We are developing methods for imaging acoustic parameters individually to obtain a more direct interpretation of the scattering medium. Two properties of the medium that describe microscopic scattering structure have been accurately measured: the average scattering particle size and the product of the number density and the scattering strength. Analysis of quantitative images show increased image contrast by as much as 6 dB between regions that vary in these acoustic properties over that of conventional B-mode imaging. It is also possible to separate the effects of scatterer size from the effects of the number density and scattering strength. The motivation for developing these techniques for diagnostic ultrasound is to improve low-contrast detectability.

Techniques exist for reconstructing current density vectors from measurements of the three components of the generated magnetic field. In order to have a practical data collection scheme, the number of measurements must be as small as possible. In addition, some technical difficulties exist in using SQUID detectors to measure all three components of a magnetic field simultaneously. This paper presents a method for extrapolating the missing x- and y-components of a biomagnetic vector field from the measured z-component, under certain readily met boundary conditions. Specifically, we address the solution of such a problem when the measured data consists of the z-component in a single plane. The boundary conditions that must be met in order to perform the extrapolation require that the z-component values fall off to zero on the edges of the measurement plane, and that the extrapolation be performed in a region free of current sources.

The magnetic lead field provides a relation between the current density distribution in the patient and the measurement of a SQUID magnetometer. This approach can be derived in a new and simple way by a derivation based on energy concepts. Using the lead field and conservation of charge conditions provides two linear, spatially invariant imaging equations relating the current density and flux measurements made by a planar array of SQUID magnetometers. These equations, plus an assumption that the current density is confined to a plane, are solved using Fourier techniques. The validity of the resulting equations is shown both analytically and with a computer model. The effects of not satisfying the planar assumption are described.

We describe demonstration experiments for a new method of medical diagnostic imaging. The method is called Chrono-Coherent Imaging (CCI) and it can be used in a transmission geometry to form images in the presence of overwhelming scattered light, which blocks conventional image formation. Future applications are for imaging inside the human body where tissue light scattering normally obscures image formation. In a transmission geometry the scattered light will take different time delays to reach a recording medium than will the very weak unscattered light which contains image information. The recording of the image for a series of times is not done as a real image with ultrashort gating devices such as streak cameras or Kerr shutters, but the recording is done coherently like a hologram with the sub-picosecond coherence properties of the laser pulse. By using a time sweep of a reference laser beam on the recording medium, similar to Light-In-Flight Holography, we can make a series of coherent images with a single laser exposure, even in the presence of very large incoherent exposure by the scattered light. These images are much like an X-ray in that cumulati've transmission effects are recorded throughout the object, but in CCI the time series of images has both refractive index and absorption information. Many other features such as tissue selectivity by wavelength tuning, depth enhancement and three dimensional image reconstruction are possible with the new imaging method of CCI.

The feasibility of using visible and/or near-infrared radiation for screening for breast cancer has been examined. Results are presented of measurements of the total attenuation coefficient of pig adipose at wavelengths of 633nm and 1.15μm, and of experiments which attempt to discover how transillumination image resolution might be improved by collimating the scattered radiation.

As a part of continuing investigation, this work aims to further characterize biological specimens, mantained in living conditions, through the use of non-invasive phase laser-microscopy. Dark-field and phase-contrast laser-microscopy and the relative signal processing techniques have been realized to recover the spatial distribution of the phase map of the object, obtainable after a proper processing. A simplified theory of image formation is discussed and some experimental results, on living CHO cells, are reported.

A Solid State Microscope (SSM) has been designed and is being developed in order to improve and optimize spatial, photometric and spectral resolution for quantitative microscopy. The SSM is an opto-electronic device for scanning and viewing microscopic objects in the visible light spectrum. Using Kohler illumination, pulsed light is transmitted through the microscope sample and focused by a single objective lens. The objective magnifies and projects the image onto a large area charge-coupled device (CCD) located at the intermediate focal plane (PIP). Sample scanning procedures and optical considerations require a CCD array of greater than 1000 x 1000 picture elements, with each element having a sensing area of approximately 7 μm x 7 μm. The signals from the picture elements are directly digitized and mapped on a one-to-one basis into a large frame memory at a rate of 20 Mpixels/s. The full digital image is continuously displayed in real-time at a rate of 60 frames/s on a gray scale monitor. The images stored in frame memory are accessed by workstation for quantitative measurements. The rationale for these specifications and in particular, how CCD characteristics relate to the optics and image display are discussed. Of particular importance is the sampling density which has been experimentally determined. The data shows that oversampling the image 3 to 4 times will be optimal for this design and a spatial resolution of at least 0.4 μm can be expected at a field of view of 1.4 Mpixels.

