Detailed subject-specific FE rib modeling for fracture prediction
Artikel i vetenskaplig tidskrift, 2019
Objective: The current state of the art human body models (HBMs) underpredict the number of
fractured ribs. Also, it has not been shown that the models can predict the fracture locations.
Efforts have been made to create subject specific rib models for fracture prediction, with mixed
results. The aim of this study is to evaluate if subject-specific finite element (FE) rib models, based
on state-of-the-art clinical CT data combined with subject-specific material data, can predict rib
stiffness and fracture location in anterior-posterior rib bending.
Method: High resolution clinical CT data was used to generate detailed subject-specific geometry
for twelve FE models of the sixth rib. The cortical bone periosteal and endosteal surfaces were
estimated based on a previously calibrated cortical bone mapping algorithm. The cortical and the
trabecular bone were modeled using a hexa-block algorithm. The isotropic material model for the
cortical bone in each rib model was assigned subject-specific material data based on tension coupon tests.
Two different modeling strategies were used for the trabecular bone.
The capability of the FE model to predict fracture location was carried out by modeling physical
dynamic anterior-posterior rib bending tests. The rib model predictions were directly compared to
the results from the tests. The predicted force-displacement time history, strain measurements at
four locations, and rotation of the rib ends were compared to the results from the physical tests
by means of CORA analysis. Rib fracture location in the FE model was estimated as the position
for the element with the highest first principle strain at the time corresponding to rib fracture in
the physical test.
Results: Seven out of the twelve rib models predicted the fracture locations (at least for one of
the trabecular modeling strategies) and had a force-displacement CORA score above 0.65. The
other five rib models, had either a poor force-displacement CORA response or a poor fracture
location prediction. It was observed that the stress-strain response for the coupon test for these
five ribs showed significantly lower Young’s modulus, yield stress, and elongation at fracture com-
pared to the other seven ribs.
Conclusion: This study indicates that rib fracture location can be predicted for subject specific rib
models based on high resolution CT, when loaded in anterior-posterior bending, as long as the
rib’s cortical cortex is of sufficient thickness and has limited porosity. This study provides guide-
lines for further enhancements of rib modeling for fracture location prediction with HBMs.