Welcome to the setup guide for "Robotic Catheter Ablation: An Evaluation and Prototyping Platform." In this guide, we will walk you through the steps of setting up an evaluation platform for catheter ablation. Following this guide will help you calibrate and register your camera system effectively and set up the physical platform. Let's start by gathering all the necessary requirements.
- 2x endoscopic cameras
- 1x 3D-printed STL-Files
- 1x Printed Apriltag
- 1x Printed Checkerboard
- 1x Printed Validation Marker
- 8x M3 bolts
- 6x M3 nuts
The two endoscopic cameras are placed in the main part of the physical setup as follows:
Figure 1: An illustration of the stereo camera setup showcasing the correct placement of endoscopic cameras.
Utilize M3 bolts to secure the cameras firmly in the model to prevent any movements during operation.
To stream the videos from the two cameras, we utilize the usb_cam package in ROS Noetic. The following launch file starts the two cameras and publishes the video streams to stereo_cam_right/image_raw and stereo_cam_left/image_raw:
<launch>
<!-- Arguments -->
<arg name="frame_rate" default="25" /> <!-- Frame rate of the video streams -->
<arg name="pixel_format" default="yuyv" /> <!-- Pixel format of the video streams -->
<!-- Launch right camera -->
<node name="stereo_cam_right_node" pkg="usb_cam" type="usb_cam_node">
<param name="camera_name" value="stereo_cam_right" />
<param name="camera_frame_id" value="/stereo_cam_right" />
<param name="video_device" value="/dev/<PATH_RIGHT_CAMERA>"/> <!-- Replace with the actual path to your right camera -->
<param name="pixel_format" value="$(arg pixel_format)" />
<param name="framerate" value="$(arg frame_rate)" />
</node>
<!-- Launch left camera -->
<node name="stereo_cam_left_node" pkg="usb_cam" type="usb_cam_node">
<param name="camera_name" value="stereo_cam_left" />
<param name="camera_frame_id" value="/stereo_cam_left" />
<param name="video_device" value="/dev/<PATH_LEFT_CAMERA>"/> <!-- Replace with the actual path to your left camera -->
<param name="pixel_format" value="$(arg pixel_format)" />
<param name="framerate" value="$(arg frame_rate)" />
</node>
</launch>
Please ensure to replace <PATH_RIGHT_CAMERA> and <PATH_LEFT_CAMERA> with the appropriate paths to your right and left cameras, respectively.
Calibrating the stereo camera involves determining the intrinsic and extrinsic parameters of the individual cameras. To accomplish this, we utilize Matlab's Stereo Camera Calibrator. The checkerboard is used to calibrate the cameras by placing it at different orientations and distances with respect to the cameras. Capture multiple pairs of synchronized images of the checkerboard using both the left and right cameras. Make sure to cover various angles and distances to get a comprehensive set of data for calibration:
Figure 2: An illustration of the stereo camera calibration using Matlab's Stereo Camera Calibrator.
To obtain the transformation from the stereo camera setup to the model, we utilize aprilags, a type of fiducial marker system. Follow the steps below to set up the registration:
- Print the Apriltag marker.
- Attach it to the registration unit as illustrated in Figure 3.
Figure 3: AprilTag attached to the registration unit.
- Attach the registration unit to the main part of the model to allow for the extraction of transformation data.
- Utilize the apriltag_ros package for detecting the apriltags in both the left and right camera streams. Below is the launch file that initiates the apriltag detection nodes:
<launch>
<!-- Arguments -->
<arg name="camera_name_left" default="/stereo_cam_left" />
<arg name="camera_name_right" default="/stereo_cam_right" />
<arg name="apriltag_definition_file" default="<PATH_TO_DEFINITION>" />
<arg name="apriltag_settings_file" default="<PATH_TO_SETTINGS>" />
<!-- Launch apriltag detection for the right camera -->
<group ns="$(arg camera_name_right)">
<node pkg="image_proc" type="image_proc" name="image_proc" />
<node pkg="apriltag_ros" type="apriltag_ros_continuous_node" name="apriltag_right_cam" clear_params="true" output="screen">
<rosparam command="load" file="$(arg apriltag_settings_file)" />
<rosparam command="load" file="$(arg apriltag_definition_file)" />
<param name="camera_frame" type="str" value="$(arg camera_name_right)" />
</node>
</group>
<!-- Launch apriltag detection for the left camera -->
<group ns="$(arg camera_name_left)">
<node pkg="image_proc" type="image_proc" name="image_proc" />
<node pkg="apriltag_ros" type="apriltag_ros_continuous_node" name="apriltag_left_cam" clear_params="true" output="screen">
<rosparam command="load" file="$(arg apriltag_settings_file)" />
<rosparam command="load" file="$(arg apriltag_definition_file)" />
<param name="camera_frame" type="str" value="$(arg camera_name_left)" />
</node>
</group>
</launch>To evaluate the accuracy of the stereo camera setup, we triangulate two color markers. We estimate the 3D coordinates of the markers and compare them to their known positions to determine the system's accuracy. The following figure illustrates how the color marker pattern is placed in the registration unit:
Figure 4: Evaluation color markers attached to the registration unit.
