X-RAY DETECTOR FOR A

LOW DOSAGE SCANNING BEAM

DIGITAL X-RAY IMAGING SYSTEM

i. TECHNICAL FIELD OF THE INVENTION

The present invention relates to diagnostic x-ray imaging equipment. More particularly, the present invention relates to a real time scanning beam digital x-ray imaging system having improved resolution and reduced x-ray emissions provided by incoφoration of a multi-apertured collimation grid and a segmented x-ray detector array.

2, BACKGROUND ART

Real time x-ray imaging is increasingly being required by medical procedures as therapeutic technologies advance. For example, many electro-physiologic procedures in cardiology, peripheral vascular procedures, urological procedures, and orthopedic procedures rely on real time x-ray imaging.

Unfortunately, current clinical real time x-ray equipment produces high levels of x-ray exposure to both patients and attending staff. The United States Food and Drug Administration (F.D.A.) has reported anecdotal evidence of acute radiation sickness in patients, and concern among physicians of excessive occupational exposure. (Radiological Health Bulletin, Vol. XXVI, No. 8, August 1992).

A number of real time x-ray imaging systems are known in the prior art. These include fluoroscope-based systems where x-rays are projected into an object to be x- rayed and shadows caused by relatively x-ray opaque matter within the object are displayed on the fluoroscope located on the opposite side of the object from the x-ray source. Scanning x-ray tubes have been known in conjunction with the fluoroscopy art since at least the early 1950s. Moon, Amplifying and Intensifying the Fluoroscopic Image bv Means of a Scanning X-rav Tube. Science, October 6, 1950, pp. 389-395.

Scanning beam digital x-ray imaging systems are also well known in the art. In such systems an x-ray tube is employed to generate X-ray radiation. Within the x-ray tube, an electron beam is generated and focussed upon a small spot on the relatively large anode (transmission target) of the tube, inducing x-ray radiation emission from that spot. The electron beam is deflected (electromagnetically or electrostatically) in a raster scan pattern over the entire anode. A small x-ray detector is placed at a distance from the anode of the x-ray tube. The detector converts x-rays which strike it into an electrical signal in proportion to the detected x-ray flux. When an object is placed between the x-ray tube and the detector, x-rays are attenuated and scattered by the object in proportion to the x-ray density of the object. While the x-ray tube is in the scanning mode, the signal from the detector is modulated in proportion to the x-ray density of the object.

Examples of prior art scanning beam digital x-ray systems include those described in United States Patent No. 3,949,229 to Albert; United States Patent No. 4,032,787 to Albert; United States Patent No. 4,057,745 to Albert; United States

Patent No. 4,144,457 to Albert; United States Patent No. 4,149,076 to Albert; United States Patent No. 4,196,351 to Albert; United States Patent No. 4,259,582 to Albert; United States Patent No. 4,259,583 to Albert; United States Patent No. 4,288,697 to Albert; United States Patent No. 4,321,473 to Albert; United States Patent No. 4,323,779 to Albert; United States Patent No. 4,465,540 to Albert; United States Patent No. 4,519,092 to Albert; and United States Patent No. 4,730,350 to Albert. In a typical prior art embodiment of a scanning beam digital x-ray system, an output signal from the detector is applied to the z-axis (luminance) input of a video monitor. This signal modulates the brightness of the viewing screen. The x and y inputs to the video monitor are derived from the same signal that effects deflection of the x-ray signal of the x-ray tube. Therefore, the luminance of a point on the viewing screen is inversely proportional to the absorption of x-rays passing from the source, through the object, to the detector.

Medical x-ray systems are operated at the lowest possible x-ray dosage level that is consistent with the resolution requirements for the procedure. Accordingly, both dose and resolution are limited by the signal to noise ratio.

The term "low dosage" used herein refers to an x-ray exposure level during operation of less than or equal to about 2.0 R/min (Roentgens/min) at the entrance to the patient.

Time and area distributions of x-ray photons follow a Poisson distribution and have an associated randomness which is unavoidable. The randomness is expressed as the standard deviation of the mean flux, and equals its square root. The signal to noise ratio of an x-ray image under these conditions is therefore equal to the mean flux divided by the square root of the mean flux. I.e., for a mean flux of 100 photons, the noise is +/- 10 photons, and the signal to noise ratio is 10.

Accordingly, the spatial resolution and the signal to noise ratio of x-ray images formed by scanning x-ray imaging systems are dependent, to a large extent, upon the size of the sensitive area of the detector. As the detector aperture is increased in area, more of the diverging rays are detected, effective sensitivity increases and the signal to noise ratio is improved. At the same time, however, the larger detector aperture reduces attainable spatial resolution as the "pixel" size (measured at the plane of the object to be imaged) becomes larger. This is necessarily so because most objects to be imaged in medical applications (e.g., structures internal to the human body) are some distance from the x-ray source. In the prior art, therefore, the detector aperture size must be selected so as to effect a compromise between resolution and sensitivity, it not being possible to maximize both resolution and sensitivity simultaneously. In medical imaging applications, patient dosage, frame rate (the number of times per second that the object is scanned and the image refreshed), and resolution of the image of the object are key parameters. A high x-ray flux may easily yield high resolution and a high frame rate, yet result in an unacceptabiy high x-ray dosage to

the patient and attending staff. Similarly, low dosage may be achieved at the cost of an invisible image or an inadequate refresh rate. A successful medical imaging system must provide low dosage, high resolution and an adequate refresh rate of up to at least about 15 images per second - all at the same time. Therefore, systems such as the prior art scanning beam digital x-ray imaging system described above will not work with diagnostic medical procedures where exposure times are relatively long and where, as is always the case with real patients, the x-ray dose received by the patient must be kept to a minimum.

It is therefore an object of the present invention to provide a scanning beam digital x-ray imaging system capable of use in medical diagnostic procedures undertaken on living human patients.

It is also an object of the present invention to provide a scanning beam digital x-ray imaging system which provides high resolution images at adequate frame rates while minimally exposing the object under investigation to x-ray radiation. It is a further object of the present invention to provide a scanning beam digital x-ray imaging system having improved resolution at a distance from the plane of the source of the x-rays while maintaining decreased x-ray flux levels.

These and many other objects and advantages of the present invention will become apparent to those of ordinary skill in the art from a consideration of the drawings and ensuing description of the invention.

DISCLOSURE OF THE INVENTION A scanning beam digital x-ray imaging system ("SBDX") according to the present invention includes an x-ray tube having an electron beam source and a target anode. Circuitry is provided for focussing the beam and directing or scanning the beam across the target anode in a predetermined pattern. For example, the predetermined pattern may be a raster scan pattern, a serpentine or "S" shaped pattern, a spiral pattern, a random pattern, a gaussian distribution pattern centered on

a predetermined point of the target anode, or such other pattern as may be useful to the task at hand.

A collimating element, preferably in the form of a grid, may be interposed between the x-ray source and an object to be x-rayed. The collimating element may, for example, comprise a round metal plate having a diameter of about 25.4 cm (10 in) and including an array of apertures numbering 500 by 500 at the center row and column of the collimating element. The collimating element is preferably placed immediately in front of the emitting face of the x-ray tube. Other configurations of collimating element could also be used. In one preferred embodiment of the present invention each of the apertures in the collimating element is constructed so that it is directed toward (or points at) a detection point on a plane located a selected distance from the collimating element. That distance is selected to allow placing the object to be x-rayed between the collimating element and the detection point. The function of the collimating element is to form thin beams of x-rays, arrayed like pixels, and all directed from a point on the anode of the x-ray tube toward the detector.

A segmented detector array containing an array of detector elements (preferably an area array such as a DET _X by DET _Y rectangle or square, or, more preferably, a relatively round array) is centered at the detection point. The detector array preferably comprises a plurality of densely packed x-ray detectors. Such an array can be designed, positioned and applied according to the present invention in a manner that yields high sensitivity without loss of resolution, resulting in an x-ray system having a resolution comparable to or better than that of prior art x-ray units at a dosage at least an order of magnitude less than that of prior art x-ray systems. This feature of the present invention has important implications in medical and other fields. Exposure to patients and attending medical staff involved in current procedures will be reduced. Procedures now impossible due to the radiation exposure risk will become possible.