MRI has recently gained popularity for non-invasively detecting and measuring blood flow. This is largely attributable to the development of several new flow sensitive MRI techniques. All of these techniques take advantage of either time-of-flight flow effects or the dependence of the spin phase on motion. While these phenomena have been understood for several decades, conventional MR imaging techniques have ignored them and assumed that the volume being imaged remains stationary for the duration of the exam. Consequently, conventional MRI images often have signal artifacts from flowing spins. These artifacts typically appear in the images as bright or dark signal regions which may be repeated along the phase encoding direction. New MRI sequences have been developed which exploit the source of these artifacts to reveal information about the nature of flow ranging from the anatomical mapping of vessels (MR angiography) to a quantitative analysis of velocity and higher order motion components. Unfortunately, the myriad of parameters of the imaging sequence (slice thickness, resolution, magnetic field gradient strength and duration, TR, TE) along with physical parameters of the image volume and its flow (relaxation times, distribution and direction of flow, velocity, acceleration, pulsatility) make the accurate measurement of flow a difficult task.

The experimentally determined performance characteristics of selected mammographic screen-film combinations are described and compared in detail, including absolute sensitometry, modulation transfer functions, noise equivalent quanta, and detective quantum efficiency. Advantages of each screen-film combination for various mammographic applications are also described.

The AMBER system is an equalization-radiography system (Oldeift), which uses a horizontal slit-beam. The system allows local exposure control by means of a feed back loop between detectors in front of the film and modulators in front of the X-ray tube. The AMBER-system reduces the Scatter-to-Primary-Ratio (SPR) in the mediastinum. The SPR reduction is related to the compression-curve of the system. In this study the SPR-reduction was 37%. A contrast-detail phantom was designed and used in combination with a chest-phantom. Contrast-detail-studies were performed to evaluate the improvement of image quality when film and stimulable phosphor plates were used as detector. Film images(150) and stimulable-phosphor-plate images (300) obtained with a conventional exposure technique and with the use of AMBER, were evaluated by four observers (radiologists). Half of the stimulable pho'sphor plates were digitally enhanced (adaptive filtration). Application of the AMBER-system significantly improves the image quality of film and stimulable phosphor plates in the mediastinum while no change of image quality in the lung was found. The discernability of low contrast details was not significantly improved by adaptive filtration in regions with adequate exposure, a small improvement was seen in "underexposed" regions.

Multivariate moment-generating functions provide a useful method for analyzing the influence of stochastic amplifying and scattering mechanisms on the transfer of signal and noise through multistage imaging systems. In this paper we apply this method to several cases of amplifying and scattering mechanisms in which the parameters are themselves stochastic variables. These include input-labeled amplification, input-labeled scattering, position-labeled amplification, and position-labeled scattering. Our results are general in that they are applicable to a variety of imaging chain models and can be used to analyze stationary as well as nonstationary processes. In the special case of stationary processes, relationships between the input and output NPS can be derived. For the case of photon-limited inputs, one can also readily express the DQE. For each case, an example illustrating the physics of radiographic intensifying screens is given. Previously published theoretical results are shown to follow as special cases of these new and more general results.

In both computed tomography (CT) and magnetic resonance (MR) imaging, continuous video-rate data acquisition, with retrospective cine-mode viewing of the reconstructed image sequence, is possible. With video-rate image reconstruction, a videographic capability for these imaging modalities is achieved, whereby dynamic events within the human body can be viewed in real-time. The computational requirement of real-time image reconstruction is extreme and has not been demonstrated to date using conventional digital-electronic circuitry. The high degree of parallelism of analog optical computers can be exploited to achieve very high computational rates and accomplish the goal of video-rate reconstruction of these medical images. Systems for CT and MR image reconstruction are described, results are presented, and performance of the systems is discussed.

A theory is presented which analyzes the performance of a computerized tomography system as a detection system given the detector geometry, optics, and reconstruction algorithms. The algorithmic effects are seperated from the intrinsic signal-to-noise frequency characteristics of the acquired data. This allows one to: (1) Compare the potential image quality that can be achieved by different CT systems independent of the reconstruction algorithm applied. (2) Predict the performance of new detector configurations. (3) Design the appropriate reconstruction algorithms to achieve the desired clinical results on a given scanner. Examples will be given of how different detector geometries can be compared using this theory and how specific reconstruction filters can be designed given both the scanner characteristics and desired image quality parameters.

In order to improve the performance of the instrumental variable method (IVM) in calculating regional cerebral blood flow (rCBF) using Single Photon Emission Computed Tomography (SPELT), and inert diffusible tracer such as 133Xe, we use Learning Algorithms for Multivariate Data Analysis (LAMDA) to classify the voxels of the images of local concentrations in the brain. The LAMDA method correctly distinguished between extra and intra-cerebral voxels. However the topography of the intra-cerebral classes did not match the Regions Of Interest (ROI) defined on an anatomical basis. Provided that all the intra-cerebral classes contaminated by bone and air passage artefact were rejected, the results given by the NM are in good agreement with those derived by the bolus distribution principle. We thus conclude that LAMDA methods can improve the reliability of images of CBF estimates.

