Subject-specific planning of femoroplasty: an experimental verification study

Abstract

The risk of osteoporotic hip fractures may be reduced by augmenting susceptible femora with acrylic polymethylmethacrylate (PMMA) bone cement. Grossly filling the proximal femur with PMMA has shown promise, but the augmented bones can suffer from thermal necrosis or cement leakage, among other side effects. We hypothesized that, using subject-specific planning and computer-assisted augmentation, we can minimize cement volume while increasing bone strength and reducing the risk of fracture. We mechanically tested eight pairs of osteoporotic femora, after augmenting one from each pair following patient-specific planning reported earlier, which optimized cement distribution and strength increase. An average of 9.5(±1.7) ml of cement was injected in the augmented set. Augmentation significantly (P<0.05) increased the yield load by 33%, maximum load by 30%, yield energy by 118%, and maximum energy by 94% relative to the non-augmented controls. Also predicted yield loads correlated well (R(2)=0.74) with the experiments and, for augmented specimens, cement profiles were predicted with an average surface error of <2 mm, further validating our simulation techniques. Results of the current study suggest that subject-specific planning of femoroplasty reduces the risk of hip fracture while minimizing the amount of cement required.

Keywords: Cement augmentation; Femoroplasty; Mechanical test; Planning.

Figure 1. Schematic of an augmentation suggested by BESO simulations (green) and planned path and locations of injection (blue)

Figure 2

(A): Injection setup, (B): Initial registration using the transformation between the picked points (yellow dots) and pre-determined landmarks (numbered red dots), (C): ICP registration results using surface traced points (yellow).

Figure 3

Representative force-displacement curves for a pair of specimens.

Figure 4

Two-dimensional schematics describing quantification of intra-operative (offset) and simulation (shape) errors. A: initial overlay of simulated (red) and experimentally obtained (blue) cement isosurfaces, after CT volume registration. The two isosurfaces are essentially triangular surface meshes that enclose the cement volumes (triangles and vertices not shown for clarity). B: corrective rotation (Θ) and translation (d) identified by ICP registration that align post-operative surface onto the simulated one. In three dimensions, there are two additional rotations along X and Z axes, and the translation is a three-dimensional vector. These numbers represent injector placement error. C: after alignment, average surface distance is measured to quantify shape discrepancy, represented by the light grey area, between the two isosurfaces.

Figure 5

Effect of augmentation on yield and maximum load (left) and yield and maximum energy (right). Error bars represent standard deviations.

Figure 6

Measured vs. predicted yield loads. The overall correlation is y = 1.00 x + 147, R²=0.74.

Figure 7

Comparison of cement isosurfaces before (left) and after (right) ICP registration. Red indicates the planned cement distribution while blue indicates the cement cloud obtained from post-operative CT scan.