The output of the detector array is an intensity value for each element of the

detector array at each point in time that the x-ray beam is emitted through an aperture in the grid. Because each aperture is located at a different point in space relative to the object under investigation and the detector array, a different output will be available from the detector array for each aperture that the x-ray beam travels through. The detector array outputs may be converted into an image in a number of ways. One method is to perform a simple convolution upon the array output, i.e., summing the intensity values of the array elements corresponding to each aperture scanned, and then normalizing. The output array could then be used to drive a video or other display. More preferred are the multi-image convolution and the multi-output convolution methods, described below, which provide enhanced visual output. The SBDX imaging system is also capable of stereo imaging where a collimation grid having two groups of apertures is used. In this case, one group of apertures is constructed to point to a first detection point where a first segmented detector is located and a second group of apertures is constructed to point to a second detection point where a second segmented detector is located. By constructing two images from the two segmented detectors and using conventional stereoscopic display methods, a stereo image may be produced.

The SBDX imaging system is also capable of highlighted imaging of materials which exhibit different x-ray transmissivities at different x-ray photon energies. Accordingly, for example, microcalcification, a precursor of breast cancer, may be imaged. By constructing the grid and/or anode to emit two or more groups of beams of x-rays each having different x-ray energy spectra, and directing each group to the detector array (plural detector arrays could also be used), the difference of transmissivities of the object under investigation at the various x-ray photon energies can be turned into an image, thus highlighting only those materials within the object under investigation which exhibit differential x-ray transmissivity. Optimized for the detection of calcium, for example, such an imaging system is a powerful tool for the early detection of breast cancer and other anomalies.

Utilizing a segmented array which intercepts the entire collimated x-ray beam and image processing the array output provides maximum sensitivity without sacrificing the resolution provided by using small surface area detectors. A non- segmented detector of the same size as the segmented array would provide the same sensitivity at lower resolution.

Additionally, sub-sampling techniques may be used for processing the data from the array detector which reduce the complexity of the system and the required processing speed and energy consumption while providing virtually the same image quality. The system described herein is capable of being used in conjunction with the

"Catheter Including An X-ray Sensitive Optical-Sensor Locating Device" described in U.S. patent application serial number 08/008,455 (CAM-003) filed January 25, 1993 and which is hereby incorporated herein by reference. U.S. patent application serial number 08/008,455 is owned by the assignee herein.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is a diagram showing the basic components of a low dosage scanning beam digital x-ray imaging system.

Fig. 2 is a diagram showing the distribution of x-rays from an SBDX system in the absence of a collimation grid. Fig. 3 is a magnified view of the grid and anode of an x-ray tube for a low dosage scanning beam digital x-ray imaging system.

Figs. 3A, 3B and 3C are partial cross-sections of collimation grids useful in the inventive device.

Fig. 4 is a diagram of an x-ray tube for a low dosage scanning beam digital x- ray imaging system.

Fig. 5 is a cross sectional diagram showing the fabrication of an x-ray tube for a low dosage scanning beam digital x-ray imaging system.

Fig. 6 is a diagram of a stereoscopic scanning beam digital x-ray imaging system.

Fig. 7A is a diagram of an apertured x-ray source interacting with a simple non- segmented detector. Fig. 7B is a diagram of x-rays from a single aperture of an apertured x-ray source interacting with a segmented detector array.

Fig. 7C is a diagram of x-rays from a number of apertures of an apertured x-ray source interacting with a segmented detector array.

Fig. 7D is a diagram of x-rays from two apertures of an x-ray collimation grid interacting with an object under investigation and then a segmented detector array.

Fig. 8 is a diagram of the exposed surface of a 5 x 5 detector array for a low dosage scanning beam digital x-ray imaging system.

Fig. 9 is a diagram of a 5 x 5 detector array for a low dosage scanning beam digital x-ray imaging system. Fig. 9A is a diagram depicting a scintillator element according to one preferred embodiment of the present invention.

Fig. 10 is a diagram of a detector element for a low dosage scanning beam digital x-ray imaging system.

Fig. 11 is a diagram showing an array of pencil-type detector elements for a non-planar detector array.

Fig. 12 is a diagram of a 3 x 3 detector array for a low dosage scanning beam digital x-ray imaging system.

Fig. 13 is a diagram showing the basic components of a low dosage scanning beam digital x-ray imaging system utilizing negative feedback to control x-ray flux. Fig. 14 is a perspective diagram showing the grid seal assembly.

Fig. 15 is a diagram of the layout of a 96 element detector array according to a presently preferred embodiment of the invention.

Fig. 16 is a diagram showing the interaction of the collimation grid and the

detector array.

Fig. 17 is a diagram showing a presently preferred embodiment of the detector assembly.

MODES FOR CARRYING OUT THE INVENTION Those of ordinary skill in the art will realize that the following description of the present invention is illustrative only and not in any way limiting. Other embodiments of the invention will readily suggest themselves to such skilled persons.

System Overview

Turning to Fig. 1 , a scanning beam digital x-ray imaging system according to a preferred embodiment of the present invention is diagrammed. A scanning x-ray tube 10 is used as the x-ray source. An approximately -100kV to -120kV power supply is used to power x-ray tube 10 as is well known in the art. A 100 kV power supply will provide a spectrum of x-rays ranging to 100keV. As used herein, 100kV x-rays refers to this spectrum. X-ray tube 10 includes a deflection coil 20 under the control of scan generator 30 as is well known in the art. An electron beam 40 generated within x-ray tube 10 is scanned across a grounded anode 50 within x-ray tube 10 in a predetermined pattern. For example, the predetermined pattern may be a raster scan pattern, a seφentine or "S" shaped pattern, a spiral pattern, a random pattern, a gaussian distribution pattern centered on a predetermined point of the target anode, or such other pattern as may be useful to the task at hand. Presently preferred is the seφentine or "S" shaped pattern which eliminates the need in a raster scan pattern for "fly back."

As electron beam 40 strikes anode 50 at point 60, a cascade of x-rays 70 is emitted and travels outside of x-ray tube 10 toward the object 80 to be investigated with the x-ray. To optimize system performance, a cone of x-ray photons must be generated that will diverge in a manner that will just cover the detector array 110.

This is preferably accomplished by placing a collimation grid between the anode of the scanning x-ray tube and the detector. Collimation grid 90 is therefore disposed between object 80 and x-ray tube 10. Collimation grid 90 is designed to permit only those x-rays 100 which are directed toward detector 110 to pass through it. Collimation grid 90 does not move with respect to detector array 110 while the system is in operation. Thus, as electron beam 40 is scanned across anode 50, at any given moment, there can only be a single x-ray beam 100 passing from anode 50 to detector array 110.

Fig. 2 depicts the distribution of x-rays in the absence of a collimation grid. The output of detector array 110 is then processed and may be displayed on monitor 120 as a luminance value at an x,y location on monitor 120 corresponding to the x,y location on anode 50. This may be accomplished by using the same scan generator to drive the x,y position of electron beam 40 and the position of the electron beam within the video monitor 120. Alternatively, image processing techniques can be used to produce a computer driven image on an appropriate display or photographic medium.

The inventive system disclosed herein is a low dosage system in that it typically doses the patient at a rate of about 0.15 R/min with a 15 frame/sec refresh rate to about 0.33 R/min with a 30 frame/sec refresh rate - again - measured at the entrance to the patient. Whole body exposure with a 30 frame/sec refresh rate with this system will be about 0.50 R/min. Thus the practical range of dosage at the entrance to the patient for operation of the invention disclosed herein is within the range of 0.15 R/min to 2.00 R/min.

