Summary of Research Projects

Contemporary transrectal ultrasound (TRUS)-guided biopsy cannot identify or target lesions. Improved biopsy targeting with MRI could potentially overcome the shortcomings of ultrasound for the diagnosis and local therapy for prostate cancer. My Contributions: Conducted MRI compatibility study with multiple sequences (T1, T2, and TFE) Implemented a program for the calculation of the signal to noise ratio (SNR) based on National Electrical Manufacturers Association (NEMA) for a fully actuated robot Designed the motor control circuit boards and wiring CAD drawing of the actuated robot for prostate intervention SNR results averaged for 10 slices with 3 squences My current role (Needle Artifact Study): The purpose of this study is to quantify needle placement accuracy by comparing needle artifacts with multiple sequences, frequency encoding directions, needle orientations and different needle materials (clinical titanium needle, stainless steel alloy needle, glass rod, carbon fiber stick) using 3T scanner (Philips Achieva at NIH) in MR-Guided interventions. Based on the pre-study experiment (the images above), I have implemented a 3D reconstruction and interpolation image processing program for the sagittal view to track the centers of the needle voids and to evaluate the true needle trajectories. Cropped T1 Axial images showing needle voids and artifacts of 16G titanium needle (1 st and 3 rd rows) and 16G glass rod (2 nd and 4 th row) Nathan Bongjoon Cho [email protected] 3 of 9 Project Title: FBG Force Sensing Tool of the Steady-Hand Robot for Retinal Surgery Period: Dec. 2008 Jul. 2009 Advisors: Prof. Iulian Iordachita, Prof. Russell H. Taylor Financial Supports: NSF ERC9731748, NIH R01-EB007969-A1 Abstract: Retinal surgery requires extremely precise and delicate manipulation. However, human hand tremor is often too big to make precise movements. Forces at the tissue-tool interaction are frequently below human perceptual thresholds. Visualization techniques are also inadequate. Retinal surgery requires extremely precise and delicate manipulation. However, human hand tremor is often too big to make precise movements. Forces at the tissue-tool interaction are frequently below human perceptual thresholds. Visualization techniques are also inadequate. Microsurgical Assistant Workstation known as the Eye Robot is a framework that combines cooperative robotic manipulator and 3D stereo display. The robotic microsurgical tools are integrated with micro optical sensors for realtime tissue interrogation to accurately measure tissue-tool contact forces. 3D Stereo Display with integrated sensing information is for optimal visualization. Force Sensing: Force sensing tool is based on Fiber Bragg Grating (FBG) which strain sensors that are integrated into the shaft of the instrument. It can measure as small as 0.25 mN force resolution at the tip of the tool and is insensitive to interference. And it can also be mass-produced at low cost while maintaining high accuracy. My Contributions: Found a solution for the lag during realtime performance Optimized the software written in MATLAB Ported the current MATLAB program to C++ Calibrated 2D force sensing tool based on 3 channels of FBG Studied the temperature effect on each FBG using constant temperature oven and optical sensing interrogator (sm130, Micron Optics, Inc.) Microsurgical Assistant Workstation (The Eye Robot) Optical Force Sensing Concepts Nathan Bongjoon Cho [email protected] 4 of 9 Project Title: Image Overlay for MRI-Guided Needle Insertions Period: Sep. 2008 present Advisor: Prof. Iulian Iordachita Financial Support: NIH R01-CA118371 Abstract: MR arthrography (MRAr) is the imaging gold standard to assess small ligament and fibrocartilage injury in joints. In practice, MRAr consists of two consecutive sessions: 1. An interventional session where a needle is driven to the joint space and MR contrast is injected under fluoroscopy or CT guidance. 2. A diagnostic MRI imaging session to visualize the distribution of contrast inside the joint space and evaluate the condition of the joint. MR arthrography (MRAr) is the imaging gold standard to assess small ligament and fibrocartilage injury in joints. In practice, MRAr consists of two consecutive sessions: 1. An interventional session where a needle is driven to the joint space and MR contrast is injected under fluoroscopy or CT guidance. 