A prototype volume CT system for use in angiography has been modified and tested using three different phantoms. This system consists of a fixed x-ray tube, a conventional image intensifier coupled to a charge-coupled device (CCD) camera, a computer-controlled turntable on which phantoms were placed. In order to explore the imaging performance of the system for reconstructing a three dimension (3D) vascular structure, two sets of projection images of a vascular phantom, acquired over 250 projection angles with two different sizes of image intensifier were digitized and used for a direct 3D cone beam reconstruction. The spatial resolution limits of the system were measured from the 3D reconstructed images of a specially-designed resolution phantom. The direct 3D reconstructions of a Humanoid chest phantom were obtained using this system to show the perspective of the system for a general medical application with 3D imaging. The quality of the reconstructed images indicates that the system can be used for achieving a direct 3D reconstruction of a vascular structure as well as a general 3D object. The measured spatial resolution of the system is nearly half of the nominal resolution of the image intensifier and is reduced around the edge of the image intensifier, primarily because of the pincushion distortion.

A method for optimizing image-recovery algorithms is presented that is based on how well the specified task of object localization can be performed using the reconstructed images. The task performance is numerically assessed by a Monte Carlo simulation of the complete imaging process including the generation of scenes appropriate to the desired application, subsequent data taking, image recovery, and performance of the stated task based on the final image. This method is used to optimize the constrained Algebraic Reconstruction Technique (ART), which reconstructs images from their projections under a nonnegativity constraint by means of an iterative updating procedure. The optimization is performed by finding the the relaxation factor, which is employed in the updating procedure, that yields the minimum rms error in estimating the position of discs in the reconstructed images. It is found that the optimum operating points for the best object localization are essentially the same as those obtained earlier when the performance of simple object detection is to be optimized.

The simple neural network composed of layers of "neurons" which threshold the sum of weighted input values received from the neurons of the immediately preceding layer is directly applicable to many signal detection tasks. Use of the maximum likelihood "ideal observer" formalism allows for a quantitative measurement of neural-network performance for such tasks, in contrast to measures of network convergence such as sum of squares error. For the signal known exactly (SKE) problem, for which a known signal is detected on a noisy background, a single-layered neural network with ideal weights performs at 100% efficiency, since it is isomorphic to the matched-filter ideal observer. Measured efficiencies of less than 100% primarily reflect idiosyncrasies of the training method (e.g., back projection) and incomplete training of the network. For more complicated problems which have terms in the ideal test statistic that are quadratic or higher in the data, more complex neural-network architectures are required. Even then, however, convergence to optimal solutions is not assured. Using the feed-forward processing, back-projection training paradigm leads to far from ideal performance for these higher order tasks, at least when training is carried out on noise-free data or small sets of noisy data.

Performance for several apertures is presented for a number of Rayleigh discrimination tasks with signal and background exactly specified. Performance is defined as the squared signal-to-noise ratio of an ideal observer determined from statistical decision theory. The conclusions of Wagner, Brown, and Metz (1981) are shown to hold for different source-pair orientations and some other well-known (but non-ideal) figures of merit. When the background is assumed to be a known constant, and the source width and separation are also known, the performance of a simple open aperture increases as the aperture is enlarged. For a known source width a complex aperture can be designed which will give performance superior to a large open aperture for these simple discrimination tasks. For any of these apertures to be clinically relevant, performance comparisons over a wider range of clinically realistic tasks, including signal and object variability, must be considered.

Two studies of the effect of background inhomogeneity on observer performance in radionuclide emission imaging are presented. In the first, the task is detection of a Gaussian blob, and the imaging aperture is a pinhole of Gaussian profile. In the second, a simple discrimination task called the Rayleigh task is considered, and the aperture has a rectangular profile. In both cases performance of a suboptimal linear observer is calculated; in the first study the observer is one derived in a classic paper by Harold Hotelling, while in the second study the observer is a simple non-prewhitening matched filter. In both studies an important variable is the aperture size, and a key question is whether a small aperture or compact point spread function is advantageous. The main result is that a large aperture may perform very well or even optimally with a spatially uniform background but fail badly when the background is non-uniform. Thus predictions of image quality based on stylized tasks with uniform background must be viewed with caution.

