consistency of 0.600 or higher,including 3D Cube blurred images show the consistency of the lowest(ICC = 0.600),2D image artifacts has the highest consistency(ICC=0.905).Including the supraspinatus tendon,labrum,bone marrow,image artifacts,fat suppression,the overall image quality of 2D sequence,and the ICC value of labrum display,image artifacts,fat suppression of 3D sequence were > 0.75,shows that in terms of the above observations with high consistency between the observer and other aspects of the ICC value is 0.60 or higher,that has the high consistency between the observer.2.Subjective evaluation: The results of conventional 2D sequences were all higher than3 D Cube in all aspects,and the differences in the scoring results of suprapassus tendon,labrum,biceps longhead tendon,bone marrow,image blur,fat suppression and overall image quality were not statistically significant(P > 0.05).Compared with conventional 2D sequences,3D Cube sequences had a lower subjective score of image blur and the difference was statistically significant(3.93±0.57;F=4.207,P = 0.023).3.Objective evaluation: The 2D sequence showed significantly higher SNR of supraspinatus tendon,bone marrow,and liquid compared with 3D Cube sequence(P <0.05).The liquid of 2D sequence was the highest SNR(39.02±6.89).The SNR of the supraspinatus tendon of the 3D Cube sequence was the lowest(2.80±0.57).The liquid-marrow of 3D Cube sequence was the highest CNR(21.63±8.57),and the tendon-marrow of 2D sequence was the lowest CNR(1.14 ±0.22).The CNR between 2D sequence and 3D Cube sequence of liquid-tendon,liquid-marrow,tendon-marrow were statistically significant(P<0.05).4.Consistency of rotator cuff injury: The ICC value of consistency between the two physicians’ diagnosis results was 0.834(P < 0.001),and there was a high degree of consistency between the two physicians and between the conventional 2D sequence and the3 D Cube sequence in the diagnosis of rotator cuff injuries.5.The value of the two groups in the diagnosis of rotator cuff injury: This study included a total of 30 patients with clinically diagnosed rotator cuff injury,29 cases of rotator cuff injury were diagnosed by shoulder arthroscopy,including 14 cases of simple supraspinatus tendon tear,6 cases of supraspinatus tendon combined with biceps long head tendon tear,4cases of supraspinatus tendon combined with subscapular tendon tear,2 cases of simple subspinatus tendon tear,and 2 cases of supraspinatus tendon tears with biceps long head tendon and subscapular tendon,1 case with simple biceps long head tendon tear,with or without subacromial deltoid bursa,subcoracoid bursa,and brachial fluid accumulation in the tendon sheath of the biceps tendon.Among them,there were 16 cases of partial tearing and 13 cases of full-thickness tearing/fracture.The AUC values of the two sets of sequences were both ≥0.75(P > 0.05),among which the 3D Cube sequence of physician I was the highest(0.854),and the sensitivity,specificity and accuracy of the diagnosis of rotator cuff injury by physician I and physician II in the two sequences were both ≥96.6%,there was no statistical significance between the two physicians diagnose partial rotator cuff tears and full-thickness tears(P > 0.05).Conclusion1.3D Cube sequence showed lower scores and SNR values in all aspects of shoulder joint display than conventional 2D sequence,but the CNR value of 3D Cube sequence was significantly higher than that of conventional 2D sequence.2.3D Cube sequence can be used to observe the surrounding structures through multiplanar reconstruction.3.There was no statistical significance between the two sequences diagnose partial rotator cuff tears and full-thickness tears.The sensitivity,specificity and accuracy of the diagnosis of rotator cuff injury were ≥96.6%.4.3D Cube sequence can be used as a supplementary sequence for shoulder joint scanning,through its thin-layer scanning and multi-planar reconstruction to improve the imaging diagnosis of rotator cuff injuries.