The X-Rav Tube Fig. 3 depicts a magnified view of the grid and anode structure. Anode 50 is preferably fabricated of a target layer of a material having good vacuum characteristics and the ability to withstand high heat and electron bombardment formed upon a

beryllium anode support 130. Aluminum or other relatively x-ray transparent materials could be used for the anode support 130 as well. The presently preferred construction of the target layer is, in order of preference: (1) a first layer of niobium approximately 1 micron thick sputter-deposited upon the anode support to which is then sputter- deposited a second layer of tantalum approximately 5 microns thick (this structure is preferred because the niobium has a thermal coefficient of expansion intermediate to the coefficients of thermal expansion of beryllium (the anode support 130) and tantalum, thus reducing or preventing microcracking due to thermal cycling of the target anode between the ON and OFF conditions of the tube; (2) a sputter-deposited layer of tantalum approximately 5 microns thick; (3) a sputter-deposited layer of tungsten-rhenium approximately 5 microns thick; and (4) a sputter-deposited layer of tungsten approximately 5 to 7 microns thick. Tantalum, tungsten and tungsten- rhenium are preferred for anode 50 because, having relatively high atomic numbers and densities, they readily emit x-rays when irradiated by an electron beam. Tungsten's high melting point of 3370°C and good vacuum characteristics make it suitable for the high temperature and hard vacuum conditions within the tube. Tantalum and tungsten-rhenium have similar characteristics as known to those of skill in the art. The thicknesses of the anode layers are selected so that they are approximately equivalent to the distance necessary to efficiently convert 100 kV electrons to x-rays. Beryllium is preferred for anode support 130 because it is strong and does not significantly attenuate or scatter the x-rays emitted from anode 50. The thickness of beryllium anode support 130 is preferably about 0.5 cm. Anode support 130 should be as thin as possible subject to the physical constraint that it must be strong enough to withstand the pressure gradient of one atmosphere across it. Collimation grid 90 preferably consists of an array of apertures 140, each of which according to one preferred embodiment of the present invention is oriented or pointed toward detector array 110. That is to say that the apertures within the collimation grid 90 are not parallel to each other and for use with, for example, a chest

x-ray application, may have an angle with the frontal plane 260 of the collimation grid 90 of between 0° at the center of the collimation grid 90 to as much as 20° at the edge of the grid 90. In use of the invention with a mammogram application, the grid 90 may be fabricated so that the apertures form an angle with the frontal plane ranging to 45° at the edge of the grid. The number of apertures 140 in collimation grid 90 may correspond to the number of pixels, e.g., 500 by 500 to 1024 by 1024 at the center of preferably round collimation grid 90 and, in part, determines the system resolution. Alternatively, fewer apertures than pixels may be used in conjunction with the technique of sub-sampling discussed below. The thickness of grid 90 and size of apertures 140 are determined by the distance of the detector array 110 from x-ray tube 10 (here, preferably 91.4 cm (36 in)), the need to attenuate all x-rays not heading to the detector, and the size of detector elements 160 of detector array 110 (not shown in this figure). Although it is not critical for this invention, apertures 140, as viewed from frontal plane 260, are preferably laid out in a rectangular row and column pattern having a circular boundary 25.4 cm (10 in) in diameter. The aperture array may be of any convenient layout which may be correlated using the detection and convolution techniques outlined below to resolve the image of object 80. This aperture array is called the "circular active area". At the center of the circular active area the aperture count is preferably, according to one preferred embodiment of the present invention, 500 by 500. The non-aperture portion 150 of collimation grid 90 is designed to absorb errant x-rays so that they do not illuminate object 80. This is accomplished by fabricating the grid so that x-rays striking the non-aperture portion 150 will see at least ten times the "1/2 value" (the amount of material necessary to attenuate 1/2 the x-rays striking it at the system energy, here, 100 keV). Errant x-rays would provide the object and attending staff with x-ray dosage but contribute no meaningful information to the image. Collimation grid 90, as shown in FIGS. 3A and 3B, may be formed from a number of sheets 143, 144 of x-ray absorbing materials having apertures 140 therethrough to allow the x-ray beam 100 to pass through to the

detector. The collimation grid 90 is preferably fabricated of 50 thin sheets of 0.0254 cm (0.010 in) thick molybdenum which are stacked and held together. Molybdenum is preferred because it readily absorbs x-rays so that x-rays generated by x-ray tube 10 which are not directed to detector 110 will be stopped before they impinge, uselessly and potentially harmfully, upon object 80, which, of course, may be a human patient. Lead or similar x-ray dense materials could also be used.

The apertures 140 of collimation grid 90 are preferably square in cross section in order to obtain the maximum packing density and be compatible with the preferred square shape of the detector array elements 160. Other shapes could also be used, particularly hexagons. The square apertures 140 are preferably 0.0381 cm (0.015 in) by 0.0381 cm in dimension which yields a cross sectional area that is about 1/100 the cross sectional area of conventional collimators used with fluoroscopes. Because of this tighter collimation, a smaller beam width for x-ray beam 100 is achieved. This means that the cross sectional area of the face of the detector may be correspondingly much smaller than in conventional systems. As a result, x-rays scattered at the object miss the detector and do not fog the image as they do in conventional systems which utilize relatively large surface area detectors.

A preferred method for fabricating the collimation grid 90 is by photo-chemical milling or etching. Photo-chemical milling is presently preferred because it is cost effective and accurate. According to this method, a set of 50 photo masks is created to etch holes or interstices into 50 thin sheets of 0.0254 cm (0.010 in) thick material. The etched sheets are then stacked and aligned and held together to form a grid assembly having a plurality of stepped apertures, each of a predetermined angular relationship with respect to the sheets. FIG. 3A shows a variation of the inventive collimation grid 90. This variation includes a number of x-ray absorbing sheets 143 having individual apertures with a constant cross-section (however, the cross-section need not be constant). The resulting aperture 14 is stepped, as shown, but allows the x-ray beam 100 to pass through to the detector array 110. The variation shown in

FIG. 3B is quite similar to that shown in FIG. 3A except that the individual apertures formed in x-ray absorbing sheets 144 are themselves stepped. These stepped apertures may be made by milling or chemical etching from each side of sheet 144 with a slight offset so as to result in the configuration shown as would be obvious to one of ordinary skill in the art. The FIG. 3B configuration is highly desirable because less x-ray energy need be absorbed within the stepped apertures 140 of collimation grid 90 and consequently, the x-ray flux at the edge of the x-ray beam 100 is not attenuated as much as in the variation shown in FIG. 3A.

One presently preferred method for holding the etched sheets that grid 90 is formed from is shown in Fig. 14. Etched sheets 91 (preferably 50) are each provided with alignment holes or alignment apertures 94. Alignment pegs 95 are placed in each alignment aperture 94 to align the etched sheets 91. The assembly of sheets 91 and pegs 95 is then placed in aluminum ring 359. Aluminum ring 359 is provided with a vacuum port 370 which may be sealed with pinch off 375. Aluminum sheet 365 which is 0.1 cm in thickness is then bonded and sealed with a vacuum adhesive to upper surface 380 of ring 359. Aluminum sheet 360 is similarly bonded to a lower surface 385 of ring 359. A partial vacuum is then pulled through port 370 and the port 370 is then sealed at pinch off 375 as is well known in the art. In this manner, relatively x-ray transparent aluminum sheets 360, 365 provide a clamping action tending to hold etched sheets 91 together and in alignment as grid assembly 90.

The apertures 140 furthest from the center of grid 90 have a stepped surface and are preferably square in cross section. X-rays are generally unaffected by the roughness of the channels due to the stepped surface, and even if they are scattered, they will not measurably affect the resultant beam. The material used for the collimation grid 90 as discussed above can be molybdenum, brass, lead, or copper with molybdenum presently preferred. Presently preferred tolerances for the positions of the holes are +/- 0.00127 cm (0.0005 in) center to center without cumulative error and for the hole sizes are +/- 0.00254 cm (0.001 in).

Alternative methods for fabricating collimation grid 90 which could be used include electron beam machining, drilling or mini-machining, and laser drilling. Drilling and laser drilling have the drawback that they generate round rather than square holes. Circular apertures are not presently preferred although they should work as well.

