Towards automatic computer-aided knee surgery by innovative methods for processing the femur surface model

The femoral shaft (FDA) and transepicondylar (TA), anterior-posterior (WL) and posterior condylar (PCL) axes are fundamental quantities in planning knee arthroplasty surgery. As an alternative to the TA, we introduce the anatomical flexion axis (AFA). Obtaining such axes from image data without any manual supervision remains a practical objective. We propose a novel method that automatically computes the axes of the distal femur by processing the femur mesh surface.

Surface data were processed by exploiting specific geometric, anatomical and functional properties. Robust ellipse fitting of the two-dimensional (2D) condylar profiles was utilized to determine the AFA alternative to the TA. The repeatability of the method was tested upon 20 femur surfaces reconstructed from CT scans taken on cadavers.

At the highest surface resolutions, the relative median error in the direction of the FDA, AFA, PCL, WL and TA was < 0.50 degrees, 1.20 degrees, 1.0 degrees, 1.30 degrees and 1.50 degrees, respectively. As expected, at the lowest surface resolution, the repeatability decreased to 1.20 degrees, 2.70 degrees, 3.30 degrees, 3.0 degrees and 4.70 degrees, respectively. The computed directions of the FDA, PCL, WL and TA were in agreement (0.60 degrees, 1.55 degrees, 1.90 degrees, 2.40 degrees) with the corresponding reference parameters manually identified in the original CT images by medical experts and with the literature.

The proposed method proved that: (a) the AFA can be robustly computed by a geometrical analysis of the posterior profiles of the two condyles and can be considered a useful alternative to the TA; (b) higher surface resolutions leads to higher repeatability of all computed quantities; (c) the TA is less repeatable than the other axes.