Cartilage volume quantification with multi echo data image combination sequence in swine knee at 3.0 T MRI

Objective To investigate the accuracy and reproducibility of multi echo data imagine combination (MEDIC) sequence with water excitation at 3. 0 T in swine knee cartilage. Methods Sagittal MEDIC sequences (0.6 mm slice thickness, isotropic) were acquired twice at 3. 0 T MRI in 30 swine knees. The knee cartilage was then removed and the volume was directly measured with water substitution method. The cartilage volume was also determined with a validated open source image software OsiriX by two observers ( A and B). The cartilage volumes obtained by two methods were compared. The reproducibility of MEDIC for quantitative measurement was accessed by the root-mean-square (RMS) of variation coefficient. Interobserver and intraobserver precision errors were compared using a paired students t-test. The accuracy of MEDIC for quantitative measurement was determined by the random pairwise differences, systematic pairwise differences and the Pearson correlation coefficients. Time of semiautomatic and manual segmentation were recorded. Results Time was saved about 75% by using semiautomatic segmentation methods [(4.0 ±1.5) min] versus manual segmentation [(16.0 ±0.9) min]. Interobserver precision errors (RMS CV%for paired analysis) between A and B for cartilage volume measurement were (2. 66 ±0. 82) ml and(2. 61 ±0.81) ml, t=0.24, P=0. 81 (patella); (2.40 ±0.69) ml and(2.49 ±0. 85) ml, t= -0.45,P =0. 65 (medial femoral condyle) ;(2. 28 ±0.74) ml and(2.41 ±0. 78) ml, t = -0. 66, P = 0. 51 (lateral femoral condyle);(3.43±1.28) ml and(3. 51 ±1. 08) ml, t= -0.26, P=0.79 (femora trochlea) with sagittal MEDIC. Intraobserver precision errors ( RMS CV% for paired analysis) of observer A for the first and second cartilage volume measurement were(2. 64 ±0. 62) ml and(2. 67 ±0. 60) ml, t = -0.19, P =0. 85(patella); (2.43 ±0.60) ml and(2. 39 ±0.59) ml, t =0.26, P = 0. 80 (medial femoral condyle);(2. 26 ±0. 56) ml and (2. 30 ±0. 57) ml, t = - 0.27, P = 0. 78 (lateral femoral condyle); (3.40 ±1.20) ml and(3.47 ±1. 11) ml, t = -0.23, P = 0. 82 (femora trochlea) with sagittal MEDIC. The cartilage volume of direct measurement were (2.73 ±0.97 ) ml ( patella ) , (2. 28 ±0. 66 ) ml ( medial femoral condyle) ,(2. 18 ±0. 55) ml (lateral femoral condyle) and (3. 39 ±1.31) ml (femora trochlea).Correlation coefficients between MEDIC and direct measurement for cartilage volume measurement were high [r = 0.95, t = 16. 10, P<0.05 ( patella); r=0.92, t = 12.42, P<0.05 (medial femoral condyle); r =0.94, t = 14.58, P < 0.05 (lateral femoral condyle) and r = 0.98, t = 26.06, P < 0. 05 (femora trochlea) ]. Conclusion Using image software OsiriX, sagittal MEDIC sequence permits accurate and precise measurement of cartilage volume at reasonable acquisition and segmentation time at 3.0 T MRI in swine knee. Key words: Cartilage, articular;  Organ size;  Magnetic resonance imaging;  Image processing,computer-assisted