More details of the preferred scanning x-ray tube 10 are shown at Figs. 4 and 5. Electron gun 161 is located opposite the face of x-ray tube 10 and is operated at a potential of up to about -100kV to -120kV Grounded anode 50 is located at the face of the tube and an electron beam 40 travels between electron gun 161 and anode 50. A grounded electron aperture plate 162 is located near electron gun 161 and includes an aperture 163 at its center for electron beam 40 to pass through. A magnetic focus lens 164 and deflection coil 20 position the beam spot on anode 50 utilizing dynamic focussing as is well known in the art. The tube is fabricated to have a 25.4 cm (10 in) diameter circular active area in which electron beam 40 may intersect anode 50 with the electron beam deflected up to about 30° at the extremities of the circular active area. When the beam is not being "fired" through a particular aperture it is preferably left off, resulting in a power savings of up to about 25%.

Turning to Fig. 5, a cross sectional view of the front portion of a suitable x-ray tube 10 is depicted. The interior of the x-ray tube 340, maintained at a vacuum is rearward of anode 50. Anode 50 is a coating of anode material as discussed above. Forward of anode 50 is beryllium anode support 130 which is 0.5 cm thick. Forward of beryllium anode support 130 is cooling jacket 350 which is preferably 0.4 cm thick and may be adapted to carry water or forced air. Aluminum grid supports 360, 365 are each 0.1 cm thick and help support collimation grid 90 which is preferably 1.27 cm (0.5 in) thick.

When the x-ray tube 340 is in use, no more than one aperture 140 of collimation grid 90 will be passing substantial amounts of x-rays at any given instant. According to one preferred embodiment, the electron beam 40 may be shut off when

the electron beam 40 is not positioned directly in front of an aperture 140. Thus the x- ray tube may be operated effectively in a scanned-pulsed mode to reduce power consumption and wear and tear on the target anode 50.

Stereoscopic X-rav maging Turning now to Fig. 6, according to another preferred embodiment of the present invention, a grid having more than one focal point may be provided so that stereoscopic x-ray images may be obtained. If, for example, every other row of apertures in grid 90 were pointed at focal point F1 (92) and the remaining apertures were pointed at focal point F2 (93), by placing a first sensor array at F1 (92) and a second sensor array at F2 (93) it is possible to scan the apertures in a raster or seφentine pattern and thereby create a "line" of data for the first sensor array, then a line of data for the second sensor array. Repeating this, it is possible to build up two complete images, as seen from two distinct points in space, F1 and F2, and thereby display them with conventional stereoscopic imaging display systems to provide a stereoscopic x-ray image. Turning now to Fig. 3C it is shown how one may construct such a stereoscopic collimation grid out of layers 144 of x-ray absoφtive material. In this case the apertures 140A, 140B may in fact be shaped like a "V ^* as shown providing separate paths along the "legs" of the "V" for x-ray beams 100A, 100B. There is no requirement, however, that apertures 140A, 140B be joined as shown, but an advantage of the "V'-shaped aperture where the x-rays enter at the apex of the "V is that both detectors will be illuminated simultaneously, the "V' acting as a signal splitter with some of the x-rays going to F1 and some to F2. This halves the power required for the beam and deflection currents. The price is a small but acceptable increase in scattering and thus fogging of the image.

The Array Detector

To achieve resolutions of several lines per millimeter at the object plane, as are

required in some medical applications, the spatial resolution limit is in large part determined by the size of the detector. This is because, with today's x-ray tube technology, it is not feasible either to produce the very high power levels that would be required in order to obtain sufficiently intense highly directionalized x-ray emissions or to develop the associated x-ray directionalizing means.

When the detector is made smaller than the area intersecting the cone of emitted x-rays, a large proportion of the x-rays emitted by source 50 miss detector 250, as shown in Fig. 7A. This is, in fact, how industrial scanning beam digital x-ray inspection systems are designed, where dose is usually not an issue. As a consequence, the dose is increased in order to maintain the desired resolution.

Accordingly, resolution improves with the use of smaller detectors, but x-ray dose is minimized when the area of the detector equals or exceeds the area defined by the cone of emitted x-rays intersecting detector plane 270.

Resolution of a scanning x-ray imaging system is determined by the cross sectional area of a detector element projected onto the object plane 280 (the plane peφendicuiar to a line between the center of anode 50 and the center of detector 110 in which object 80 is located). Thus, if a large area detector is subdivided into smaller array elements, e.g., as shown by the front view of the detector array in Fig. 8, the large capture area of the aggregate detector is maintained, while simultaneously retaining an image resolution that is proportional to the size of an individual small detector element 160.

The resolution defined by the individual detector elements 160 is maintained by distributing and summing the readings from the individual elements into a memory buffer in which each address, i.e., pixel, corresponds to a specific location in the object plane 280. As the x-ray beam 100 is moved discretely across collimation grid 90, which is positioned in front of the x-ray emitting anode 50, the address, to which the output of a given detector element is added, changes. The imaging geometry is shown in Fig. 7B and 7C. In Fig. 7B a single beam position is shown along with how

it is divided among 5 pixels. In Fig. 7C the sequential beam positions are shown along with how they are added together within a single pixel.

In other words, the signal for each of the detector elements is stored in an image buffer, at a memory address that corresponds to a very small specific region in the object plane 280, i.e., a single pixel. Accordingly, the memory storage address for each detector element changes with the location of the scanning x-ray beam in an ordered fashion such that each pixel in memory contains the sum of the radiation passing through a specific spot in the object plane 280. In this way the resolution of the system is determined by the size of a single detector element, while the sensitivity of the system is optimized, since virtually all of the x-rays reaching the detector plane 270 are recorded.

An additional benefit of this array detector imaging geometry is that the object plane 280 is narrowly defined. Structures lying in front of or behind it will be blurred (out of focus). X-rays from a first aperture 141 and a second aperture 142 are depicted in Fig. 7D passing through an object plane 280 a distance SO from apertures

141, 142 and passing through a plane 281 twice distance SO from apertures 141,

142. As can be readily seen, the resolution degrades to about 1/2 that available at SO at the distance twice SO. This feature provides for improved localization and visualization of detailed structures in the plane of interest 280, while providing an adequate depth of field that may be modified by the system geometry.

The array of the presently preferred embodiment is a 96 element pseudo-round array of square detector elements 0.135 cm on a side disposed within a circle of diameter about 1.93 cm (0.72 in). It need not be so large and could be three or more detectors disposed so that not all are in a line within a circle of radius equal to the length of a side of one of the detectors, here 0.135 cm.

The X-rav Detectors

Conventional image intensifier technology has basic constraints that limit a

system's sensitivity. The thickness at which the scintillator material can be applied is limited by its optical transmission properties. Typically, it is made thick enough to capture about 50 percent of the incident x-ray photons. Of the emitted light photons, only about half reach the photocathode. At the photocathode, only about 10 percent of the incident light photons produce photoelectrons. Thus, only about 2.5 percent (.5 x .5 x .1) of the incident x-ray photon energy is conserved in an image intensifier system. In addition to this limited conversion efficiency, light photons are scattered laterally by the scintillator material and create haze that reduces the system's resolving power at a given dose level. One of the primary objects of the present invention is to provide an SBDX imaging system which will ensure that the subject under examination is exposed to the lowest possible level of x-rays commensurate with achieving image quality adequate to meet the requirements of the procedure being performed. This means that the system used to detect the x-ray photons emerging from the subject must have the highest possible photon to electrical signal conversion efficiency. In order to achieve this, the material used for the detector must have a length in the direction in which the photons travel that is sufficient to ensure that no photons emerge from the end farthest away from the incident x-rays, i.e., the photon energy must be adequately dissipated in the material in order to maximize the output of the detector. There are several types of detectors which could be used in the presently described SBDX system. That which is currently preferred is the scintillator in which x-ray photon energy is converted to visible light energy and the light intensity is then converted to an electrical signal by means of a photomultiplier, photo diode, CCD or similar device. Because each pixel in the SBDX image must be generated in a very short time period, about 140 nanoseconds, the scintillator material must have a fast response and a minimum afterglow time. Afterglow is the phenomenon wherein the scintillator continues to emit light after the stimulating incident x-radiation has ceased. Plastic scintillators, such as organic loaded polystyrene, are suitable in that they have the required fast response

characteristics but they have a relatively small x-ray photon interaction cross section so that their linear x-ray absoφtion coefficients are also small in value. The consequence is that a considerable thickness is required to stop all the x-ray photons. For 100kV x-rays, as presently preferred, a typical plastic scintillator must be about 28 cm (11 in) thick to capture 99% of the incident x-rays. More preferred at present (and in order of preference) are: (1) YSO cerium doped (yttrium oxy-orthosilicate, available from Airtron (Litton) of Charlotte, NC); (2) LSO cerium doped (lutetium oxy- orthosilicate, available from Schlumberger, Inc.); and (3) BGO (bismuth germanate, available from Rexon Components, Inc. of Beachwood, OH). YSO and LSO are advantageous in that they may be used at room temperature. BGO must be heated to about 100° C in order to achieve a suitable light output decay period of the order of 50 nanoseconds. These scintillator materials need not be as long as the plastic scintillator and are effective at a length of 0.10 cm.

