Ultrasound examination had an overall detection rate of 48% for the presence of lesions without significant differences between surgeons (A 173/360 and B 172/360) (Figure 8). This is not significantly different to the probability of detecting the lesion correctly by chance alone. The severity of the accurately identified lesions was correctly identified by the surgeons in 81-90% (surgeon A identified 81%, and surgeon B identified 90 %) of the scans.
Surgeon A and Surgeon B agreement with each other on if a lesion was present or not was fair (k 0.2856). Surgeon A had fair agreement with gross examination on the presence of the lesion on ultrasound (A=k 0.0461). Surgeon B had poor agreement with gross examination on the presence of the lesion on ultrasound (B=-0.1161). Surgeon A
had slight agreement with gross examination on the score of the lesion on ultrasound (k =0.1813). Surgeon B had slight agreement with gross examination on the score of the lesion on ultrasound (k= 0.0393).
Table 8 - Ultrasound findings compared with gross lesions
Surgeon A
Gross lesion y/n 0 1 Total
0 87 146 232
1 41 87 128
Total 127 233 360 Surgeon B
Gross lesion y/n 0 1 Total
0 132 100 232
1 88 40 128
Total 220 140 360
Surgeon A sensitivity = 87/232 = 0.375 and specificity = 87/128= 0.68 Surgeon B sensitivity = 132/232= 0.569 and specificity = 40/128 = 0.31
Therefore, Surgeon A was less likely to correctly detect that a tendon did not have a lesion than he was to detect a lesion was present. Surgeon B was more likely to correctly determine that a lesion was not present than to correctly detect a lesion.
More severe lesions were more consistently identified correctly (Appendices 16 and 17). These data confirm the clinical knowledge of the relative insensitivity of ultrasonography alone for the detection of subtle lesions.
Ultrasound examination has been the current main clinical tool for imaging tendon injuries since it was first described in 1986 (Yoo et al., 2012). Computer assisted
tomography and magnetic resonance imaging are becoming more popular but are not mobile or cost effective for general clinical practice. Clinical examination of tendon injuries involves ultrasound examination and determination of size and echogenicity changes within the tendon (Reef, 2001, Marr et al., 1993b, Marr, 1992). Acute tendon injury results in haemorrhage and serum exudation into the injured area and therefore swelling of the tendon. This presents on the ultrasound scan as an increase in tendon diameter and reduced echogenicity of the injured area. It is well recognised that ultrasound examination of tendons is a relatively insensitive tool (van Schie et al., 2009), particularly in mild lesions. In the clinical situation, these surgeons would have been expected to have a higher lesion detection rate, as they would use history, clinical examination, lameness investigation and ultrasound examination together to come to a diagnosis, particularly in subtle cases. In a clinical ultrasound examination, the surgeon can view the skin and subcutaneous tissues as well as the tendons and this can assist with lesion detection. In this study as the presence of subcutaneous haemorrhage from the skin incisions, and the variation in degree of this haemorrhage between horses, was thought likely to bias the examination of the images. The surgeons were therefore only provided with the section of the image that showed the superficial digital flexor tendon. Video ultrasound images of the whole length of the tendon would have enabled construction of a 3 dimensional sonogram of the tendon (Wood et al., 1994). A 3- dimensional image may have improved the detection rate of the lesions but this has not been fully validated. Ultrasonographic tissue characterisation has also been recommended for lesion monitoring (Bosch et al., 2011) although this method cannot be used for determining prognosis and imaged lesions were not compared with gross ex vivo and histological findings. On the video sonograms it would also have been harder
to correlate the sites of lesions on ultrasound with the lesions on the skin and therefore with histopathology.
4.4.4
S
AMPLEC
OLLECTION TECHNIQUESTendon is very fibrous and hard to cut without tearing. Collection and processing of both thoracic-limb superficial digital flexor tendons takes 4-6 hours. In all limbs there was significant discolouration and haemorrhage apparent when the skin was removed (Figure 20a). Once the subcutaneous tissues were dissected the sites of stab incisions into the tendon could be visualised and although fibrin was adherent to them no subcutaneous tissues were adherent (Figure 20b). Once the superficial and deep digital flexor tendons were separated, it was practically easier to divide the tendon in a coronal plane using the cutting board and a Feather® trimming blade (No.130 type S) (Figure 20c). Most but not all lesions were exactly in the centre of the tendon (Figure 20d and e) therefore it might have been better if the tendon had been divided into smaller lengths before splitting.
Figure 20 - Gross appearance of tendons
a) Thoracic limb with skin reflected to expose the tendons
b) Thoracic limb with subcutaneous tissue removed and stab incisions into tendon visible c) Cutting board used to hold tendon prior to coronal sectioning
d) End on view of tendon section to show offset lesion
e) Lengthwise view of 4cm section of tendon following coronal sectioning showing lesion offset from centre f) Tendon with discolouration proximal and distal to stylet insertion and fibre disruption proximal to stylet
insertion. Proximal on the horse is towards the top of the page and section is 4cm long.