Spy Cameras and Robotics Track Patient Movements During RT
Spy Cameras and Robotics Track Patient Movements During RT
BOSTONFor the cancer patient who can’t hold a breath or stay still during radiation therapy, a team of medical experts and engineers is working on a tracking solution worthy of James Bondspy cameras and robotic vision.
Andre Kalend, PhD, of the University of Pittsburgh Medical Center, described the Robotics Artificial Vision (RAV) system at the 42nd annual meeting of the American Society for Therapeutic Radiology and Oncology (ASTRO).
RAV "anticipates" slight movements, such as tremors, breathing, coughing, and sighing. Then, by moving the treatment table, it realigns the patient for more precise delivery of radiation beams or alerts the operator in time to halt therapy until the movement stops.
"The accelerator itself can ‘see’ the patient, recognize the patient’s skin anatomy, and treat only when the robotic eye sees that the patient is in alignment with the isocenter of the beams," Dr. Kalend told ONI (see Figures 1 and 2). The full RAV-Venturi system can distinguish normal breathing from movements triggered by normal physiologic urges.
A radiotherapy medical physicist, Dr. Kalend developed the RAV system prototype with Joel Greenberger, MD, chairman of the Department of Radiotherapy at the University of Pittsburgh Medical Center, and Takeo Kanade, PhD, of the Carnegie Mellon University Robotics Institute, also in Pittsburgh.
The group began work 5 years ago on ways to make radiotherapy more dynamic and more accurate with respect to the tight target margins required for 3D conformal radiation therapy and intensity modulated radiation therapy (IMRT).
Cancer patients are often weak, Dr. Kalend said, and cannot tolerate today’s invasive respiration-gated treatments. Tracking breathing is important, however, because the margins of error for precision beams are so small that coughing, sniffing, sneezing, or burping can cause misalignment.
A Wobbly Dart Board
"Radiation beams and fields are tailored to the irregular shape of the mobile tumor with margins no longer of single centimetersnow it’s a few millimeters," Dr. Kalend said. With this small margin of error, he said, "the patient, the target, is like a wobbling dart board."
He summarized his thinking behind RAV as, "If I can’t stop [movement], I must be able to track it. If it’s normal, I treat; if it’s abnormal, I stop." To do that, he equipped the accelerator with "vision" or "eyes" to see patient movement and "ears" to hear the patient breathe.
The System’s Eyes
RAV has eight independently operating cameras that serve as eyes. They are mounted on the walls of the treatment room and the accelerator. The cameras use patented technology called CCD-based Robotics Vision Tracking to lock alignment lasers from the accelerator onto the patient’s natural skin features or standard radiotherapy tattoos (see Figure 2).
Dr. Kalend stressed that CCD tracks patient anatomy rather than the treatment table (traditional mechanical fixtures). He calls this process "virtual fixation" as opposed to traditional "mechanical fixation."
Like a cruise missile, he said, the system finds its targets via image, or shape, matching, rather than alignment via x,y,z coordinates. The RAV system homes in on the target (the patient’s skin rather than a missile target) and only "fires" when the image matches that of the radiation therapy treatment plan. To get that match, the automated treatment table responds like a joystick to video commands to move and rotate until the alignment is correct.
The System’s Ears
The Venturi "ears" of the system listen for the patient’s laryngeal Venturi waveforms, made by the jet streams of air passing through the lower orifice of the larynx, to distinguish normal breathing from abnormal movements.
Tumors in the lung and abdomen are mobile and deform with breathing; consequently, they are almost impossible to target with precision conformal radiation therapy or IMRT beams. Dr. Kalend and a third-year University of Pittsburgh medical student, Kenneth Clark, last year invented the Venturi respiration tracking system.
It is sort of an FM stethoscope that tunes in and listens to the patient breathing, and gates the radiation machine to fine beams that are on only during very specific breathing cycles or phases of the patient breathing freely under treatment, Dr. Kalend said.
This summer, Dr. Karen Shimoga from Carnegie Mellon University joined Dr. Kalend, and together they are perfecting the Venturi’s FM wave tuning decision logic so that it operates faster than a single accelerator beam pulse.
The first treatment session is basically a learning session, during which RAV learns the normal sequence of movements and skin characteristics of the patient.
The system studies how a person coughs and learns the "signature of coughing," he said, as it affects the position of the tumor.
"We know that no matter how you constrain the patient from breathing, natural respiration urgeslike swallowing, coughing, sniffing, and even just lifting a leg or taking a deep inspiration differentlyaffect the position of the diaphragm and therefore throw out your beam port position," he explained.
Dr. Kalend set four requirements when his group began developing the RAV-Venturi technology, he said. The system had to be noninvasive, competitive economically, accommodate patients who could not hold their breath for 12 seconds, and operate in real time.
In benchmark studies, RAV is reported to have achieved an alignment accuracy of 2 mm (limited by the width of the laser) and 0.95 tracking correlation with Venturi interrupts measured in milliseconds, Dr. Kalend said.
So far, the system has been tested on volunteers, and the group has applied for NIH funding for a clinical trial with Internal Review Board approval to start testing the system in patients during actual radiation treatments.