According to a presently preferred embodiment of the present invention, the SBDX array detector 110 comprises a 12 by 12 pseudo-round array of 96 densely packed discrete x-ray detectors 160 spaced at a distance of 91.4 cm (36 in) from the x-ray source 50. (A 5 by 5 and a 3 by 3 array are also contemplated as is a non- square array having square detectors filling a circle about the x-ray target: see, e.g., TABLE I below). The geometry of the collimation grid aspect ratio, its spacing from x- ray source 50 and the size of x-ray source 50 gives a square section pyramid of x- rays with a 1.46° total included angle which is about 2.23 cm (0.9 in) across at the entrance to detector array 110. Each individual scintillator 170 must therefore be about 0.152 cm (0.06 in) center-to-center within the detector plane. If scintillator 170 has parallel sides, x-rays entering near to the edges will not be able to travel the requisite distance without striking the scintillator walls. These x-rays may therefore pass through to the adjacent scintillator if it is not shielded causing it to generate an output seemingly from the wrong spatial position in the subject with consequent degradation of the image quality. As shown in FIG. 9A, to avoid this effect, the

individual scintillators according to one preferred embodiment of the invention are preferably tapered so that their bounding faces 173 have an included angle α equal to that of the most peripheral of the incident x-rays 100'. This is particularly useful with long plastic scintillators. In the preferred example cited above, each scintillator 170 would thus preferably be a frustrum of a pyramid 28 cm long, with an entrance face (172) 0.285 cm across and a face at the photodetector end 174 which is 0.37 cm across. The whole bundle of 81 detectors therefore has multifaceted ends, each facet being tangential to the surface of a sphere centered at x-ray source 50.

A further improvement in detection efficiency for scintillators may be achieved by tapering the scintillator at an angle that is greater than that of incident x-ray beam 100. Photoelectrons and scattered x-rays produced by interaction between the incident x-ray and scintillator atoms near to the edges of the scintillator may be lost to the shielding material which preferably separates adjacent scintillators. These lost photoelectrons will not produce any light so they will not contribute to the light output from the scintillator. Their loss therefore reduces the efficiency of the scintillator. The maximum distance travelled by a photoelectron depends upon its energy and on the material in which it is travelling. For 100kV x-rays interacting with the atoms in a plastic scintillator, there will be no photoelectrons which can travel a distance greater than about 0.01 cm. If the scintillator pyramid frustrum is made with an included angle greater than that of x-ray beam 100 so that its dimensions become greater than the beam envelope by 2 x 0.01 cm, in a distance short compared with the detector length (28 cm) then the efficiency reduction due to the lost photoelectrons will be minimal. In this case, the center of the sphere tangential to the facets would no longer coincide with x-ray source 50 but would be nearer to detector array 110. Scattered photons will travel greater distances than the photoelectrons; so, to prevent these from escaping to an adjacent scintillator the rate of taper of the scintillator pyramid may be made greater than is needed for complete photoelectron capture in order to maximize the scattered photon capture.

Referring now to Fig. 9, according to a presently preferred embodiment of the present invention, in contact with each scintillator element 170 is a light pipe or fiber optic cable 180 which optically couples each scintillator element 170 with a corresponding photomultiplier tube 190 or solid state detector. Alternatively scintillators 170 may be located in close physical proximity to appropriate photodetectors.

Fig. 10 shows a preferred configuration of a detector element 160. An x-ray opaque sheet 200 with apertures 210 corresponding to each detector element 160 is disposed in front of detector array 110. Each detector element 160 is enclosed in a light tight enclosure 220 which may also be x-ray opaque. A light blocking window 230, preferably made of thin aluminum sheet is located at the front of light tight enclosure 220. Light blocking window 230 is x-ray transparent. Within light tight enclosure 220 is a scintillator element 170 in close proximity to a photomultiplier tube 190 which is electrically connected to a pre-amplifier 240. Preferably the analog signal from the pre-amplifier 240 is converted to a digital signal in a conventional manner for further processing.

Alternatively, scintillators could be placed in direct or close contact with an array of photo diodes, photo transistors or charge coupled devices (CCDs) to achieve a more rugged and compact detector. Where solid state devices, particularly CCDs, are used, cooling, such as with a Peltier-type cooler, or the like, may be employed to increase the signal to noise ratio of the device.

Alternatively, the scintillator array could be placed in direct or close contact with one or more position sensitive photomultiplier tubes which provide an output signal which identifies the position coordinates of the light source as well as its amplitude. In another preferred embodiment, the sensor array may comprise a collection of pencil-type detectors 285 arrayed as shown, for example, in Fig. 11. In Fig. 11 tapered scintillators 290 are arrayed in the path of x-ray beam 100 so that the scintillator corresponding to a particular cross-sectional area of beam 100 will fully

absorb x-rays within that cross-sectional area. Photo multiplier tubes 300 are located in close physical proximity to scintillators 290 so that an electrical signal will be generated in response to the absoφtion of x-rays by scintillators 290. Solid state devices could also be used in place of photo multiplier tubes 300. According to a presently preferred embodiment of the present invention, the scintillators are coated along their lengths and the input face with a material which reflects light such as silicon dioxide to prevent light from escaping (or entering) and to aid in internal reflection within the scintillators.

According to another preferred embodiment of the present invention, each scintillator element 179 is isolated from its adjacent scintillator elements 170 by a thin sheet 171 of a highly x-ray opaque material such as, for example, gold or lead. Sheets 171 may preferably be about 0.0102 cm (0.004 in) to 0.0127 cm (0.005 in) thick. The position of sheets 171 between the scintillators 170 is shown in Fig. 12. As shown herein, the area of the circular active area of collimation grid 90 is larger than the area of detector array 110. Thus the x-ray pencil beams emitted from the respective apertures 140 of collimation grid 90 all converge toward the detector array 110 while each individual x-ray beam 100 diverges, or spreads, as would a flashlight beam.

Image Processing An important refinement of the present invention concerns the application of an image processing system to further reduce the required dosage. In practice, the signal from the detector is not usually applied directly to the "z" or luminance input of a video monitor. Instead, digitized intensity data for each pixel are stored in a discrete address in a "frame store buffer". More than one such buffer may be used in certain applications. Pixel addresses within the buffer can be randomly accessed and the numeric intensity value can be manipulated mathematically. This function has application in applying various image enhancement algorithms and it allows for pixel

assignment of the data from discrete segments of the detector array.

According to a preferred embodiment of the present invention an SBDX image would consist of up to about 250,000 pixels, arranged in 500 rows and 500 columns (corresponding to the 500 rows by 500 columns of apertures at the center of collimation grid 90). For the puφose of the explanatory example below, it is assumed that the scanning x-ray source is momentarily centered upon the pixel, P, located at row 100 and column 100 of collimation grid 90. It is further assumed in regard to this embodiment that the detector array 110 consists of a 3 by 3 array 110 containing 9 segments 179 (Fig. 12) and that each segment 179 is sized so as to intercept ail of the x-ray emissions associated with a single pixel. Other array configurations obviously may be used as are detailed herein.

The numerical values, digitized from the individual segments of the detector array 110, are assigned to pixel addresses as follows:

Segment 1 - row 99, column 99 Segment 2 - row 99, column 100

Segment 3 - row 99, column 101 Segment 4 - row 100, column 99 Segment P - row 100, column 100 Segment 6 - row 100, column 101 Segment 7 - row 101 , column 99

Segment 8 - row 101, column 100 Segment 9 - row 101, column 101 The same pattern of data assignment is repeated as the scanning x-ray beam passes ail of the pixels. In the displayed image, the numerical value of each pixel is equal to the sum of

"n" parts where "n" is the number of segments 179 in the array 110 (in this example, n=9).

