EDUCACIÓN NO FORMAL

A two-stage registration process was developed to register the coordinate system of the robot to the micro-CT scanner (eXplore Ultra Locus, General Electric Healthcare Biosciences London, ON, Canada). The two-stage registration was developed to achieve a balance between a high quality registration and the time requirements of an end user to complete a pre-clinical intervention. Completion of the primary first step of the registration requires that a removable 6.35 mm borosilicate fiducial bead (McMaster- Carr, Cleveland, OH) be mounted onto the device at the RCM. In addition, an array of six more borosilicate bead secondary fiducials was mounted onto the robot below the animal bed (fiducial array visible in Figure 5.2). With the RCM fiducial bead attached, the robot was positioned at four different locations within the micro-CT bore, and a CT image was obtained each time. Using registration software developed in MATLAB (The Mathworks Inc., Natick, MA), the rigid body transformation between the two sets of coordinates was calculated by comparing the position of the RCM fiducial in robot coordinates to its position in each of the four images. This primary registration can be used alone to guide to the robot to targets in micro-CT images. However, if the robotic system is removed and then reattached to the micro-CT bed, this primary registration is no longer valid due to variability in robot reattachment.16 Unfortunately, repeating the primary registration each time the robot is reattached to the micro-CT to complete an intervention is time consuming and laborious. To avoid constant repetition of the primary registration a secondary registration was developed.

The secondary registration takes advantage of the six borosilicate secondary fiducials attached to the robot below the animal bed (Figure 5.2). During primary registration, these six fiducials are imaged along with the RCM fiducial bead. One of the scans acquired for the primary registration is of the robot at its home position. To complete the secondary registration, the reattached robot and six secondary fiducials are imaged with the robot at its home position. The registration software is then used to calculate the rigid body transformation using the six secondary fiducial positions in the primary registration home scan and the secondary fiducial positions in the secondary registration home scan. This secondary registration will account for and correct the variability introduced when the robot is reattached to the micro-CT scanner. A target in the micro-CT images can then be localized in robot coordinates by applying both registrations. Through the secondary registration, the end user is only required to acquire one image at the initiation of an intervention, rather then four. Furthermore, imaging of the fiducials for secondary registration can be simultaneously acquired while imaging the small animal, to localize targets. Since the small animal must always be imaged, the secondary registration does not increase the total number of scans required and allows for the primary registration to be reused across multiple interventions. Unfortunately, use of the secondary registration will also reduce overall targeting accuracy since two registration errors, from the primary and secondary, will be combined. The primary registration can be used by itself to improve targeting accuracy; however, this will be at the expense of increasing total registration time. Use of the combined registration reduces time for procedures at the expense of accuracy. The end-user must determine which registration process will best suit their application needs.

5.2.3.2

Registration Software

The fiducial bead mounted onto the robot RCM was segmented and its centroid was determined in each of the four images acquired for the primary registration. Segmentation was accomplished through a threshold based region-growing. The center of the segmented RCM fiducial was then calculated using a squared-intensity-weighted centroiding. The centroiding algorithm used was found in simulated images to have an error of < 5% of the image voxel size in localizing 3D centroids.23 The centroided RCM

fiducial positions were then rigidly registered to their homologous robot coordinates using a least squares algorithm.24 The results of the primary registration represented the translation and rotation required to align the coordinate system of the micro-CT scanner with the robotic system.

The secondary registration first required that during the primary registration, with the robot at its home position, the centroids of the six secondary fiducial beads are determined using the technique previously described. During the secondary registration, again with the robot at its home position, these six secondary fiducials are reimaged and their centroids determined. The two sets of centroids are then registered using a rigid body transformation. This secondary registration represents the transformation required to realign the robot axes at the time of the intervention with the robot axes at the time of the primary registration. The errors of the primary, secondary and combined registration were characterized by calculation of the fiducial registration error (FRE) and target registration error (TRE).25 The TRE of the primary registration was calculated by acquiring five additional images, not used in the registration, of the RCM fiducial at positions in the robot’s full range of motion. The secondary and combined TREs were calculated by attaching and reattaching the robot to the micro-CT scanner bed five times with the RCM fiducial still attached. In each image, five of the six secondary fiducials were used to calculate the secondary registration. The sixth secondary fiducial in each of the five images was used to calculate the secondary TRE. The combined TRE in these five images was calculated using the position of the RCM fiducial in the images following robot reattachment to the scanner bed.