A Technique For The In Vitro Registration Of CT, MR And Stereophotogrammetric Data

To analyze the complex interaction among bone, cartilage and other tissues during the motion of a joint without opening the capsule and thus altering its behavior, data coming from different sources, for instance from medical images and stereophotogrammetry, have to be registered [1]. The registration process often relays on the optimal alignment of a set of points, anatomical reperi or artificial landmarks, observed in all the data set. With this approach, the measurement error may considerably affect the registration accuracy, in particular in terms of orientation. For instance, the definition of the tibia reference system according to the ISB standard [2] requires the identification of the two malleoli: assuming the two points are 60 mm apart, a measurement error of 1 mm on each point may result in 1.9° of error in the orientation of the line through the malleoli for each dataset, leading to a maximum registration error of 3.8°. The purpose of this paper is to present an alternative approach based on axes rather than points. The axes are defined by drilling through holes in the bones, where cylindrical rods are inserted. The approach has several potential advantages. Axes can be easily identified and reconstructed both on medical images and on kinematic data. In fact, rods can be made of different materials that can be modified between measurement techniques to optimize the axis visibility and reconstruction. Rods can be removed and reinserted, also during the same experiment, without losing accuracy. Accuracy relies on the fitting of a very simple feature such as a line; moreover, accuracy does not depend on the rod axial position on the holes. This approach is here presented to register CT and MR to stereophotogrammetry data. The registration accuracy is evaluated for a wooden object with known geometry and for a bovine femur.