When constructed as shown herein, the detector array 110 has the effect of

fixing the working distance at which optimum focus is obtained and providing a plane of optimum focus not available in prior art non-segmented detector array SBDX imaging systems.

The following parameters must be taken into consideration in design of the detector

1. The size and shape of the collimated beam from the x-ray source (anode target) 50;

2. The distance between the source 50 and the detector array 110, "SD"; 3. The distance between the source 50 and the center of the object of interest 80, "SO";

4. The desired resolution, or pixel size at the object of interest 80;

5. In medical applications, the total area of the array must be large enough to intercept all of the x-rays from the collimation grid 90. In an SBDX system according to a presently preferred embodiment of the present invention, the distance between the x-ray source 50 and the exit side 260 of collimation grid 90 is about 2.271 cm (0.894 in) (see Figs. 3, 5). Apertures 140 are 0.0381 cm (0.015 in) by 0.0381 cm square. The spot size of electron beam 40 on anode 50 is about 0.0254 cm (0.010 in) in diameter. The detector array 110 is 91.4 cm (36 in) from anode 50. Thus, the beam width of x-ray beam 100 is

2 ^* ARCTAN((spot diameter/2)/ ((aperture width/2)+(spot diameter/2))* 2.271 cm (0.894 in), or 1.6°. At a distance of 91.4 cm (36 in) from anode 50, the projected x-ray beam diameter is 91.4 * TAN(1.6°) cm. Therefore, the detector array 110 should be about 2.54 cm (1 in) on a side for this preferred embodiment. For example, if the object to be imaged is 22.86 cm (9 in) from anode 50 and the desired pixel size is 0.0508 cm (0.020 in) at the object, and the distance from source to detector, again, is 91.4 cm (36 in) with an optimal detector array size of 2.54 cm (1 in) square, the projected size of pixels at the detector plane 270 is simply (SD/SO) ^* pixel size at object, or 0.2032 cm

(0.080 in). Dividing 2.54 cm (1 in) by 0.2032 cm (0.080 in) we see that the desired resolution may be obtained with a square segmented detector having 12 to 13 segments on a side. Obviously, many other configurations could be used depending upon the circumstances in which the SBDX system is to be used. Outside of the plane of optimum resolution, SO (280 in Fig. 7D), resolution will degrade to one half at 0.5 x SO and at 2 x SO (281 in Fig. 7D). This allows for a reasonable depth of focus for most applications. In some applications, such as imaging the human heart, degraded focus outside of this range of depth is seen as being advantageous. Blurring of detail outside of the area of interest tends to increase the perception of details within the area of interest.

A number of methods can be used to obtain a useable image from the data obtained as described above. As described above, a simple convolution may be used, however, in this case, resolution will not be fully optimized. Two additional methods are presently preferred for obtaining maximal resolution and sensitivity from the captured data. These are called the multi-image convolution method and the multi- output convolution method. For both cases, the following is assumed:

There are AP _X rows of apertures and AP _Y columns of apertures in collimation grid 90. Each intersection of a column and row is a "pixel." Those pixels outside of the circular active area of collimation grid 90 are treated as if they contribute no measured intensity to the image, i.e., they are treated as if they are "dark". Pixels not illuminated by x-ray beam 100 during a scan are similarly treated as if they contribute no measured intensity to the image, i.e., they are also treated as if they are "dark".

Turning now to FIG. 15, there are a maximum of DET _X rows of sensor elements 160 in detector array 110 and a maximum of DET _Y columns of sensor elements 160 in detector array 110 in a pseudo-round sensor array 110.

ZRATIO is a real number between 0 and 1. If ZRATIO=1, the focus is set at the sensor plane. If ZRATIO=0, the focus is set at the x-ray source plane. If ZRATIO=0.5, the focus is half way between the x-ray source plane and the sensor

plane, and so on. PIXELRATIO is the number of image pixels per physical distance between adjacent sensors in a column or row. For example, if the spacing between pixel centers at object plane 280 is 0.01 cm, and the spacing between sensors at detector plane 270 is 1.0 cm, then PIXELRATIO=10. FOCUS=ZRATIO ^* PIXELRATIO.

IMAGE is a data array of dimension DET _X x DET _Y containing the intensity information for a particular scan and corresponding to a particular pixel. PIXEL is a 4 dimensional array of dimension AP _X x AP _Y x DET _X x DET _Y which contains the DET _X x DET _Y IMAGE data arrays obtained by scanning all (or part of) the apertures. PIXEL is refreshed after each scan according to one preferred embodiment of the present invention.

As the beam is scanned across the anode surface, it is, in effect, positioned before the center of selected apertures 140, "fired," and then repositioned. Thus for each firing, an IMAGE array of data will be acquired. While these images could be constructed into a displayable image having some use directly, more resolution and sensitivity is obtained by combining them. The first preferred method for combining the images is called the multi-image convolution method. In the multi-image convolution method, an OUTIMAGE array of intensities of dimensions AP _X x AP _Y , which can be displayed on a CRT or like display means, is formed by assigning to OUTIMAGE(y.x) the value of:

The second presently preferred method for combining the AP _X x AP _Y IMAGE data arrays into a useful picture is called the multi-output convolution method. In this case, with a sensor array of DET _X x DET _Y sensors there will be DET _X x DET _Y digitizers (or their equivalents, multiplexed) and the same number of pixel summing circuits. The digitized values from each sensor are called SENSOR(j,i). The final OUTIMAGE array is computed as follows - for each pixel in the output image array OUTIMAGE(y,x) [for y=1 to AP _Y and x=1 to AP _X ] one pixel from each of the DET _X x DET _Y source images SENSOR(j,i) is summed [for j=1 to DET _Y and i=1 to DET _X ] into destination image pixel OUTIMAGE(y-j*FOCUS,x-i*FOCUS). Normalization is then carried out over the OUTIMAGE array by dividing each element thereof by DET _X ^* DET _Y .

A further improvement upon these techniques may be obtained by performing linear inteφolation based upon the fractional part of the FOCUS factor.

An advantage of the multi-image convolution method over the multi-output convolution method is that the former allows the plane of optimum focus to be selected in software after the data is captured while the latter does not. The latter method, however, may be performed quicker where timing is a limitation.

Three Dimensional Image Reconstruction From SBDX Data

The SBDX system described herein may be used to generate a set of sequential planar images which can then be used to form a tomograph or a three dimensional display of the object 80. An image set can be analyzed to produce a 3D image consisting of a series of images at various depths by re-analyzing the data set with various values of FOCUS. The natural FOCUS values to use are n/DET _x or n/DET _γ where n is an integer from 0 to DET _X or DET _Y , respectively. Normally, only those focus values would be analyzed that correspond to planes of interest within the object 80. For example, in the SBDX system described in TABLE I (below), the planes of focus would be spaced at approximately 2.54cm (1 in) intervals near the

normal focal plane of 22.86cm (9 in) (plane of optimum focus).

The following formula shows where the sequential planar images are located in terms of distance from the anode 50.

EQ. 2 FA FOCUS) = — ≤ — ≤ -^ ^c FOCUS *λ _d

Where F _t (FOCUS) = Distance from the anode to the particular focal plane of interest

F _d = Distance from the detector to focal plane (distance from the anode to the detector less F^ λ,= Center-to-center spacing between adjacent collimation grid apertures λ _d = Spacing between centers of adjacent detectors 160 within detector array 110. When using the technique of sub-sampling, the computation does not change - just the data from the collimation grid apertures which are not "skipped" is processed. But A, remains the same even if intervening collimation grid holes are missing.

