Development of Virtual Photographic and Radiographic Imaging Simulator (ViPRIS)

Diagnostic x-rays are the largest source of man-made radiation exposure. A major goal of radiography is to maximize the amount of diagnostic information while minimizing the radiation risk to the patient. Optimization of the x-ray imaging chain is difficult because, the chain contains a large number of variables including noise, energy, patient anatomy, scatter removal grid, scintillation detectors, lesions, radiation dosimetry, and the ability to judge the quality of generated images.

We have been developing the Virtual Photographic and Radiographic Imaging Simulator (ViPRIS). The objective of ViPRIS is to allow the user to generate simulated images of various conditions. The user can then determine which settings provide images of diagnostic quality while minimizing dose to the patient. ViPRIS utilizes two sets of anatomically identical images: photographs of the sliced VHP-Male cadaver and Computed Topography data sets from the Visible Human Project (VHP). ViPRIS utilizes a Projection Phantom constructed from the CT images to simulate primary unscattered x-rays. The VIP-Man phantom has been constructed from segmented high-resolution color photographs. The VIP-Man phantom is used to generate dosimetry information and a scatter image. The phantoms are of the same individual and allows for the generation of simulated radiographs in two phases as detailed below. An illustration of the ViPRIS system may be found in Fig.1.

Materials & Methods

ViPRIS simulates an x-ray in two phases. First, for a given x-ray tube setting, ViPRIS calculates the primary x-ray image by tracing each x-ray through the CT Projection Phantom. Simultaneously, ViPRIS calculates the scattered x-rays using a stored data file. The data file has been pregenerated using a Monte Carlo simulation of the VIP-Man Phantom. Organ doses and effective doses are simultaneously calculated based on the user-desired fluence and stored dosimetry data.

a) Primary X-ray Image

Primary x-ray images are formed by photons that transmit through the patient anatomy without interactions. VHP CT images are converted into attenuation coefficients. Primary x-ray transmission is calculated using I=I0e-tSm.

b) Monte Carlo Scatter Image

Compton scattering towards the detector adds a background fog to the primary unscattered image. A critical method used to increase the signal-to-noise ratio in radiography is attenuating the scattered signal from the image receptor. The scattered photon pattern within a radiograph is a phenomenon that is still not fully characterized. Monte Carlo techniques are ideally suited for modeling this random process. We used the Monte Carlo computer code, EGS4nrc to simulate the detailed photon interactions in the Dosimetry phantom. The scatter energy profile is stored for later scaling based on the specifications of each simulation. By storing the scatter energy profile, multiple simulations may be completed in a reasonable time frame.

c) Combining Primary and Scatter Images

The square root of the raw calculated photons are assumed to be the standard deviation or noise component of each pixel. Because the simulation is scale based on a stored profile, additional Gaussian quantum noise is added based the user specified fluence. The transmitted photon fluence is converted into a photon energy fluence. The log of the scatter and primary energies are then combined based on the properties of the filtration grid and detector.

Results

When the photon fluence used to create a radiographic image is increased, the number of photons that reach the sensor will also increase, leading to an improved signal-to-noise ratio. Unfortunately, increasing the fluence will simultaneously increase the radiation dose to the patient. When a small number of photons are simulated, the projected image is grainy (noisy) and it is difficult to observe fine details.

Total effective doses and organ dose have been calculated for photons incident for several views and charted as Sv/particle

Conclusions

We have constructed ViPRIS to study optimization of the x-ray imaging chain. The database created will be combined with observer studies. This will allow us to provide insight into settings at which the radiation dose to patients will be minimized while simultaneously providing adequate diagnostic information to physicians.

Documents

  • Winslow, M. Radiographic Image Simulation Using Monte Carlo Image Projection Techniques. Rensselaer Polytechnic Institute PhD Dissertation. May 2004 view view
  • Sample Images view view
  • EDE for Chest Examinations view view
  • Winslow, M. Radiographic Image Simulation Using Monte Carlo Image Projection Techniques. Doctoral Candidacy Exam. view view
  • Presentation at HPS 2004 view view
  • Presentation at AAPM 2004 view view