The linear prewhitening matched filter (PWMF) is the optimal decision function for discriminating exactly-specified signals in additive Gaussian noise. When the signals are less well specified the optimal decision function contains higher than linear terms in the data. Several examples of detection and discrimination tasks are presented for which only the linear term is required and other examples for which higher order terms are necessary. Even in the nonlinear case decision functions can often be approximated by linear operations on the data followed by logical operations. This combination of linear weights plus logic operations is typical of neural network models, which are thought to be elementary models of human processing mechanisms. A number of experiments suggest that there is no decrease in performance of humans for some complex tasks that ideally require such nonlinear operations. However, there are other such tasks where human performance is degraded and this appears to be due more to the complexity of the task and the nature of the correlations in the image than the order of the task. This suggests applications in diagnostic imaging where it might be advantageous for a machine viewer to substitute or work in conjunction with the human observer.

We have investigated the feasibility of an optical beam attenuator (OBA) for image equalization in the optical light path of a camera-based x-ray imaging system. In our OBA design, a liquid crystal device (LCD) operated in transmissive mode is used to equalize low spatial frequencies. In most camera systems, compression of image dynamic range will result in improved performance. The advantages of optical equalization include bit compression for digital applications, video-rate equalization, detective quantum efficiency (DQE) improvement at high dose rates, and high-pass frequency filtration for film-based applications. A prototype OBA system has been constructed and installed in our DSA imaging laboratory. Equalized phantom images have been obtained which demonstrate improved visualization of structures in the darker regions of the image. The effect of the OBA on system limiting spatial resolution and system noise performance has also been investigated.

In the area of high resolution X-ray TV cameras, the present needs are better image quality and more functional capabilities. These requirements can be net with the new PRIMICON pick-up tube that includes up-to-date technologies electrostatic deflection, noncrossover diode gun, integrated light bias and improved photoconductive target. The paper describes how these technological features enhance basic performance such as resolution uniformity between the center and the edges of the image, fully controlled lag and tube compactness which pave the way to cameras combining "High Definition" quality and multimode capability (fluoroscopy/DSA).

Many general vascular examinations require a large field of view. For this purpose, a new 16" X-ray image intensifier has been developed incorporating most recent technologies. In operation, the radiologist will appreciate its useful view of over 36 cm, and its ability to zoom to 16 cm in four steps down. In spite of this large field of view, the tube is light and compact. Its small, 30 mm output image grants a high light collection efficiency with small-volume lenses. The use of high density output phosphor and a new stray-light-absorbing output window result in important improvements in resolution and contrast, especially on small objects. The overall performance are described in this paper.

This investigation concerns the sources and the characteristics (importance and structure) of veiling glare in angiographic systems. The veiling glare is modelled by empirical PSFs. The parameters of these PSFs are fitted from contrast ratio measurements. Various acquisition configurations are examined. Theoretical expressions have been developed to study the spatial variance of the PSFs and the contribution of the image intensifier (II) input screen in the global veiling glare. The results are both qualitative and quantitative. The spread of the veiling glare is well modelled by the sum of short and long distance terms. The importance of the II input screen is shown and is related to both short and long range effects.

We are developing a flash x-ray scanner for stereoradiography and CT which will be able to produce a stereoradiograph in 30 to 70 ns and a complete CT scan in one microsecond. This type of imaging device will be valuable in studying high speed processes, high acceleration, and traumatic events. We have built a two channel flash x-ray system capable of producing stereo radiographs with stereo angles of from 15 to 165 degrees. The dynamic and static Miff 's for the flash x-ray system were measured and compared with similar MIT's measured for a conventional medical x-ray system. We have written and tested a stereo reconstruction algorithm to determine three dimensional space points from corresponding points in the two stereo images. To demonstrate the ability of the system to image traumatic events, a radiograph was obtained of a bone undergoing a fracture. The effects of accelerations of up to 600 g were examined on radiographs taken of human kidney tissue samples in a rapidly rotating centrifuge. Feasibility studies of CT reconstruction have been performed by making simulated Cr images of various phantoms for larger flash x-ray systems of from 8 to 29 flash x-ray tubes.

It is shown, that the Image Quality parameters, Modulation Transfer Function (MTF) and noise can be evaluated digitally with high accuracy. The systems MTF is measured by the use of a 5 slits phantom. Simulation studies are showing the optimal phantom dimension and indicate the expected precision of the measurement. By comparison studies with conventional techniques the MTF measurement procedure is validated for the effects of pixel coarseness, noise and limiting resolution. An analytical expression for signal-to-noise ratio in a digital radiographic system is developed. Predicted values are compared with measured data.

Linear solid state X-ray detectors are used to produce digital images. The detectors produce the X-ray image line by line as the patient is scanned with a narrow X-ray beam. One of the decisive advantages of this principle of imaging is that it rejects scattered radiation. This translates to an increase in image contrast. The main functions of a solid state X-ray detector are : image detection ; signal multiplexing ; DC offset and gain calibration. The output signal is a digital signal coded on 8 bits or 12 bits. The dynamic range is 4000:1. The avantages of solid state linear detectors are : large field of view (for example 18") ; high contrast ; direct digital output ; high dynamic range. Multilinear detectors will be described. They integrate a TDI function (time delay and integration) in synchronism with the scanning.

