Computer based quantification of the mean Achilles tendon thickness in US-images — effect of tendinosis

Background and purpose

: B-mode measurement of the sagital diameter of the Achilles Tendon (AT) based on manual tracing (MT) procedures is partly dependent on the subjectivity of the reader. The aim of this study is to compare the tracing results of the AT boundaries with an automatic identification (AI) process, already established for the detection of the Intima-Media-Thickness (IMT), and computerassisted MT.

Methods and Results

: The detection of the AT boundaries was performed in 115 US images of the AT including the anterior boundary of the calcaneus bone. The measured section (starting point 4cm away from the anterior boundary of the calcaneus bone) amounted 3 cm and was divided in 3 sub segments (1 cm each). Intra- and inter-reader/ - observer variability for the mean and the maximum AT Thickness (ATT) with AI and MT were evaluated. A normal group and a group with manifest tendinosis were compared concerning mean ATT (mATT) and maximum ATT (maxATT).

Using MT the intra- and inter-reader variability amounted 3.0 % and 6.8 %, respectively, using the AT 1.6 % and 3.9 % mm. Mean and maximum ATT were measured systematically lower by AI compared with MT in all regions by 0,4 mm. The AI procedure was most suitable in the 2

nd

segment. The mATT and maxATT were correctly detected in 93.9 % and 96.6 % of the images.

Conclusion

: The AI procedure detects the ATT with a high precision in all three segments. The most robust measurement was reached in the 2

nd

segment. It eliminates most of the inter-/intra-reader variability of the ATT measurement using MT. We suggest using the new method for new gold-standard of the ultrasound analysis of the AT.