The known positions of the markers in the apriltag frame are:
- Green marker: [0,0,0]
- Red marker: [0.005,0,0]
In this section, we cover loading the heart model and setting up the static transformations. It is essential to correctly set up the GUI to visualize and interact with the heart model appropriately.
To load the heart model, define the color and the necessary visual parameters using the following file:
<robot name="visual">
<!-- Define color -->
<material name="red">
<color rgba="1 0.3 0.3 1"/>
</material>
<link name="pulmonary_vein_main_fixture">
<!-- Main fixture -->
<visual>
<geometry>
<mesh filename="<PATH_TO_MAIN_PART_STL>" />
</geometry>
<material name="red"/>
</visual>
<!-- Ablation section -->
<visual>
<geometry>
<mesh filename="<PATH_TO_ABLATION_SECTION_STL>" />
</geometry>
<material name="red"/>
</visual>
</link>
</robot>Next, establish the necessary launch parameters and transformations using the snippet below:
<launch>
<!-- Load model -->
<arg name="model" default="<PATH_TO_URDF>"/>
<arg name="gui" default="true"/>
<param name="pulmonary_vein" command="$(find xacro)/xacro $(arg model)" />
<!-- Transformation from the model to the world frame -->
<node pkg="tf" type="static_transform_publisher" name="pulmonary_vein_broadcaster" args="0 0 0 0 0 0 1 mns model 60" />
</launch>The following XML script helps in establishing the necessary static transformations from the model to the apriltag to the camera setup:
<launch>
<!-- These nodes publish the static transformations form the apriltag -> aprilttag Mount -> model. They are know from the geometry of the calibration unit and the AprilTag -->
<node pkg="tf" type="static_transform_publisher" name="model_to_apriltag_center_tf" args="0.0908 0.0844 0.08045 0 0 0 1 model tag_0_center 60" />
<node pkg="tf" type="static_transform_publisher" name="apriltag_center_to_apriltag_tf" args="0 0.007 0 0.5 -0.5 -0.5 0.5 tag_0_center tag_0 60" />
<!-- Broadcast the transformation from the apriltag to the camera frame-->
<node pkg="tf" type="static_transform_publisher" name="apriltag_to_camera_tf" args="<PLACEHOLDER> tag_0 stereo_cam_right 60" />
</launch>The model and the corresponding transformations can then be visualized in Rviz as illustrated below:
Figure 6: Heart model and relevant transformations visualized in Rviz.
In this section, we explain the placement of the catheter and the complete arrangement of the setup.
First, a sheath is positioned inside the model such that the opening is located in the left atrium. The following image illustrates the overall setup:
Figure 7: The complete setup, showcasing all the components properly assembled.
The catheter is advanced through the sheath such that its color-marked tip is located inside the left atrium in front of the cameras:
Figure 8: The catheter as positioned in the physical setup.
This setup allows for the tracking of the catheter tip inside the model as illustrated in the following GIF:
Figure 9: A GIF demonstrating the catheter tracking system in the model.
For more detailed instructions, the reader is referred to the following publication:
@inproceedings{Heemeyer2023,
doi = {10.1109/ismr57123.2023.10130271},
url = {https://doi.org/10.1109/ismr57123.2023.10130271},
year = {2023},
month = apr,
publisher = {{IEEE}},
author = {Florian Heemeyer and Christophe Chautems and Quentin Boehler and Jos{\'{e}} L. Merino and Bradley J. Nelson},
title = {An Evaluation Platform for Catheter Ablation Navigation},
booktitle = {2023 International Symposium on Medical Robotics ({ISMR})}
}