The analysis of spatial-resolution properties in terms of the modulation transfer function (MTF) has been presented in a computed radiographic (CR) system (FCR-101) with the photostimulable-phosphor plate (imaging plate, IP). The newly devised method of determining the presampling MTF which includes the x-ray detector (IP) unsharpness and the unsharpness of the sampling aperture is described in which an image of a slightly-angulated lead slit relative to a horizontal or vertical direction is employed. The IP MTFs as an analog MTF in the system, the presampling MTFs for different types of IPs, different sampling distances, different versions of IPs, simultaneous multisection tomography and magnification radiography, and the laser-printer MTFs as display MTF are measured and shown. The effective sampling aperture MTFs calculated indicate that the noticeable degradation of resolution occurs at the stage of image data sampling. The usefulness of the magnification technique for mammography and bone radiography is demonstrated. It is shown that both of the digital MTF and the overall MTF are difficult to use for general purpose due to the aliasing artifacts. The effect of glare on the contrast is also characterized by lead-disk method. The glare fraction is found to be approximately 6.5%.

We have integrated a dedicated digital projectional radiographic system for the Chest Radiology Section at UCLA. The system has been running since September, 1988 and is undergoing clinical evaluation. This paper describes the image acquisition, communication, storage, and display components of this system. An implementation of the system as a full PACS module in the Chest Radiology Section is in progress.

This paper presents results of the initial in-vivo investigations with the system NIKOS II (NIKOS = Nicht-invasive Koronarangiographie mit Synchrotronstrahlung), an advanced version of NIKOS I which was developed since 1981. Aim of the work is to be able to visualize coronary arteries down to 1mm diameter with an iodine mass density of lmg/cm2, thus allowing non-invasive investigations by intravenous injection of the contrast agent. For this purpose Digital Subtraction Angiography (DSA) in energy subtraction mode (dichromography) is employed. The two images for subtraction are taken at photon energies just below and above the iodine K-edge (33.17keV) After subtraction the background contrast from bone and soft tissue is suppressed and the iodinated structures are strongly enhanced because of the abrupt change of absorption at the K-edge. The two monoenergetic beams are filtered out of a synchrotron radiation beam by a crystal monochromator and measured with a two line detector. One scan (two images) lasts between 250ms (final version) and ls (at present ). The images from the in-vivo investigations of dogs have been promising. The right coronary artery (diameter 1.5mm) was clearly visible. With application of better image processing algorithms the images illustrated in this paper have a definite potential for improvement.

An experimental set-up of a selenium-based detector with a scanning fan-beam exposure and improved spatial resolution has been realized. The imaging performance in terms of MTF, image noise and signal-to-noise (S/N) ratios (NEQ,DQE) is measured as a function of dose and spatial frequency for 40 kVp and 50 kVp tungsten X-ray spectra. The measured S/N-performance is interpreted in terms of a simple model which is used to study the effect of geometric magnification techniques and the behaviour of the system for various beam qualities. The object detectability for weak and high contrast objects has been calculated for different beam qualities. The results are compared with the data of high-resolution screen-film systems.

We have developed an experimental digital mammography system which incorporates a beryllium-window x-ray image intensifier (XRII), allowing efficient detection of low-energy diagnostic x rays. The optical image produced by the XRII is recorded by a low-noise charge-coupled device (CCD) operating in the time-delay integration (TDI) mode. Images are obtained by translating a 7 mm x 15 cm slot of radiation across the breast and integrating the signal over 96 image lines. By using a 7 mm slot-beam, we are able to obtain a scatter-to-primary photon fluence ratio of 0.1 for a 4 cm compressed breast. At the same time, the TDI approach results in more efficient usage of available x-ray fluence than a line-scan system, so that imaging times of about 4 seconds are possible. Detector noise is minimized by cooling the CCD to about 5° C. We report on measurements of physical parameters such as modulation transfer function, noise power spectrum and detective quantum efficiency. Our results are compared with equivalent measurements on screen-film systems. Although high-contrast spatial resolution with our digital system is currently limited to 4 mm-1, the improved signal-to-noise ratio allows better low-contrast detectability than screen-film systems.

This paper describes the use of entropy to quantify leakage of large molecules in a microvascular system. This measure can be used as a global parameter to characterize leakage. A software package for analysis of a sequence of images comprising leakage in rat cremaster tissue has been developed. The analysis is based on the statistics of both gray level components and frequency components of the images. Results show that entropy provides a better measure of leakage because it does not depend on variation in illumination or translation and rotation of image objects. Moreover entropy based on frequency components provides a more sensitive leakage measure than entropy based on gray level components.

