Tuning and validation of a virtual mechanical testing pipeline for condylar stress fracture risk assessment in Thoroughbred racehorses

Abstract

Condylar stress fracture of the third metacarpal/metatarsal bone (MC3/MT3) in Thoroughbred racehorses is a common catastrophic injury, putting both the racehorse and the jockey at risk. Microdamage forms in the distal MC3 and may result in strain elevation and higher risk of stress fracture, directly and through focal osteolysis resulting from the associated damage repair by remodelling in the condylar parasagittal grooves (PSG). Standing computed tomography (sCT) is a practical screening tool for detecting fatigue-induced structural changes, but clinical interpretation remains subjective. The goal of this study was to develop and validate an sCT-based, subject-specific finite element analysis pipeline for virtual mechanical testing of the distal MC3. Twelve (n = 12) MC3 condyles with available ex vivo strain were selected. Half of the specimens were used for tuning Young's modulus in fatigue sites, and the other half were used to evaluate the validity of the pipeline in predicting PSG strain. The tuned model improved the prediction of strain and predicted higher strain in bones with PSG lysis over the untuned model, which is an essential feature for a diagnostic screening tool. The presented pipeline can assist clinicians with the interpretation of sCT-detectable structural changes and, ultimately, with the assessment of condylar fracture risk in racing Thoroughbreds.

Keywords: Thoroughbred racehorses; condylar stress fracture; finite element analysis; functional adaptation; third metacarpal bone.

Figure 1.

Segmentation of the sclerotic and lytic regions. Adaptive and damage factors were used for tuning Young’s modulus in sclerotic and lytic regions, indicated with orange (wide) and red (thin) arrows, respectively. Fatigue cracks are visible on the palmar aspect of the parasagittal grooves of the condyle PSG-SBI-1 after dissection and removal of soft tissue (A). Sclerotic and lytic regions of PSG-SBI-1 were segmented from the standing CT image (B,C). A representation of the measurement of the volume of a condyle with its sclerotic region is shown (D).

Figure 2.

Finite element (FE) modelling pipeline. (A) Standing CT (sCT) imaging was used for segmentation of the distal MC3 volume. (B) Surface strain data from previous ex vivo testing data were used as the reference data for comparison with the FE results. (C) Element-wise heterogeneous HA density was captured from the sCT image set. (D) Young’s modulus was assigned to individual elements, and the ex vivo loading and boundary conditions were replicated in the model. (E) The surface for which ex vivo mechanical testing data were available (light grey) and the associated PSG region (dark grey) were registered on the three-dimensional segmented bone end for comparison of the surface strain.

Figure 3.

Adaptive factor A_SCL showed a positive relationship with the normalized subchondral sclerotic volume V_SCL. This relationship was found by linear regression.

Figure 4.

Comparison of the FE-predicted and the ex vivo-measured (DIC) mean PSG strain. (A) Tuned model on the validation group. (B) Non-tuned model on the validation group. (C) Non-tuned model on all specimens. Triangular (red) and circular (black) datapoints represent the PSG-SBI and CTRL condyles, respectively. The non-tuned pipeline leans towards underestimating PSG strain (slope < 1), especially in bones with PSG-SBI, and explains very little variance in the experimental data (small R²). These limitations were resolved by the tuned modelling pipeline, which predicts PSG strain with about 4% error on average.

Figure 5.

Ex vivo-measured and FE-predicted surface strain for the condyles CTRL-6 and PSG-SBI-3. Ex vivo strain data were not available for the dorsal aspect of the condyles and the area under the indenter. Red arrows show the location of the strain concentration in the PSG in the vicinity of the lysis that matches between the ex vivo and the FE-predicted results.