A merger of advanced mathematical techniques with cutting-edge computer technology may give doctors a powerful new weapon in the battle against prostate cancer. Researchers have developed a computerized expert system that would help radiation oncologists optimize placement of radioactive "seeds" for prostate brachytherapy, a non-surgical treatment that has been growing in popularity. Beyond providing treatment more precisely tailored to each patient, the system targets escalated doses of radiation at tumor pockets, and accounts for changes that occur in the prostate volume during treatment. Utilization of the system has the potential to improve tumor control and reduce uncomfortable side effects.
For the cost-conscious medical industry, the automated system offers a dramatic reduction in the time required to design radioactive seed treatment, allowing optimized plans to be created in minutes -- and revised as the procedure proceeds.
"The system allows us to effectively manipulate the large number of variables involved, something that is far too complex for even the best human experts," said Eva K. Lee, assistant professor of industrial and systems engineering at the Georgia Institute of Technology and of radiation oncology at Emory University School of Medicine. "We can deliver better precision and create the optimal plan for each patient. This system should help cut the recurrence rate for prostate cancer and reduce toxicity to healthy tissue."
Lee will present details of her treatment planning system February 19 at the 167th annual meeting of the American Association for the Advancement of Science (AAAS) in San Francisco. The project will also be part of a news briefing at 1 p.m. PT on February 19.
Prostate brachytherapy involves implantation of tiny radioactive seeds in the cancerous prostate. Continuous radiation from the seeds kills the cancer cells, allowing patients to avoid surgery that can produce such complications as incontinence and impotence.
To successfully treat the cancer, however, physicians must carefully design the radiation dose, balancing the high radiation levels needed to eradicate the cancer against the need to protect nearby tissue, including the urethra and rectum. Further complicating treatment is the edema that occurs as needles are inserted to place the seeds. Resulting changes in prostate volume can mean delivering too little radiation at the beginning of treatment and too much as the swelling subsides.
"It is very complicated to produce a successful implant," explained Lee, who collaborated on the work with Dr. Macro Zaider, professor and head of brachytherapy physics at Memorial Sloan Kettering Cancer Center. "Proper coverage of the entire prostate is very important, but it can be very difficult to carry out the plan. The seeds cannot always be placed in the location you want, so you must be able to compensate for that. Our system allows real-time planning, and corrections can be made as you proceed."
Ultrasound images of the patient's prostate are used by the system to help determine optimal radioactive seed placement based on such variables as prostate volume, location of tumor pockets, radioactivity levels of the seeds, location of the urethra (which passes through the prostate) and regions of the organ that may be unreachable by placement needles. Mixed integer programming and computational optimization techniques are the core technological tools utilized.
Woven into the system are a dose-calculation engine, a modeling module, an optimization engine, and a graphical evaluation tool. Features include (1) real-time automatic design of seed configuration for use in the operating room, (2) incorporation of magnetic resonance spectroscopy imaging information to target escalated doses to tumor regions, (3) planning that accounts for changes in prostate volume, and (4) an ability to correct seed configuration intra-operatively to account for seed displacement, needle distortion, or unforeseen difficulties encountered during implantation.
"To the physician, this will be a black box," Lee explained. "They will not need to know what is going on with the mathematics. All they will have to do is tell the system what they want in the plan."
Less experienced oncologists working in remote areas could use the system to aid in producing high quality treatment plans.
In most current treatments, seed placement is determined manually based on a simulation of the patient's prostate. Done days or weeks ahead of the operation, this "pre-plan" takes hours to produce. By cutting the planning time to as little as 15 minutes, the system should reduce costs and allow physicians to spend more time with their patients, Lee noted.
Though the expert system is ready for commercialization, it will have to receive FDA approval before being made available to treatment centers. However, Lee has used real patient data to compare her system against treatment plans designed by radiation oncologists. Those results suggest the system will provide significant improvements in treatment outcomes.
The system operates on a wide range of computing platforms, including Windows NT personal computers.