We have been developing a digital fluoroscopic imaging device to replace the portal films that are currently used to verify patient positioning during radiotherapy treatments. Our system differs in two ways from previously reported devices. Our x-ray detector, which consists of a 1.0-mm copper plate with a 400-mg/cm2 Gd2O2S:Tb phosphor layer bonded to the plate, has a much thicker phosphor layer than used previously. Thus, the light output from the detector is also much larger. The increased light output reduces the contribution by the secondary (light) quanta to noise in the images. In addition, the operation of the T.V. camera has been modified so that the light from the phosphor screen is integrated on the target of the T.V. camera for periods of 0.2-2.0 seconds. Integration of the light increases the video signal relative to the fixed noise current generated by the camera, and thus minimizes the camera's contribution to noise in the images. Image quality improves if either a thicker phosphor detector or target integration is used to form the image, and the best images result if these two are used in combination. The images obtained from the imaging system are comparable to film, and show that our fluoroscopic imaging system represents a definite alternative to film as a method of verifying patient positioning in radiotherapy.

A scanned projection radiography sensor has been developed using a time-delay-and-integrate imaging CCD optically coupled to an X-ray intensifying screen. A mathematical model has been developed to predict the efficiency of information transfer between intensifying screen and CCD. This model predicts that for an 8-inch field of view the CCD should capture one electron for each X-ray photon absorbed in the screen. Variance in the number of electrons captured should reduce the DQE of the screen by a factor of 0.7. This almost exactly offsets the effect of scattered radiation in film/screen systems, suggesting that the new detector could have image quality equivalent to a screen/film/grid system. Experimental data shows that the model correctly predicts the light photon transfer efficiency of the system but that the image noise is dominated by switching noise from the CCD. With a low-noise camera design the system should be equal or superior to a screen/film/grid system for a wide range of practical imaging situations.

At present, many dental acts involve operations carried out without visual monitoring, for example in mouth surgery or during the course of a dental root canal treatment. While this kind of operation is in progress, the dental surgeon is guided by his tactile sense alone. Only the traditional pre and post-operative radiographic examinations are possible. Waiting for a least ten minutes is the time required for these radiographs. This article describes a new material giving the dentist the means of a real time visual monitoring for the intervention being carried out. The device implemented in our laboratory is a dental fluoroscopy set with low level X-ray doses, small field, suitable resolution, and primarily usable for kinetic images acquisition. This dental fluoroscopic device consists of : - a special X-ray generator (low exposure and collimated flow), - a fiber optics bundle and its X-ray / visible converter, - an image intensifier associated to a charge coupled device (CCD), - and finally, a digital processing device for image treatment and display. A preliminary/ dosimetric study completes this work.

Although there is considerable interest in gated cardiac ECT (Ref. 1), the large amount of data makes it difficult to use this technique clinically with existing computer systems. We have developed a clinically practical technique to acquire and analyze gated SPECT data of the cardiac blood pool by performing a buffered beat acquisition at each camera position, sorting and normalizing the images, serially reconstructing sets of transverse slices, and post-processing those slices for real-time review and analysis. Processing steps are designed so that a minimum of human interaction is required at the beginning, and the bulk of the processing is subsequently done automatically by the computer in the background. We acquire 64x64 word images over 180 degrees at 32 camera angular positions for 120 seconds per position, gating the R-R interval into 16 frames, yielding a total of 4 megabytes of raw data. Standard software (Siemens MicroDelta) was modified to automatically sort, normalize, and reconstruct all 16 frames with the same parameters following a standard reconstruction of the first. After 2 minutes of user interaction, the unattended reconstruction takes 1.5 minutes per frame. Studies are reviewed by displaying the moving transverse or oblique slices in either of two cine modes, and/or defining the blood pool surface for each chamber and viewing the moving surfaces from virtually any angle selected with a joy stick. The volume calculation is based on our previously reported technique for non-gated data. The first slice-cine mode displays one slice level, with interactive control of level and rate. The second displays all slices in a "spread stack" format. Surface definition requires 20 minutes of initial operator assistance to help separate the chambers. We have acquired and analyzed data from ten patients and a rotating phantom. Patients have tolerated the one hour acquisition time well. The automated processing makes this technique feasible for daily use in a busy clinic.

Feasible images in tomographic image reconstruction are defined as those images compatible with the data by consideration of the statistical process that governs the physics of the problem. The first part of this paper reviews the concept of image feasibility, discusses its theoretical problems and practical advantages, and presents an assumption justifying the method and some preliminary results supporting it. In the second part of the paper two different algorithms for tomographic image reconstruction are developed. The first is a Maximum Entropy algorithm and the second is a full Bayesian algorithm. Both algorithms are tested for feasibility of the resulting images and we show that the Bayesian method yields feasible reconstructions in Positron Emission Tomography.

