DOS 2017

DOS Kongressen 2017 ·

PRECISION OF BONE MODELS IN DYNAMIC RSA OF THE

ELBOW AND DISTAL FOREARM

Sepp De Raedt, Janni Thillemann, Chalotte Vestergaard Hemmingsen, Maiken

Stilling

esearch and Development, Nordic Roentgen Technique; Department of Clinical

Medicine, Aarhus University; Orthopedic Research Unit, Aarhus University Hos-

pital; Department of Clinical Medicine, Aarhus University

Background:

Dynamic radiostereometric (dRSA) analysis can be used to quan-

titatively measure changes in the three-dimensional in-vivo movements of

bones, which can provide insight in the kinematics. However, analysis of the ra-

dius and ulna bones in both the elbow and forearm is challenging due to the long

cylindrical shape. Using digitally reconstructed radiograph (DRR) based RSA, the

position and orientation of the bones can be determined without markers.

Purpose / Aim of Study:

To validate the precision of DRR based RSA com-

pared to marker-based analysis.

Materials and Methods:

Custom motorized fixtures to perform clinically rel-

evant wrist (11 arms) and elbow (8 arms) motions were made. Subject spe-

cific bone models were created from CT and tantalum markers were inserted.

Model- based RSA (RSAcore) was used to calibrate the first frame and initialize

the bones. The complete recording was subsequently automatically analyzed

by custom developed AutoRSA software. Marker analysis was independently

performed in 3 images per dynamic recording. Precision was evaluated as sys-

tematic bias (mean difference) and random error (1.96*SD) for translations and

rotations.

Findings / Results:

In the elbow the mean systematic bias for translations

(mm) was <0.17 for the proximal radius (p<0.05), <0.15 for the ulna (p<0.05).

and <0.02 for the humerus (p>0.21). No systematic bias was found for rota-

tions. Precision was ≤0.55 mm and ≤1°. In the distal forearm the mean sys-

tematic bias (mm/°) was <0.06 for the ulna (p>0.17) and <0.03 for the radius

(p>0.26). Precision was ≤0.18mm and ≤1°.

Conclusions:

DRR based RSA analysis using bone models provides a good pre-

cision for investigation of kinematics in the elbow and forearm. The method can

be used for automated analysis of markerless dRSA studies for both pre-oper-

ative diagnostics and to evaluate kinematics after ligament or implant surgery.

No conflicts of interest reported

25.