James R. Symon

The tetrad and hexad: Maximum beam separation as a starting point for noncoplanar 3D treatment planning: Prostate cancer as a test case

PURPOSE In contrast to computer optimized three-dimensional (3D) treatment planning, we have used maximally separated, noncoplanar beams as the starting point for 3D treatment planning of prostate cancer to maximize the rate of dose fall off from the target volume and minimize dose to surrounding tissues. MATERIALS AND METHODS A planar four-field plan, a planar six-field plan, a tetrad plan, and a hexad plan are analyzed using a 3D treatment planning system which is capable of displaying real-time 3D dose distributions within volume reconstructed data sets (VISTAnet--an extension of the virtual simulator). The tetrad plan is based on the methane molecule and the hexad plan has a minimum separation of 58 degrees on beam entrance. All fields are conformal. The irradiated volume equals the clinical target volume plus a 1 cm margin. Competing plans are compared using cumulative dose-volume histograms and normal tissue complication probabilities. RESULTS The crossover point, the isodose surface that conforms more to the beams than the target, is introduced and described. The hexad and tetrad plans result in tighter dose distributions when compared to the planar plans with the same number of beams. The tetrad plan treats a volume less than or equal to the planar six-field plan at isodose surfaces above 18% except between 37% and 44% where the tetrad volume is slightly larger. As expected from integral dose considerations, the amount of normal tissue receiving some radiation increases, but the amount receiving clinically significant amounts of radiation decreases as the number of beams increase. The plan involving the largest number of noncoplanar beams results in the tightest isodose distribution. Analysis of rectal and bladder cumulative dose volume histograms does not reveal a clearly superior plan based on normal tissue complication probabilities. CONCLUSIONS Using basic principles of solid geometry, maximally separated beams without significant overlap on exit or entrance can be designed which minimize clinically significant dose to surrounding tissues and tighten the isodose distribution around the target volume. The emphasis of this treatment plan optimization is geometric in contrast to methods using computer optimization or artificial intelligence.

Using Custom-designed Vlsi Fuzzy Inferencing Chips For The Autonomous Navigation Of A Mobile Robot

An approach merging the concepts of fuzzy logic and sensor-based behaviors has been developed to add a qualitative reasoning capability to the real-time control of autonomous mobile robots. The approach is implemented using custom-designed computer boards which include recently developed VLSI fuzzy inferencing chips. The design and operation of these boards are first described and an example of their use for the autonomous navigation of a mobile robot is presented. The development of qualitative reasoning schemes emulating human- like navigation in a-priori unknown environments is discussed. The approach using superposition of elemental sensor-based behaviors expressed in the fizzy Sets theoretic framework is shown to allow easy development and testing of the inferencing rule base, while providing for progressive addition of behaviors to resolve situations of increasing complexity. The efficiency of such schemes, which can consist of as little as a dozen qualitative rules, is illustrated in experiments involving an autonomous mobile robot navigating on the basis of very sparse and inaccurate sensor data.

VISTAnet: interactive real-time calculation and display of 3-dimensional radiation dose: An application of gigabit networking

Three-dimensional treatment planning can allow the clinician to create plans that are highly individualized for each patient. However, in lifting the constraints traditionally imposed by 2-dimensional planning, the clinician is faced with the need to compare a much larger number of plans. Although methods to automate that process are being developed, it is not yet clear how well they will perform. VISTAnet is a 3 year collaborative effort between the Departments of Radiation Oncology and Computer Science at the University of North Carolina, the North Carolina Supercomputing Center, BellSouth, and GTE with the medical goal of providing real-time 3-dimensional radiation dose calculation and display. With VISTAnet technology and resources, the user can inspect 3-dimensional treatment plans in real-time along with the associated dose volume histograms and can fine tune these plans in real-time with regard to beam position, weighting, wedging, and shape. Thus VISTAnet provides an alternate and, possibly, complementary approach to computerized searches for optimal radiation treatment plans. Building this system has required the development of very fast radiation dose code, methods for simultaneously manipulating and modifying multiple radiation beams, and new visualizations of 3-dimensional dose distributions.

Digitally reconstructed fluoroscopy and other interactive volume visualizations in 3-D treatment planning

PURPOSE Add radiographic context to the beams-eye-view used in 3-dimensional treatment planning. Improve methods for interactive visualization of anatomy and dose distributions. METHODS AND MATERIALS Most 3-dimensional treatment planning systems feature a beams-eye view that includes only graphical representations of patient anatomy. With input devices such as a mouse or trackball, the user interactively shapes the treatment field using the graphical models to provide geometric information. Radiographic context provides additional geometric information important for determining field shape. We have implemented digitally reconstructed fluoroscopy in the beams-eye view by increasing the efficiency for computing digitally reconstructed radiographs. In addition we have improved algorithms for real-time surface and volume rendering for anatomy and doses using an experimental graphics supercomputer. RESULTS Without radiographic context in the beams-eye-view, field shapes were sometimes changed after simulation or portal images were obtained. Digitally reconstructed fluoroscopy has essentially eliminated these changes. Higher quality interactive three-dimensional displays improve the comprehension, confidence and efficiency of the user. Our improvements have already been implemented on one model of a new generation of commercial graphics workstations. CONCLUSION Addition of radiographic context to the beams-eye-view is recommended. Incorporation of higher quality interactive graphics is rapidly becoming practical and is encouraged.

VISTAnet: radiation therapy treatment planning through rapid dose calculation and interactive 3D volume visualization

VISTAnet, an experimental gigabit network test bed, ties together a CRAY Y-MP, the Pixel-Planes 5 graphics engine, and an SGI host machine to create a metacomputer capable of real-time radiation therapy dose calculation and display. We report on the methods used to manipulate and examine the 3D radiation dose distribution, with emphasis on the visualization, which uses a parallel, interactive, multimodal renderer implemented on Pixel- Places 5. The real-time display is designed to facilitate comprehension of spatial relationships among the geometrically complex anatomy and radiation dose structures that characterize a 3D radiation treatment scenario. The currently ongoing clinical evaluation of VISTAnet has already yielded encouraging results.

Applications of image processing and visualization in the evaluation of murder and assault

Recent advances in image processing and visualization are of increasing use in the investigation of violent crime. The Digital Image Processing Laboratory at the Armed Forces Institute of Pathology in collaboration with groups at the University of North Carolina at Chapel Hill are actively exploring visualization applications including image processing of trauma images, 3D visualization, forensic database management and telemedicine. Examples of recent applications are presented. Future directions of effort include interactive consultation and image manipulation tools for forensic data exploration.