Negative Feedback X-Rav Flux Control

Turning now to Fig. 13 an SBDX imaging system employing a negative feedback path 305 to control the x-ray flux of x-ray beam 100 is depicted. Preferably negative feedback from the sensor array is utilized to control x-ray flux so that the sensor array always sees approximately the same flux level. In this way when soft tissue (which is relatively transparent to x-rays) is being scanned, the x-ray flux will drop, reducing the overall dosage to the patient (or object). Improved contrast and dynamic range are provided by using negative feedback flux control. According to this

embodiment, differential amplifier 310 has an adjustable reference level 320 which may be set by the user. Negative feedback loop 305 feeds back to x-ray tube 10 to control the x-ray flux.

Time Domain Scanning Mode A time domain x-ray imaging system may also be implemented using the principles disclosed herein. In such a system, the time to reach a predetermined measured x-ray flux from the various pixels could be computed and mapped. Negative feedback control could then be employed to turn off or reduce x-ray flux from apertures corresponding to pixels which had reached the predetermined flux level for the scan period in question. In this case, the information gathered would be time to flux level and the mapped or imaged information would correspond to time rather than intensity. Such a system has the potential to provide much higher signal to noise ratios, improved contrast, drastically reduced x-ray dosage to the object under investigation, and improved dynamic range.

Multiple Energy X-rav Imaging Mode

According to one preferred embodiment of the present invention, two or more groups of x-ray beams 100 are directed toward one or more detector arrays. A first group of x-ray beams has a first characteristic x-ray energy spectrum. A second group of x-ray beams has a different second characteristic x-ray energy spectrum. By comparing the measured transmissivities of the first and second group of x-ray beams, the presence of certain materials in the object under investigation may be detected. The basic concept of use of differential x-ray imaging is known in the art and is disclosed, for example, in U.S. Patent No. 5,185,773 entitled "Method and Apparatus for Nondestructive Selective Determination of a Metal" which is hereby incoφorated herein by reference.

The two groups of x-rays may be generated in a number of ways. One such

way is by fabrication of a special anode 50 having a first material or first thickness of a material adjacent to the apertures of the first group of apertures and a second material or second thickness of material adjacent to the apertures of the second group of apertures. In this manner, the apertures associated with the first group will emit x-rays having a first characteristic energy spectrum and the apertures associated with the second group will emit x-rays having a second characteristic energy spectrum. Alternatively, K-filtering (or K-edge filtering) can be used by placing filter material (such as, for example, molybdenum) within a portion of the apertures 140 to produce a similar effect. In this case, a first group of apertures would comprise a first filter inserted therein and a second group of apertures would comprise a second filter inserted therein. The second filter could be no filter at all. As in the previous case, two groups of x-rays having different characteristic energy spectra would be associated with the two groups of apertures.

Once at least two groups of apertures are associated with different characteristic x-ray spectra, it is now possible to detect micro-calcification (a precursor of breast cancer) and other abnormalities not normally visible with broadband x-rays. For example, by performing a scan of the first group of apertures to form a first image, then performing a scan of the second group of apertures to form a second image, and dividing the images to highlight their ratios, it is possible to detect micro-calcification and other such abnormalities with a low dosage scanning beam x-ray imaging system - in real time. Similarly, a multiple detector array arrangement could be used with group 1 apertures directed toward a first detector array and group 2 apertures directed toward a second detector array, etc.

Another embodiment of multiple energy imaging is now described. Because the amplitude of the electrical pulse from a detected x-ray photon is proportional to the energy (KV) of the photon, it is possible to count separately the pulses coming from photons in two or more energy bands. The pulses are separated by intensity and then counted and processed separately, forming two or more separate images. Those

images can be displayed as ratios.

It is also possible to change the selected energy levels on the fly to distinguish different density regions in the object. The advantage of this embodiment is that it is more flexible than those described above, and does not require special collimation grids, anode materials, or dual detectors.

While a number of preferred embodiments have been discussed above for various configurations of the present invention, the following specifications are illustrative of a presently preferred SBDX imaging system according to the present invention:

TABLE

Grid

Shape: round

Diameter: 25.4 cm (10 in)

Aperture Pitch: 0.0508 cm (0.020in)

Number of Apertures across a diameter: 500 (166 for sub-sampling version)

Area of grid: 506.45 sq. cm (78.5 sq. in)

Number of apertures: 196,350 approx. (approx. 21 ,630 fo sub-sampling version)

Aperture cross-sectional shape: round Aperture width: 0.0381 cm (0.015 in) Space between apertures: 0.0127 cm (0.005 in) Dist. between anode surface and Collimation Grid Output Face: 2.5 cm (0.98 in)

B. Source-Detector Distance: 91.4 cm (36 in)

Location of Plane of Optimum Focus: 22.86 cm (9 in) from Grid

C. Scan Frequency: Adjustable to 30 Hz D. Operating voltage on x-ray tube: 70-1 OOkV

E. Detector Array:

Overall shape: pseudo-round (per Fig. 15)

Shape of input face of detector elements: square

Size of input face of detector elements: 0.135 cm x 0.135 cm

Number of detector elements: 96 in pseudo round array of diamet

12

Array diameter: 1.83 cm (0.72 in)

Total included angle subtended from the detector center point to the collimation grid outside diameter: 15.8°

Field of view at Plane of Optimum Focus: 19.05 cm (7.5 in)

Pixel size at Plane of Optimum Focus: 0.038 cm

Pixel size at Detector Plane: 0.152 cm center-to-center detector spacing

Resolution: 13 line pairs/cm

Accordingly, an SBDX imaging system utilizing a segmented detector array has

been shown and described which simultaneously provides high resolution, high sensitivity, and low x-ray dosage to the object under investigation. The system also permits the point of optimum focus to be set at any point between the source 50 and the detector array 110, and provides an effective working depth of field.

Beam Sub-sampling Technioue

The following relates to a particular preferred embodiment of the present invention which uses the technique of beam sub-sampling in order to reduce the computer processing overhead, and power consumption of the scanning beam digital x-ray system. Standard video quality images use 640 x 480 pixels and are updated at 30Hz.

This requires a pixel sample rate of about 12MHz. Positioning the high voltage electron beam of the x-ray tube accurately behind 250,000 sequential different apertures at that rate requires high precision and relatively high power consumption. Digitization of signals from a large array of x-ray detectors at a 12MHz rate is similarly expensive and power intensive. Thus reduction of the pixel sample rate below 12MHz without significant reduction of the spatial or time resolution of the SBDX is useful in reducing initial unit costs, operating costs due to electric power consumption, and cooling requirements for the waste heat developed by the x-ray tube.

Accordingly, a mechanism for reducing the pixel sample rate while providing virtually the same spatial and time resolution has been developed. This mechanism is referred to as sub-sampling and is best implemented with the embodiment of the SBDX described in this section, although it could obviously be adapted to be used with other configurations of the SBDX. Advantages of this embodiment include reduced power consumption and simpler circuitry for electron beam deflection within the x-ray tube, reduced cost of fabrication of the collimation grid 90, reduced complexity of the calculations needed to resolve an image of the object 80 and other advantages as would be obvious to those of skill in the art.

Pursuant to this embodiment a collimation grid 90 is fabricated having a reduced number of apertures, preferably AP _X =AP _Y =166, rather than 500, although other numbers could obviously be used. The advantage of this reduction from a computational point of view will become apparent below. From a manufacturing point of view, however, it is a much simpler structure with approximately one-ninth the number of apertures which need to be fabricated. Because of the reduction in the number of apertures, it is easier to fabricate grids with higher deflection angles (i.e., the angle that the aperture makes with respect to the front face 260 of the collimation grid) without running into problems of having apertures intersect with adjacent apertures. This is particularly useful when stereo grids are to be manufactured as adjacent apertures in a stereo grid are directed to different detector arrays and hence require more physical separation than non-stereo grids to avoid aperture intersection.

The apertures of the collimator grid are arranged in a circle of maximum dimension AP _X rows by AP _Y columns. For computational puφoses this may be treated as a rectangle of dimension AP _X rows by AP _Y columns with the elements outside of the circle contributing no information, i.e., always being "dark" or non- illuminated by x-rays.

