Heang Ping Chan

Image feature analysis and computer‐aided diagnosis in digital radiography. I. Automated detection of microcalcifications in mammography

We have investigated the application of computer-based methods to the detection of microcalcifications in digital mammograms. The computer detection system is based on a difference-image technique in which a signal-suppressed image is subtracted from a signal-enhanced image to remove the structured background in a mammogram. Signal-extraction techniques adapted to the known physical characteristics of microcalcifications are then used to isolate microcalcifications from the remaining noise background. We employ Monte Carlo methods to generate simulated clusters of microcalcifications that are superimposed on normal mammographic backgrounds. This allows quantitative evaluation of detection accuracy of the computer method and the dependence of this accuracy on the physical characteristics of the microcalcifications. Our present computer method can achieve a true-positive cluster detection rate of approximately 80% at a false-positive detection rate of one cluster per image. The potential application of such a computer-aided system to mammographic interpretation is demonstrated by its ability to detect microcalcifications in clinical mammograms.

Digital mammography: ROC studies of the effects of pixel size and unsharp-mask filtering on the detection of subtle microcalcifications

We investigated the spatial resolution requirement and the effect of unsharp-mask filtering on the detectability of subtle microcalcifications in digital mammography. Digital images were obtained by digitizing conventional screen-film mammograms with a 0.1 X 0.1 mm2 pixel size, processed with unsharp masking, and then reconstituted on film with a Fuji image processing/simulation system (Fuji Photo Film Co., Tokyo, Japan). Twenty normal cases and 12 cases with subtle microcalcifications were included. Observer performance experiments were conducted to assess the detectability of subtle microcalcifications in the conventional, the unprocessed digital, and the unsharp-masked mammograms. The observer response data were evaluated using receiver operating characteristic (ROC) and LROC (ROC with localization) analyses. Our results indicate that digital mammograms obtained with 0.1 X 0.1 mm2 pixels provide lower detectability than the conventional screen-film mammograms. The detectability of microcalcifications in the digital mammograms is improved by unsharp-mask filtering; the processed mammograms still provide lower accuracy than the conventional mammograms, however, chiefly because of increased false-positive detection rates for the processed images at each subjective confidence level. Viewing unprocessed digital and unsharp-masked images in pairs resulted in approximately the same detectability as that obtained with the unsharp-masked images alone. However, this result may be influenced by the fact that the same limited viewing time was necessarily divided between the two images.

The validity of Monte Carlo simulation in studies of scattered radiation in diagnostic radiology

A computer program using Monte Carlo methods has been developed for the simulation of photon scattering in tissue-equivalent phantoms for incident x-rays in the diagnostic energy range. The study verified that the sampling schemes used in the program can produce random samples according to the theoretical probability distribution functions which describe the photon-scattering process. The Monte Carlo program was applied to determinations of the scatter fractions and edge responses for various phantom sizes, X-ray energies, and recording systems. These quantities were also measured experimentally under comparable imaging conditions. Excellent agreement was obtained between the predicted and experimental results. This investigation established the validity of our Monte Carlo calculations for studies of the physical characteristics of scattered radiation in diagnostic radiology.

Physical characteristics of scattered radiation in diagnostic radiology: Monte Carlo simulation studies

We applied Monte Carlo methods for the simulation of x-ray scattering in water phantoms. The phantom thickness was varied from 5 to 20 cm, and the monoenergetic incident x rays were varied from 15 to 100 keV. Eight screen pairs and a total absorption system were used as x-ray receptors. We determined the angular, spectral, and spatial distributions of the scattered radiation and the scatter fractions recorded in the image plane. The dependence of these properties on the incident x-ray energy, the phantom thickness, and the energy response of the recording system was examined. The results of this study provide useful information for the development of antiscatter techniques and for the evaluation of radiographic procedures.

Energy and angular dependence of x-ray absorption and its effect on radiographic response in screen-film systems

The relationship between the film density and the x-ray energy absorbed in the screens, and the effect of the components of a screen--film system on x-ray absorption in the screen phosphor has been studied. It is concluded that, to a good approximation, the total absorbed x-ray energy can be calculated ignoring components other than the phosphor, and that the absorbed energy is related to the film density through the characteristic curve of the film. This simple relationship provides a basis for the theoretical prediction of the radiographic performance of an imaging system. Analytical expressions have been derived for the energy absorbed in model screens containing one or two high-atomic-number elements. For incident energies above the K-edge of the high-Z elements, the contribution of the K x-rays to energy absorption was included by use of a K-reabsorption factor. This factor was determined for eight screen pairs as a function of incident photon energy and incident angle.