Visual pattern recognition aspects emerging as collective properties of systems of neurons are considered. In this sense, the firing activities of groups of individual neurons are seen as the elementary entities for cooperative phenomena processing. Specifically we show as massively coupled neural assemblies with antisymmetrical synaptic junctions can exhibit orientational selective properties according to the behaviour of simple cells located into mammalian V1 area. Biological and theoretical supports suggest that information is represented in the nervous system by a small number of highly connected neurons. In this paper a neural network approach to emulate orientational sensitive simple cells behaviour is taken into account. It is based on assemblies of 32 elements like neurons arranged in an isotropic mode with antisymmetrical synaptic patterns. The dynamicisms of the system are described by a dynamical linear model producing particular oscillating trajectories in the state space. The resulting system is trained to recognize specific orientation by a Hebb-like rule reinforcing the synaptic strengths activated by the input stimulus. Experimental results using this approach are presented that will respond to test patterns of related inputs.

A novel x-ray imaging device is under development for use in high resolution imaging applications such as coronary angiography. It consists of a six inch diameter, external CsI x-ray sensor, coupled fiber optically to a proximity focussed image intensifier. The image intensifier is coupled through a coplanar fiber optic taper assembly to six CCD's for readout. Analog and digital electronics have been developed to combine the six individual images to a composite image of 1152 by 1152 pixels and display that image in full resolution on a physician's workstation. Distortions as caused by the fiber optic taper assembly are minimal (mostly less than 2% and often less than 1%) and can be corrected in software. The light output of the external sensors is lower than expected. The proximity focussed image intensifiers show field emission points and black ("dead") spots of non-response of the photocathode and are virtually useless.

Advances in Magnetice Resonance Imaging (MRI) techniques have recently made MRI the imaging modality of choice for many applications of clinical imaging. MRI provides the diagnosing clinician a non-invasive method for obtaining soft tissue differentiation with sub-millimeter resolution. Clinical MRI techniques include 3-dimensional imaging, spectroscopic imaging, arterial angiography and cardiac imaging. One MRI technique which has recently gained popularity is a class of protocols known as variable/partial flip angle MRI. Partial flip angle MRI techniques are useful because of their ability to vary contrast between tissues and/or maintain a particular level of contrast with a reduction in acquisition time [1]. Variable flip angle techniques differ from conventional MRI protocols in that the initial RF excitation/rotation pulse is not constrained to a 90 degree rotation of the longitudinal magnetization. Instead, the initial excitation flip angle is calculated to provide improved contrast between two tissues and/or maximize the intensity of a particular tissue. For tissues with reduced TR/T1 ratios, variable flip angle techniques may also be used to increase the image signal to noise within a localized region.

To evaluate the loss of information in the process of digitizing radiographs, followed by reprinting these on film, a study was performed. Radiographs were taken of a 3M chest phantom with a random distribution of lucite cylinders, simulating pulmonary nodules, using two different dose techniques: a conventional technique, and an equalization technique (AMBER). All radiographs were digitized to a 2048*2048*8 matrix by means of a Konica KFDR-S laser scanner, and reprinted on film by means of a 3M laser printer. A traditional ROC study was performed, presenting the films to four radiologists. Results indicate no information loss for the conventional chest radiographs, but slight loss of nodule detectability in the equalized radiographs.

Film based radiography accounts for about 60% of the case load in community medical imaging departments. This includes chest, gastrointestinal, breast and general radiography. To date, most of the work has been devoted to the conversion of these procedures from film to digital image acquisition techniques so as to facilitate their integration into digital Picture Archive and Communication Systems (PACS). While most efforts in this area have been directed toward the conversion of chest radiography, in this report the authors describe a technique for digital radiography of the wrist using a 33 cm image intensifier (II).

Using an anamorphic imaging technique, the solid state image sensor can replace the vacuum pick-up tube in medical X-ray diagnostics, at least for the standard quality fluoroscopic application. A video camera is described in which, by optical compression, the circular output window of an image intensifier is imaged as an ellipse on the 3:4 image rectangle of a CCD sensor. The original shape is restored by electronics later. Information transfer is maximized this way: the total entrance field of the image intensifier is available, at the same time the maximum number of pixels of a high resolution CCD image sensor is used, there is also an increase of the sensor illuminance by 4/3 and the aliasing effects are minimized. Imaging descriptors such as Modulation transfer, noise generation and transfer are given in comparison with a Plumbicon-based camera, as well as shape transfer, luminance transfer and aliasing. A method is given to isolate the basic noise components of the sensor. A short description of the camera optics and electronics is given.

Previous studies [1-3] of the capabilities of a digital radiographic imager employing a scanning strip (slot) beam and a Kinestatic Charge Detector (KCD) have focussed on spatial and contrast resolution. This paper compares the temporal resolution of such a KCD system with that of a system employing a wide area beam and a rare-earth phosphor. The results are largely based on theoretical and computer simulations, including use of a Monte Carlo x-ray transport program, and indicate that for equal image SNR, the KCD yields significantly improved temporal resolution and lower doses than the rare-earth screen system. However, the KCD suffers from tube loading problems and long scan times for patient thicknesses approaching the equivalent of 30 cm of water.