The sensors 160 of the x-ray detector array 110 are arranged in a circular array of maximum dimension DET _X rows by DET _Y columns as shown in FIG. 15. The pixel sample rate may be reduced by illuminating less than all of the apertures of the collimation grid, i.e., by sub-sampling. Preferably a collimation grid without the not-to- be-illuminated apertures is used. In order to form an image with the detector array, only every DET _x -th collimator hole in each row and every DET _γ -th collimator hole in each column needs to be illuminated and thus the image may be built up out of image tiles of pixels, each DET _X pixels by DET _Y pixels in size. This corresponds to a sub- sampling ratio of DET _X by DET _Y whereas no sub-sampling corresponds to a sub- sampling ratio of 1 by 1. The sub-sampling ratio may thus be adjusted from 1 to DET _X in the X-direction (rows) and from 1 to DET _Y in the Y-direction (columns). In

accordance with this preferred embodiment, DET _X =DET _Y =12 as shown in FIG. 15.

Where a 12 x 12 detector is used and the sub-sampling ratio is 12, the image is fabricated from a plurality of non-overlapping images which are in effect "pasted" together - much like a David Hockney photomosaic. Because real world scintillators and detectors are not all perfectly and identically responsive, the x-ray pencil beam is not perfectly uniform, the collimation grid apertures are not all exactly identical with identical areas, and because a circular, rather than a square detector is used, some degree of overlap is highly desirable in order to permit averaging out detector non- linearities and noise. If the sub-sampling ratio is less than the detector size in pixels (that is, less than 12 in this preferred embodiment), the image will be built up from overlapping "tiles", which must be summed or averaged. If the sub-sampling ratios are not even multiples of the detector size (in pixels) or if the detector array is not rectangular, there will be different numbers of samples added to each pixel, and different divisors needed to average each pixel. The techniques for handling these less than ideal circumstances are well known to those of skill in the art and need not be disclosed here to avoid over-complicating the disclosure.

In the calculations that follow, SSX represents the sub-sampling dimension in the X direction (rows), and SSY represents the sub-sampling dimension in the Y direction (columns). For example, if SSX=SSY=1 , there is no sub-sampling and the processing takes place exactly as in the other embodiments of the present invention discussed above. Similarly, if SSX=SSY=12, in this embodiment, we get back to the "photomosaic" with no pixel averaging. If SSX and SSY are 3 and the circular active area is 500 x 500, then 166 x 166 apertures will be scanned, i.e., one-third in X and one-third in Y, reducing the data obtained by a factor of 9. Note that if one is only going to use one-ninth of the apertures all of the time, there is no need for them and they need not be included in the collimation grid.

Accordingly, only 1/(SSX ^* SSY) apertures in the original collimation grid

(500 x 500 apertures) need to be used or illuminated by the electron beam for the formation of x-rays in order to be able to generate an image. If the frame rate is kept constant, e.g., 30Hz, then the number of electron beam motions is reduced by SSX ^* SSY, as is the frequency response of the circuit that drives the electron beam. The total distance travelled by the electron beam (and the number of scan lines) is reduced by 1/SSY, so that the average beam velocity across the target anode is reduced by 1/SSY. The image reconstruction pixel rate is the same as the collimation grid aperture rate (rate at which apertures are scanned or illuminated), and is also reduced by 1/(SSX*SSY). In accordance with this scheme, the number of samples averaged into each display pixel is (DET _X /SSX) ^* (DET _Y /SSY). When using the maximum sub-sampling where SSX=DET _X and SSY=DET _Y , only one digitizer sample is averaged into each display pixel (the "photomosaic" mode). The sample averaging is important for smoothing out non-uniformities in the beam, the scintillators, the detectors, and the amplifiers. The amount of sub-sampling (SSX and SSY) must be set to an appropriate level for the conditions presented in order to assure acceptable image quality. This may be adjusted by the user on the fly in accordance with the user's preference for image quality and the conditions presented by a particular set of circumstances. The detector array 110 shown in FIG. 15 is preferably an array of 96 individual detector elements 160 arranged roughly in a circular area having a diameter of about one inch. There are 12 detectors (DET _X ) in the vertical column at the center of the array and 12 detectors (DET _Y ) in the horizontal row at the center of the array. The scintillator crystals are preferably cut to a square horizontal cross-section and are supported by an "egg crate" structure made out of 0.005" thick stainless steel strips. The circle 400 of FIG. 15 within which all of the scintillator crystals (cross-hatched) are located, is preferably about 0.800" in diameter.

The length of the scintillator crystals in detector array 110

is preferably about 0.10 cm and the front input faces are preferably 0.135 cm x 0.135 cm. The scintillator crystals are preferably YSO, LSO or BGO but other materials may also be used as discussed above. For a suitably reduced decay time for its light output in this application (to about 50 nS), BGO needs to be heated to approximately 100°C. Accordingly a resistive heating element may be provided.

FIG. 17 depicts the detector assembly 402 according to a presently preferred embodiment of the present invention. X-rays enter from the top through x-ray window 404 in lead shield 406. X-ray window 404 is preferably circular and about 1.91 cm (0.75 in) in diameter in order to permit x-rays coming from the apertures of the collimation grid 90 to strike the detector array 110 while attenuating scattered x-rays. A light shield 408 is provided to shield the detector from errant light. It may be made of a thin sheet of aluminum or beryllium chosen to attenuate light without substantially attenuating the x-rays. It is 0.0125 cm thick.

Detector array 110 is located near optional heating element 410 for use with a BGO scintillator. Heating element 410 may be a resistive heating element designed to keep the detector array 110 at an operating temperature of about 100°C. A fibre optic imaging taper 412 directs photons emerging out of the bottom 414 of detector array 110 to a 96 channel photomultiplier tube (PMT) 416. The detector assembly 402 is enclosed in a light tight outer housing 418 in order to prevent stray light from generating noise. Three shoulder screws 420 and three centering screws 422 are provided for planar and linear alignment as is well known to those of ordinary skill in the art. Rotational alignment is achieved by rotating outer housing 418 with respect to PMT mount 426. The fibre optic imaging taper 412 is available from Collimated Holes of Campbell, CA and has a circular input aperture of diameter 2.03 cm (0.8 in) and a circular output aperture of diameter 3.38 cm (1.33 in). Taper 412 matches each scintillator crystal pitch dimension (0.06") to that of the PMT 416 (0.10"), i.e., it has a magnification of 1.667 times. High viscosity optical coupling fluid available from Dow Corning (Type 200) with a refractive index approximately matching that of the glass is

used at the two faces of the taper as an optical coupling medium in order to maximize the light transfer efficiency from the scintillator crystals 160 to the taper 412 and from the taper 412 to the PMT input face 424.

Photomultiplier tube 416 is a 96 channel tube (one channel corresponding to each scintillator crystal 160) available under model number XP1724A from the Philips Coφoration. It has a fibre optic face plate so that the spatial arrangement of the scintillator array is accurately carried through to the PMT photocathode located in the PMT on the other face of the faceplate. An x-ray photon striking one of the scintillators 160 produces a light pulse which is coupled to the PMT photocathode. This produces a corresponding electron pulse at the photocathode and the pulse is amplified in one channel of the PMT dynode structure up to 1,000,000 times.

The PMT output pulse is connected to the input of a 30 MHz bandwidth amplifier, the output pulse of which is in the range of 0.5 to 5.0 volts and about 30 nsec in duration. The amplifier is AC coupled to eliminate offset drift problems. The AC coupling low frequency cut-off is high, e.g., 30 MHz, so that the pulse is differentiated. This eliminates the need for a DC restorer circuit to keep the baseline reference voltage constant as the pulse rate varies.

The amplifier output feeds a comparator which gives a constant amplitude output pulse regardless of the amplitude of its input. The reference voltage for the comparator is set to a value which is slightly higher than the amplifier noise output level so that it will not trigger on the noise level. The amplifier chain is repeated 96 times, once for each scintillator crystal in the detector array. The comparator output pulses provide the raw data for the data acquisition and image re-construction system. Tests have shown that the prototype system is capable of counting randomly occurring x-ray photons up to about a 10 MHz rate.

While embodiments and applications of this invention have been shown and described, it would be apparent to those skilled in the art that many more modifications than mentioned above are possible without departing from the inventive

concepts herein. The invention, therefore, is not to be restricted except in the spirit of the appended claims.