REGISTRATION TECHNIQUES – IMAGE BASED AND IMAGE FREE – AN OVERVIEW

Abstract

Measurements of a patient’s anatomy are often made in two different forms, for instance from a computer tomography (CT) scan and by direct measurement of the anatomy, or when comparing a CT and a magnetic resonance imaging (MRI) scan or at different times. Therefore, it is almost inevitable that the patient will be measured in a different position each time, since the relative position between the patient and the measuring or scanning device will be different. To align the patient’s anatomy between these different measurement systems a process of registration is used. This is necessary in a number of fields including computer assisted navigation, robotic assisted surgery and diagnostics.

Computer assisted surgery (CAS) generally involves “patient to modality” registration, as, in any CAS application that involves planning, the relationship between the modeled space (where the procedure is planned) and the patient’s workspace (where the procedure is executed) needs to be established. Patient to modality registration involves the registration of patient-specific anatomy with an image acquired using one of many modalities. It is usually associated with intra-operative registration, where the actual patient’s position needs to be known with respect to a pre-operative or previously acquired image. Even though the acquisition of patient-specific information may itself involve the use of a modality, the purpose of the process is to register the patient’s position against the model. The two co-ordinate systems to be registered belong to the patient and to the modality used to acquire the registration image, respectively.

In “image-based” methods, identifiable features, such as fiducial marker screws or anatomical landmarks, are first extracted from the model, which is generally reconstructed from CT images, and then “sensed,” or located, in the operating theatre. This process provides the system with enough positional information for the model’s and patient’s spaces to be registered against a common co-ordinate system.

In recent years, the CAS community has seen a shift to “image-free” methods, where both the plan and registration process are carried out without any prior knowledge of the patient’s anatomy. The pre-operative image acquisition stage is avoided altogether, and the planning is executed intra-operatively during surgery. A complete functional model of the patient is reconstructed from anatomical landmarks sensed intra-operatively and, in some instances; intra-operatively acquired surface information is used to “morph” a standard anatomical atlas to resemble that of the patient.

Image-free methods offer the prospect of no pre-operative imaging or planning, however their value, in terms of intra-operative workflow and accuracy of outcome, has yet to be assessed when compared to image-based methods.