Three commercially available heavy element filters as well as more conventional filters of aluminum and copper were evaluated to determine their relative merit for the reduction of patient exposure in radiographic imaging. Specifically the filters were evaluated to determine reduction in x-ray tube exposure output and changes in beam quality, degree of entrance exposure reduction provided, associated increase in x-ray tube loading, and alteration of image contrast. The results indicate that a modest increase in aluminum filtration, above that which is required to provide the minimum beam quality requirements as specified by the Federal Performance Standard, will yield entrance exposure reductions of approximately 25 percent with x-ray tube load increases of from 5 to 10 percent. Further reduction in entrance exposure, on the order of 40 to 60 percent with associated x-ray tube load increases of less than 40 percent, may be achieved above 75 kilovolts by use of a composite filter of aluminum and copper, and two of the commercial filters. The third commercial filter, although providing significantly reduced entrance exposure, requires an unacceptable increase in x-ray tube load to obtain satisfactory image optical density. No significant changes in image quality were observed in images of either anatomical phantoms or a quantitative test object for any of the filters investigated.

An inexpensive computer imaging system is being developed which is capable of accurately recovering the three-dimensional surface of the human body. This system uses biologically safe structured white light. In structured light, a uniform square grid pattern is projected onto the skin and an image of this pattern is recorded using a single camera. The squares in the camera image appear distorted due to the curvature of the skin. By locating the intersections of the grid stripes and matching them correctly to the projected grid pattern, the three-dimensional positions of points on the skin surface can be determined by triangulation. Triangulation is the method by which an illuminated point on the object surface may be located in space as the intersection of two lines: the illuminating ray from the projector and the line of sight determined by the location of the point in the camera image. This system has been applied to measure facial swelling and the surface area of burns. This paper summarizes the steps in processing an image to reconstruct a surface patch of a portion of the skin. We discuss the geometry of the imaging system and show how three-dimensional information can be recovered. Next, we describe the actual processing steps and algorithms used to locate the data used to reconstruct the patch. A unique feature of the system is the compensation for the reflectance of the skin. Experiments are discussed figures are presented illustrating the various processing steps. These preliminary results demonstrate the feasibility of the imaging system in accurately reconstructing a surface patch of the skin. Finally, we summarize the present state of our project and discuss work in progress and future plans.

The coronary cineangiograms are used principally to show the coronary circulation (arterial tree morphology, stenosis etc.). However, it has other interesting potentials in relation to cardiac dynamics. Indeed, the coronary arterial tree provides natural landmarks to appraise the epicardial motion with cardiac contraction. This paper presents a new approach to assess epicardial dynamics, locally, using a velocity field description obtained through processing of coronary cineangiographic sequences with an optical flow algorithm. The velocity fields can be visualized by assigning to each pixel of the angiogram the corresponding 2-D velocity vector. In addition, the method allows computation of epicardial translation, rotation and deformation as a function of time, by integrating interframe motion parameters computed from the velocity field sequence. The method is illustrated with a clinical example.

In this paper, according to the H-D curve of silver halide plate, we systematically analyse and explain the various phenomena which are produced by the zeroth and other orders diffraction images. The reversal conditions are given and some important conclusions are obtained. In our experiment, the density distributions of encoded plates are measured by using microdensitometer and corresponding distributions of diffraction intensity are recorded. The experimental results are consistent with theoretical analysis. This paper will provide foundations for making use of the characteristics of grating encoded plates.

The physical and mathematical problems of X-ray computed tomography (CT) imaging have been extensively studied during the past two decades. This paper presents a statistical description of X-ray CT imaging. The asymptotic Gaussianity of the pixel image generated by convolution algorithm is proved. The conditions for the two pixel images to be statistically independent and the conditions for a group of pixel images to be a spatial stationary random process and ergodic in mean and autocorrelation are derived. These properties provide the basis for the statistical image analysis of X-ray CT image, which is briefly described in the last section of this paper.

The application of DFT and FFT techniques provides a simpler detector alignment procedure in a fourth generation computed tomography system. A single scan of a circularly symmetric object can be used to provide the alignment data for this technique. Due to the symmetry involved, filtered autoconvolution employing an FFT technique similar to that used in CT reconstruction generates the locus of the perceived detector positions (known as the alignment curve). The fundamental components of this curve are seperated using a DFT to generate the actual detector alignment correction values. Applying these Fourier techniques to the CT alignment problem has simplified the calibration procedure necessary. Other applications having accurate positioning requirements may also benefit from these techniques.