2. A diagnostic MRI imaging session to visualize the distribution of contrast inside the joint space and evaluate the condition of the joint. Our approach to MRAr is to eliminate the separate radiologically guided needle insertion and contrast injection procedure by performing those tasks on conventional high-field closed MRI scanners. Current Overlook: Our project team is scheduled to do ten cadaver experiments along with phantom experiments in between. By the end of this year, the first set of accuracy test will be completed using our current system. In the beginning of next year, a new system will be designed to overcome any drawback of the current system. My Contributions and Roles: Coordinating clinical workflows during the experiment Modifying planning software according to the needs of the clinician Developing a new planning software module for 3D Slicer (http://www.slicer.org) Calculated the SNR for the new system System concept of 2D image overlay device Close-up view of the MR image overlay system Nathan Bongjoon Cho [email protected] 5 of 9 Project Title: MRI-Guided Robot for Transperineal Prostate Intervention Period: Sep. 2008 present Advisors: Sam Song, PhD, Prof. Iulian Iordachita Financial Support: NIH R01-CA111288-01 Abstract: To date, there has been no system that maintains the workflow of traditional interventional procedures in a compact, reliable and convenient platform while enabling the use of intraoperative MRI. The overriding goal of this work is to make high-field MRI available and practical for interventional procedures. To date, there has been no system that maintains the workflow of traditional interventional procedures in a compact, reliable and convenient platform while enabling the use of intraoperative MRI. The overriding goal of this work is to make high-field MRI available and practical for interventional procedures. Our robotic system provides a method for performing image-guided interventions using real-time MR images and pneumatic actuators. Pneumatic actuators have great potential for MR-compatible mechatronic systems. Since no electronics are required, they are fundamentally compatible with the MR environment. Therefore, pneumatic actuation turns out to be an ideal solution for this robotic system, and it allows the robot to meet all of the design requirements. Current Overlook: The role of our lab at Johns Hopkins in this project is to design the hardware including the robot controller and its software. Through integrating workshops for evaluation at Brigham and Women's Hospital (BWH), this system is leaping onto the next level. Currently the new hardware is designed to overcome drawbacks including restricted 2-DOF movement and much friction on the vertical axis. My Contributions and Roles: Implementing the robot control software including servo control written in C++ Integrated the robot with the navigation software using Open IGT Link which is a simple network protocol intended for trackers, robots and other devices to send data to the main application Tested the accuracy and repeatability of the robot using Optotrak (Northern Digital Inc.) Found the function of gravity compensation for the vertical axis Calculated the SNR for the new robot Robotic needle placement mechanism Connection and data flow among the components Nathan Bongjoon Cho [email protected] 6 of 9 Project Title: Multi-Axis Force Torque Sensor Calibration using Shape-from-Motion Method Period: Jul. 2008 Sep. 2008 Advisor: Prof. Peter Kazanzides Financial Support: NSF CISST / ERC Abstract: The least-square method is a traditional way of obtaining force torque calibration matrix. However, it requires explicit knowledge of all the applied load vectors. This technique called Shape-from-Motion allows us to eliminate the need to collect all the applied loads by collecting large amount of calibration data almost without any effort. This also brings out the result much faster, more accurate calibration than the least-square method. The least-square method is a traditional way of obtaining force torque calibration matrix. However, it requires explicit knowledge of all the applied load vectors. This technique called Shape-from-Motion allows us to eliminate the need to collect all the applied loads by collecting large amount of calibration data almost without any effort. This also brings out the result much faster, more accurate calibration than the least-square method. Project Goal: My project goal was to find an alternative method for calibrating force sensor. In other words, I needed to find a 6 x 6 decoupling matrix to convert 6 strain gauge readings into 3 forces and 3 torques (ATI sensor), and recalibrate a damaged (uncalibrated) sensor. Our further goal was to find ways to determine sensor bias which consists of a constant vector of offsets that is added to each and every sample of strain gauge readings. My Contributions: I implemented the force sensor calibration algorithm using MATLAB for the case of 6 DOF (3 forces and 3 torques). Furthermore, I implemented an algorithm to find a bias if there exists any. For testing, the authors used special calibration fixture to get quick, accurate measurements where I tested using simulated data instead. That is, I used a known calibration matrix and added noise to each and every data set which is normal random numbers with = 0 and = 0.001. I used μ σ frobenius normal form