Background
Methods
Specimens
Ultrasonographic cartilage thickness measurement
Contrast-enhanced micro-CT
Validation
Statistical analysis
Results
Parameter | ICC intraobserver 1 | ICC intraobserver 2 | ICC interobserver |
---|---|---|---|
Cartilage thickness | 0.992 (0.986–0.996) | 0.994 (0.990–0.997) | 0.993 (0.990–0.995) |
Discussion
Author, year | Joint | Species | Setup |
n
| Ground truth | Ultrasonic measurement | Accuracy |
---|---|---|---|---|---|---|---|
Aisen et al., 1984 [21] | Knee | Bovine | Cadaver a | 3 | Ruler | B-mode, 7.5 MHz, 1750 m/s speed of sound, immersed mineral oil bad | Mean difference ranging from 0.1 to 0.2 mm (cartilage thickness from 1.0 to 4.2 mm) |
Modest et al., 1989 [25] | Hip | Human | Cadavera | 24 | Microscope Thawed cadaver, toluidine O staining | A-mode, 7.5 MHz, 1760 m/s speed of sound, thawed plugs | Comparison of the average of 5 ultrasonic with 12 microscopic measurements; 88% of the ultrasonic values within one standard deviation of the microscopic measurements |
Martino et al., 1993 [22] | Knee | Human | In vivob | 18 | Microscope Fixed in 90% ethanol, resection following TKA | B-mode, 7.5 MHz, fixed in 90% ethanol, transcutaneous | Significant difference of 0.3 mm for the minimum cartilage difference, no significant difference of 0.2 mm for the maximum cartilage thickness |
Jurvelin et al., 1995 [4] | Knee | Canine and bovine | Cadavera | 45 | Microscope/needle probe Temporarily stored in 4 °C Ringer solution | A-mode, 10 MHz, temporarily stored at 4 °C Ringer solution, 1760 m/s speed of sound | A strong correlation was found between ultrasonic and microscopic measurement (r = 0.91) or needle probe measurement (p = 0.90). The mean difference between microscopic and ultrasonic measurement was 0.03 ± 0.16 mm |
Ostergaard et al., 1995 [18] | Knee | Human | In vivob | 39 | MRI 1.5 Tesla, 3D fast imaging by steady-state precession, slice thickness of 1.56 mm | B-mode, 7.5 MHz, 1750 m/s speed of sound, transcutaneous | Mean difference of 0.2 mm (95% confidence interval − 1.4–1.0 mm) and a strong correlation with r = 0.82 |
Chérin et al., 2001 [19] | Patella | Murine | Cadavera | 48 | Microscope Decalcified and fixated, hematoxylin–eosin–safran staining | B-mode, 50 MHz, measurement within minutes of sacrifice in 0.9% saline solution, 1600 m/s speed of sound | A strong correlation with r = 0.8923 (least square linear fit y = 0.6589x + 79.758) was found |
Töyräs et al., 2002 [20] | Patella | Bovine | Cadavera | 60 | Needle probe Plugs obtained within few hours after slaughter | A-mode, 22 MHz, 1654 m/s speed of sound. Plugs obtained within few hours after slaughter | Mean difference of 0.02 ± 0.99 mm with a strong correlation of r = 0.943. |
Pellaumail et al., 2002 [26] | Patella | Murine | Cadavera | 16 | Microscope Decalcified, paraffin fixated and stained with toluidine blue | B-mode, 50 MHz, measurement within minutes of sacrifice in 0.9% saline solution, 1600 m/s speed of sound | Difference ranging from 0.09 to 0.79 mm depending on age and cartilage degeneration |
Tarhan et al., 2003 [8] | Knee | Human | In vivob | 206 | MRI 0.23 Tesla, proton density and T2-weighted images, 4.5 mm slice thickness with 0.5 mm interslice gap | B-mode, 5–10 MHz, transcutaneous | Correlation coefficient r ranging from 0.38 to 0.74. No difference between non-arthritic (r ranging from 0.38 to 0.74) and arthritic knee joints (0.44 to 0.61). |
