At present, orthopaedic surgeons utilize either CT, MRI or X-ray for imaging a joint. Unfortunately, CT and MRI are quite expensive, non weight-bearing and the orthopaedic surgeon does not receive revenue for these procedures. Although x-rays are cheaper, similar to CT scans, patients incur radiation. Also, all three of these imaging modalities are static. More recently, a new ultrasound technology has been developed that will allow a surgeon to image their patients in 3D. The objective of this study is to highlight the new opportunity for orthopaedic surgeons to use 3D ultrasound as alternative to CT, MRI and X-rays. The 3D reconstruction process utilizes statistical shape atlases in conjunction with the ultrasound RF data to build the patient anatomy in real-time. The ultrasound RF signals are acquired using a linear transducer. Raw RF data is then extracted across each scan line. The transducer is tracked using a 3D tracking system. The location and orientation for each scan line is calculated using the tracking data and known position of the tracker relative to the signal. For each scan line, a detection algorithm extracts the location on the signal of the bone boundary, if any exists. Throughout the scan process, a 3D point cloud is created for each detected bone signal. Using a statistical bone atlas for each anatomy, the patient specific surface is reconstruction by optimizing the geometry to match the point cloud. Missing regions are interpolated from the bone atlas. To validate reconstructed models output models are then compared to models generated from 3D imaging, including CT and MRI.Introduction
Methods
Fracture of the distal radius is one of the most common wrist fractures that orthopedic surgeons face. Quite often an injury is too severe to be repaired by supportive measures and pin or plate fixation is the subsequent alternative. In this study we present a novel method for automated 3D analysis of distal radius utilizing statistical atlases, this method can be used to design pin or plate fixation device that accurately fit the anatomy. A set of 120 bones (60 males and 60 females) were scanned using high resolution CT. These CT scans were then segmented and the surface models of the radius were added to the statistical atlas. Global shape differences between males and females were then identified using the statistical atlas. A set of landmarks were then calculated including the tip of the lateral styloid process and centroid of the distal plateau. These landmarks were then used to calculate the width of the distal plateau, the height of the distal plateau, overall radius length and the curvature of the distal plateau. These measurements were then compared for both males and females. Three of the measurements came statistically significant with p<
0.01. Curvature of the distal plateau wasn’t found to be significant, with females having slightly higher radius of curvature than males. This automated 3D analysis overcomes the major drawbacks of 2D x-ray measurements and manual localization methods. Thus, this analysis quantifies more accurately the anatomical differences between males and females. Statistically significant anatomical gender differences were found and quantified, which can be used for the design of trauma prosthesis that can fit normal anatomy.
The success of TKAs depends on the restoration of correct knee alignment and proper implant sizing and placement. The mechanical axis is considered a key factor in the restoration of knee alignment along with the transepicondylar axis and the posterior condylar axis as references for external and internal implant rotation. Accurate calculation of the distal resection plane in the femur and proximal resection plane in the tibia is crucial to determine the amount of the bone to be resected. In this study, we developed a model for mapping the thickness of the femoral and tibial articulating cartilage. We also studied the effect of cartilage presence and the absence on the accuracy of calculating the surgical landmarks, implant sizing and placement. Cartilage models were constructed using fat suppression MRI scans of healthy individuals with different body sizes. The femoral and tibial cartilages were segmented and surface models were generated. The inner and outer surfaces of the cartilage were separated, the inner surfaces were then mapped to the articulating surface of the femur and tibia to establish correspondence between the cortical bone surface and the inner surface of the cartilage. For each vertex on the normalized inner surface of the cartilage, the closest point was found on the outer surface of the cartilage and the normal distances were calculated. These distances were then averaged for each vertex across the population to calculate an average cartilage model. This average cartilage model was then used to grow a cartilage layer on our database of 300 bones from CT scans. Surgical landmarks and implant sizing and placement were then calculated for each bone before and after the cartilage and results were compared. Some of the landmarks including the mechanical and transepicondylar axes were found to be independent from the presence or absence of knee articulating cartilage, whereas the posterior condylar axis and tibial and femoral resection planes can be affected by the absence or presence of cartilage.
