Intelligent Database Helps Radiation Therapists Efficiently Develop Treatments

Using radiation therapy to attack cancerous tumors is a revolutionary treatment method that has saved lives and allowed many terminal patients to live longer. Radiation therapies attempt to kill cancerous cells while minimizing the harm to healthy tissue.

Harold Cline, a medical physicist at Lutheran Medical Center in Wheatridge, Colorado and his team have taken advantage of the data handling capabilities of IDL to develop an "intelligent" database for optimizing radiation treatment plans.

Database Saves Time

The database catalogues and retrieves radiation therapy treatment plans, giving physicians, planners and researchers a wealth of instantly available patient information. The database also saves personnel from the redundancy of creating treatment routines from scratch for each patient. Cline uses IDL to implement an artificial Neural Network Simulation, by combining components of previous treatment plans. Optimal starting points for new treatment plans are derived from the archived combinations.

Group Effort Promotes Faster Progress

The radiation treatment research is a cooperative effort between Lutheran Medical Center and the University of Colorado's Health Sciences Center (UCHSC) in Denver. Cline, along with Francis Newman of UCHSC, used IDL's array-oriented data handling architecture to develop the algorithms in the Neural Network Simulation. Using IDL, Gayle Harnisch, a graduate student at UCHSC, and Cline implemented the scheme as an interactive environment. The UCHSC group has employed a similar artificial neural network simulation to detect lesions otherwise difficult to identify in computed tomography (CT) and magnetic resonance (MR) images. This simulation facilitates screening and diagnosis of cancerous tumors.

Better Insight Through Experience

The database stores, retrieves and combines information in a manner that promotes greater insight. The system "trains" the database to associate a patient's CT scan and tumor location with a set of radiation beams to effectively treat the tumor. Because of the distributed manner in which information is stored, the retrieval process provides an optimal mixture of information based on the system's previous experience. In a way, it "studies" the archived plans and "learns" from them.

Easy Analysis With the IDL Interface

With IDL widgets, Cline and Harnisch developed a custom interface so end-users could easily analyze CT scans. The interface includes draw widgets, sliders, buttons and pointers to aim the radiation beams. Consequently, physicians interact with the system to detect and treat tumors more quickly and with greater precision. Cline likes the fact that he also can program at the command line and that "IDL is not a low-level language but a complete package, accessible to any end-user."

Using CT Scans

Furthermore, Cline says that it would take an entire day to do what he does in a half hour with IDL. To test the database, Cline uses simple geometric shapes and implements them on a CT scan for comparison. Cline contends that making spline curves to analyze data would be much more tedious with a traditional language. Cline asserts, "The group wouldn't be anywhere near this far using C++; we'd be in the Stone Age."