for error measurement. The Shape-from-Motion approach yields more precise and accurate results with much less effort than the least-square technique. References: [1] Voyles, R.M., J.D. Morrow, and P.K. Khosla, "Shape from Motion Approach to Rapid and Precise Force/Torque Sensor Calibration,'' ASME International Mechanical Engineering Congress and Exposition, San Francisco, CA, November, 1995. [2] R.Voyles, J. Morrow, and P.Khosla, "Including Sensor Bias in Shape from Motion Calibration and Multisensor Fusion," IEEE Multisensor Fusion and Integration Conference, December, 1996. Nathan Bongjoon Cho [email protected] 7 of 9 Project Title: Intraoperative Control for Ultrasound-Guided Prostate Brachytherapy Period: Feb. 2008 Jul. 2008 Mentor: Prof. Peter Kazanzides Advisor: Prof. Russell H. Taylor Clinician: Danny Y. Song, MD Financial Supports: NIH 5R44 CA088139, NIH 1R41 CA106152 Abstract: This project for robotic assistance was developed to solve current problems in prostate brachytherapy. It has passed Phase I trials, but limitations prevent from fully realizing the benefits: The existing robot control interface must be operated by a medical physicist using a laptop outside the sterile field, under the direction of a clinician. The 4-DOF robot is capable of slanted needle insertion, but clinical application is precluded by a lack of control for needle angulation. The clinician must be careful to prevent the needle guide from colliding with the ultrasound probe. This project for robotic assistance was developed to solve current problems in prostate brachytherapy. It has passed Phase I trials, but limitations prevent from fully realizing the benefits: The existing robot control interface must be operated by a medical physicist using a laptop outside the sterile field, under the direction of a clinician. The 4-DOF robot is capable of slanted needle insertion, but clinical application is precluded by a lack of control for needle angulation. The clinician must be careful to prevent the needle guide from colliding with the ultrasound probe. My Contributions: I made an improved interface for clinicians by providing an alternative to the existing robot control interface and greatly reducing the complexity of the workflow. This allows the clinician to control the robot directly which then significantly reduces procedure times, also reducing cost and risk of adverse effects from edema. Direct control of the clinician also prevents errors from relaying information between the clinician and the medical physicist. I also enabled the needle angulation to avoid pubic arch interference, making brachytherapy available to a wider segment of the population. This also improves safety because collisions between the needle guide and ultrasound probe are no longer possible. New GUI for robot control New software architecture for the robot control Nathan Bongjoon Cho [email protected] 8 of 9 Outcome: The new interface can be directly operated by the clinician inside the sterile field and is fully integrated with the existing robot control system. A new, menu-driven control interface was developed using a multi-modal input device and a robot-mounted display. Needle angulation capabilities and rotation using seed position as a remote center of motion (RCM) were added to the control software. Full collision avoidance was implemented, with the ultrasound probe registered in the robot's coordinate system. Related Tools: Input Device: MaxFire Pandora Pro USB game controller Display: CrystalFontz CFA634-YMC-KU smart LCD Software Component ⦁ SDL (Simple DirectMedia Layer, http://www.libsdl.org) for input device ⦁ pySerial for accessing to the serial port for the display ⦁ MPy3 (an embedded media player interface in Python) for a set of driver classes for display Programming Languages: C, Visual C++, Python Operating Systems: Windows, Unix, Mac OS X Input Device and LCD Display and input device inoperation Nathan Bongjoon Cho [email protected] 9 of 9 Project Title: Versatile MP3 Player using Secure Digital (SD) Card Period: Mar. 2005 Dec. 2005 Advisor: Prof. Hoon Chang Undergraduate Graduation Project Project Goal: To make an MP3 player using SD card having basic features such as playing, fast forwarding, rewinding, volume controlling and displaying on the LCD. Project Overlook: I used ATMega128 (by Atmel) as a microcontroller and VS1003 for decoding and DAC. On the LCD, I wanted to display file names and total number of songs in the SD card and the remaining battery life. For compiling, I used AVR gcc in the AVR Edit environment. PonyProg2000 was used as the main downloading program. The following pictures show how the parts are put together. Whole view Switches Back side showing circuits Microcontroller LCD SD card slot Project Outcome: Since I focused more on its functional side, it was rather bulky and not very aesthetically pleasing. However, it was overall a success.