Mathiesen et al., 2004 [27] | Knee | Human | Cadaverb | 24 | Caliper gauge Thawed at room temperature | B-mode, 10 MHz, thawed at room temperature | Mean difference was 0.0 ± 0.4 mm. |
Yoon et al., 2008 [10] | Knee | Human | In vivob | 102 | MRI 1.5 Tesla, fat-suppressed spoiled gradient echo images, 1.5-mm slice thickness | B-mode, 12.5 MHz, transcutaneous | Correlation coefficient r for the minimal and maximal cartilage thickness in arthritic knees ranged from 0.568 to 0.844 for the medial epicondyle and from 0.314 to 0.507 for the lateral epicondyle |
Naredo et al., 2009 [24] | Knee | Human | Cadaverb | 24 | Stereoscopic magnifying loupe Fixed in formalin solution | B-mode, 14 MHz, thawed knees, transcutaneous | Agreement of the two methods was assessed using the ICC (intraclass correlation coefficient). The ICC was 0.719 for the medial epicondyle, 0.285 for the lateral epicondyle, and 0.267 for the intercondyle region. Mean difference ranged from 0.266 to 0.326 mm. |
Aula et al., 2010 [17] | Knee | Bovine | Cadavera | 10 | Contrast agent CT Slice thickness of 2.3 mm, pixel size of 0.2 × 0.2 mm2, thawed and stored in phosphate-buffered saline | A-mode, 30 MHz, thawed and stored in phosphate-buffered saline | Correlation coefficient r was 0.84. |
Spannow et al., 2011 [28] | Knee, ankle, wrist, finger | Human | In vivob | 143 | MRI 1.5 Tesla, fat-suppressed T1, slice thickness 1.4–1.6 mm, voxel size 0.5 × 0.8 × 1.6 mm | B-mode, 6-14 MHz, transcutaneous | Mean difference ranging from 0.08 to 0.48 mm. |
Ohashi et al., 2012 [5] | Knee | Porcine | In vitroa | 62 | Microscope Thawed cadaver | Conventional B-mode and real-time spatial compound B-mode, 10 MHz, 1488 and 1709 m/s speed of sound, thawed, stored in saline solution | Correlation coefficient r was 0.961 for conventional and 0.976 for real-time spatial sonography. The mean difference was 0.0073 ± 0.171 for conventional and 0.0139 ± 0.131 for real-time spatial sonography |
Ohashi et al., 2012 [6] | Knee | Human | In vivob | 505 | MRI 3 Tesla, double echo steady state sequence, slice thickness 0.29 mm, in plane resolution 0.31 mm × 0.31 mm | B-mode, 10 MHz, transcutaneous | Correlation coefficient r was 0.976 for non-arthritic and 0.964 for arthritic knees. The mean difference was 0.085 ± 0.405 mm for non-arthritic and 0.154 ± 0.413 mm for arthritic knees |
Mandl et al., 2015 [7] | Finger | Human | Cadaverb | 19 | Calibrated photos Frozen cadavers | B-mode, 7-15 MHz, 1696 m/s speed of sound | Correlation coefficient r was 0.509 (R2 = 0.2595) |
Pradsgaard et al., 2015 [11] | Knee | Human | In vivob | 69 | MRI 1.5 Tesla, gradient fat saturation | B-mode, 6–14 MHz, transcutaneous, 1696 m/s speed of sound | Correlation coefficient rho was 0.86 for the medial condyle, 0.71 for the lateral condyle, and 0.70 for the intercondylar region. The mean difference was 0.31 mm for the medial condyle, 0.23 mm for the lateral condyle, and 0.33 mm for the intercondylar region |
Current study | Hip, knee, and shoulder | Bovine and porcine | Cadavera | 50 | Contrast agent micro-CT Voxel size/slice thickness of 0.016 mm, stored in phosphate-buffered saline solution for max. 18 h, iodine contrast agent | A-mode, 20 MHz, 1580 m/s and 1696 m/s speed of sound, non-frozen, evaluated within few hours of sacrifice | A strong correlation with a correlation coefficient of r = 0.973 was found. The mean difference was 0.074 ± 0.061 mm (0.002–0.256 mm) |