Computer-aided diagnosis in chest radiology

Digital radiography offers several important advantages over conventional systems, including abilities for image manipulation, transmission, and storage. In the long term, however, the unique ability to apply artificial intelligence techniques for automated detection and quantitation of disease may have an even greater impact on radiologic practice. Although CAD is still in its infancy, the results of several recent studies clearly indicate a major potential for the future. The concept of using computers to analyze medical images is not new, but recent advances in computer technology together with progress in implementing practical digital radiography systems have stimulated research efforts in this exciting field. Several facets of CAD are presently being developed at the University of Chicago and elsewhere for application in chest radiology as well as in mammography and vascular imaging. To date, investigators have focused on a limited number of subjects that have been, by their nature, particularly suitable for computer analysis. There is no aspect of radiologic diagnosis that could not potentially benefit from this approach, however. The ultimate goal of these endeavors is to provide a system for comprehensive automated image analysis, the results of which could be accepted or modified at the discretion of the radiologist.

Studies of x-ray energy absorption and quantum noise properties of x-ray screens by use of Monte Carlo simulation

The imaging properties of the phosphor layer in fluorescent screens or image intensifiers are related to its x-ray absorption characteristics. In this study, we applied Monte Carlo methods for the simulation of x-ray photon diffusion in a phosphor layer. The K-reabsorption factor, absorbed x-ray energy, quantum absorption efficiency, statistical factor, and noise-equivalent absorption were determined as a function of the incident energy and angle of the x rays for eight commonly used phosphor layers. These basic physical quantities will be useful for the prediction of the information transfer properties of a phosphor layer.

Performance of antiscatter grids in diagnostic radiology: Experimental measurements and Monte Carlo simulation studies

We have devised an experimental method with which one can accurately measure the transmission of primary radiation and the transmission of total radiation by an antiscatter grid in a setting similar to a practical radiographic examination. We measured the transmission values for 27 combinations of x-ray tube potentials, phantom thicknesses, screen-film systems, and grid parameters. The standard deviation of one measurement was estimated to be 2.9% and 1.5% for the total and primary transmissions, respectively. The measured grid transmission was compared with results predicted by our Monte Carlo calculations; 92% of the measured and calculated values agree within two standard deviations. This close agreement indicates that our Monte Carlo calculation can accurately predict the performance of antiscatter grids under diagnostic imaging conditions.

Investigation of the performance of antiscatter grids: Monte Carlo simulation studies

Monte Carlo methods have been applied to the study of the performance of antiscatter grids in radiography. Photon histories for imaging with a 20 cm thick water phantom irradiated with an 80 kV x-ray spectrum were generated by Monte Carlo simulation. The scattering and absorption of the photons transmitted from the phantom were traced in the grid. Grids were evaluated with strip densities ranging from 10 to 67 lines/cm, grid ratios from 6 to 15, and lead-to-interspace ratios from 1/9 to 1/2. The contrast improvement factor and the Bucky factor were used as the benefit and cost indicators, respectively, for evaluation of the grids. The dependence of these factors on the grid ratio, lead-to-interspace ratio, strip density, lead content, and interspace material was investigated. The results indicate that high-strip-density grids with thin lead strips and high grid ratios can provide higher contrast improvement factors than are achieved with low-strip-density grids, without an increase in patient exposure. This theoretical approach is useful for the prediction and development of optimal antiscatter grids for radiographic procedures.

Automated Tracking Of The Vascular Tree In Dsa Images Using A Double-Square-Box Region-Of-Search Algorithm

We have developed a double-square-box region-of-search algorithm for semi-automated tracking of the vascular tree. Starting from a user-specified point, the algorithm accurately and automatically tracks the connected vessels in the entire vascular tree. In this study, we used the tracking algorithm in conjunction with the iterative deconvolution technique to obtain accurate vessel sizes and contrasts along the vessels being tracked. The tracking algorithm used the vessel size information to generate square-box regions of search for subsequent tracking points. Both the vessel size and the vessel contrast were monitored to determine the point at which tracking of a particular vessel would be terminated. The vessel sizes and contrasts along the vascular tree will be useful for further quantitative analysis of the vessels. At present, the algorithm accurately tracks the vascular tree into regions in which the ratio between the vessel contrast and the root mean square (RMS) noise is approximately 3 for vessel sizes of 0.5 mm or more.