Diagnostic Performance of Multi-Detector Computed Tomography Arthrography and 3-Tesla Magnetic Resonance Imaging to Diagnose Experimentally Created Articular Cartilage Lesions in Equine Cadaver Stifles

Abstract

Background: The purpose of the study was to determine the diagnostic performance of computed tomographic arthrography (CTA) and 3-Tesla magnetic resonance imaging (MRI) for detecting artificial cartilage lesions in equine femorotibial and femoropatellar joints.

Methods: A total of 79 cartilage defects were created arthroscopically in 15 cadaver stifles from adult horses in eight different locations. In addition, 68 sites served as negative controls. MRI and CTA (80-160 mL iodinated contrast media at 87.5 mg/mL per joint) studies were obtained and evaluated by a radiologist unaware of the lesion distribution. The stifles were macroscopically evaluated, and lesion surface area, depth, and volume were determined. The sensitivity and specificity of MRI and CTA were calculated and compared between modalities.

Results: The sensitivity values of CTA (53%) and MRI (66%) were not significantly different (p = 0.09). However, the specificity of CTA (66%) was significantly greater compared to MRI (52%) (p = 0.04). The mean lesion surface area was 11 mm² (range: 2-54 mm²). Greater lesion surface area resulted in greater odds of lesion detection with CTA but not with MRI.

Conclusions: CTA achieved a similar diagnostic performance compared to high-field MRI in detecting small experimental cartilage lesions. Despite this, CTA showed a higher specificity than MRI, thus making CTA more accurate in diagnosing normal cartilage. Small lesion size was a discriminating factor for lesion detection. In a clinical setting, CTA may be preferred over MRI due to higher availability and easier image acquisition.

Keywords: arthroscopy; cartilage lesion; computed tomography; horse; magnetic resonance imaging; stifle joint.

Figure 1

Graphic illustration of a left stifle joint (A) cranial view, medial to the left with the patella removed; (B) mediolateral view. Red circles correspond to the lesion locations in the femoropatellar joint ((A) = trochlear ridges and intertrochlear groove of the femur, (B) = facies articularis patellae); blue circles show the lesion locations in the femorotibial joint ((A) = cranial femoral condyles, (B) = cranial and caudal femoral condyles). On the caudolateral femoral condyle, a comparable lesion location was chosen.

Figure 2

Distal to the top, lateral to the left. (A) Macroscopic image of a cartilage lesion on the medial trochlear ridge of the femur (white arrow) and a second lesion in the intertrochlear groove (arrowhead). (B) Corresponding arthroscopic view of the medial trochlear ridge lesion (white arrow).

Figure 3

Images showing a cartilage lesion (white arrow) on the lateral trochlear ridge of the femur. CTA images in bone reconstruction algorithm: sagittal plane (A), transverse plane (B), bottom row 3D T2W MRI images of the same lesion in sagittal (C) and transverse (D) plane. Subcutaneous gas and contrast accumulation is visible on both CTA images (black arrowhead contrast media, white arrowhead gas artifact).

Figure 4

Box plots of mean lesion surface area distribution of arthroscopically created cartilage lesions in equine stifle joints using computed tomographic arthrography (CTA) (A) and magnetic resonance imaging (MRI) (B). X-axis: 0 = lesion not detected, 1 = lesion detected. Y-axis: mean lesion surface area in mm². Asterisk (*) indicates a statistically significant difference between the mean lesion surface area of detected and not detected lesions.

Figure 5

ROC curve with optimal Youden index point using CTA (A) and MRI (B). The optimal cut-off for lesion detection with CTA was 3.9 mm², resulting in a sensitivity of 61% and a specificity of 70%. The optimal cut-off for lesion detection with MRI was 4.53 mm², resulting in a sensitivity of 55% and a specificity of 80%. TPR = true positive rate. FPR = false positive rate.