Data Acquisition System for Computed Tomography

This invention relates to radiation measurement and signal processing systems for use with computed tomography (CT) imaging equipment.

Apparatus for examining objects by means of penetrating radiation, typically X-ray or gamma-ray radiation, and for producing cross-section images thereof by the methods of computed tomography is well known and is described, for example, in U.S. Pat. No. 3,778,614 to Hounsfield. In typical CT apparatus the object undergoing X-ray examination is scanned by an X-ray source in combination with one or more X-ray detectors which rotate and/or translate about the object to measure radiation transmission along a large plurality of independent paths in an examination plane. Data obtained from the detectors is then processed and combined, typically by using well known computational algorithms in a digital computer, to generate images of transmissivity characteristics in the plane.

Several computational algorithms are available for generating image information from transmission data. A convolution and backprojection algorithm, of the type first described by Ramachandran and Hakshminarayanan, is generally believed to provide a reasonable compromise between computer equipment cost and image reconstruction time and is utilized, in various forms, in most CT equipment presently in commercial production. Convolution-backprojection algorithms of this type are most efficiently utilized with transmission data measured at discrete points which are uniformly distributed in space. However, the scanning mechanisms which are utilized in the measurement of radiation transmission generally comprise large, high inertia components and tend to translate and/or rotate with time-varying velocity profiles. When data from such mechanisms is sampled at a uniform periodic rate the corresponding transmission paths are not uniformly distributed in space.

In accordance with the present invention, measurements of X-ray transmission in CT apparatus are made at points which are equally spaced in time (and which are generally not equally distributed in space) at a sampling rate substantially higher than the cut-off frequency of the filter function utilizied in a convolution-backprojection image reconstruction algorithm. Groups of data samples which surround each of a series of points (uniformly distributed in space) are fitted to a high order curve, typically a quadratic curve, which is interpolated to calculate equivalent data signals having a uniform spatial distribution. The system allows data acquisition at a uniform rate, which generally reduces hardware cost and complexity; inherently provides low pass filtering to reduce noise, and allows effective multiplexing and transmission of data from the scanning apparatus to the digital computer.

It is, therefore, an object of this invention to reduce the complexity of scanning and data transmission apparatus in computed tomography systems. Another object of this invention is to reduce the effect of statistical signal variations and noise on the image quality in computed tomography systems.

The novel features believed to be characteristic of the present invention are set forth in the appended claims. The invention itself, together with further objectives and advantages thereof, may best be understood with reference to the following detailed description, taken in connection with the appended drawings in which:

FIG. 1 schematically represents a computed tomography system of the present invention; and

FIGS. 2 and 3 graphically describe the data sampling methods and operations of the present invention.

FIG. 1 is a computed tomography system which may incorporate the data processing equipment and techniques of the present invention. An X-ray or other radiation source 10 is mounted on a yoke 12 opposite a linear array of radiation detectors 14. An object 16 undergoing tomographic examination is disposed between the radiation source 10 and the detectors 14 in the path of radiation propagating therebetween. Motor drive 13a and 13b cause the yoke 12 to translate and/or rotate about the object 16 whereby the detectors 14 produce signals which are representative of the radiation intensity transmitted from the source 10 to the detectors 14 along a large number of separate paths through the object 16.

A position encoder 18 is connected between the yoke 12 and a fixed reference to produce signals which are representative of the rotational and translational position of the source and detectors with respect to the body 16. Signals from the individual detectors 14 and the position encoder 18 are periodicly sampled and combined, typically in digital form, in a sampler and multiplexer circuit 20 and are typically stored, for example in a disk file 22 and transmitted to a digital computer 24 which generates image data representative of the physical characteristics of the object 16. The image data may, typically, be displayed on a cathode ray tube display 26 which is driven by the computer 24.

In general the yoke 12 will undergo some velocity changes while X-ray measurements are being taken. Such changes may be necessary, as in the case where acceleration and deceleration must take place near the ends of translational motion where ray measurements are still required. Other changes may be unintentional but difficult to avoid due to the performance limitations of the yoke velocity control system. Thus the velocity of the yoke may vary substantially as it rotates and/or translates about the object 16. If the multiplexer circuit 20 operates to acquire data from the detectors 14 at a uniform periodic rate; the data thus obtained will represent radiation transmission along paths which are non-uniformly distributed in space. Such data is, generally, not suited for economical image reconstruction with a convolution-backprojection algorithm.

Image reconstruction algorithms may compute the radiation absorption at each of a large number, N, of picture elements by solution of an equal or larger number of simultaneous equations with N unknown values. A minimum of N X-ray transmission data points must, therefore, be obtained to allow a closed solution of the set of image equations. In accordance with the present invention radiation transmission data is measured by the detectors 14 over a substantially larger number of transmission paths (typically 5-10 times as many paths) than the number of desired picture elements in the computed image. The position of the yoke 12, which is in continuous motion during the course of the measurements, is monitored during the measurement process and is stored, along with the values of the detector 14 signals on the disk file 22. In a typical system the multiplexer 20 serially scans an array of detectors 14 (which may, for example, include 30 separate detectors) and samples the position encoder twice during each scan of the detector signals.

Signals representative of the sampled signals from the detectors 14 and of the corresponding position of the yoke 12 are received by the digital computer 24 from the disk file 22 and are processed by a program (an example of which is included as Appendix A) which compensates for the geometry and motion of the yoke and detector array to produce a data file representative of the actual spatial position of the transmissivity measurements with respect to known reference points in the scanner mechanism. The curves in FIGS. 2 and 3 represent typical data which might be produced by those measurements and computations. The round dots in the Figures represent data values which are transmitted by the multiplexer 20 to the disk file 22 at uniform intervals in time. FIG. 2 represents the spatial position of each of the computed measurement points which, because of the non-uniform motion of the yoke 12, are not uniformly distributed in space. FIG. 3 represents corresponding transmission values obtained at each of the data points which values vary as a function of the structure of the object 16 and which are also subject to statistical variation and the effects of additive noise. Points A, B and C in FIG. 2 represent the position of a set of uniformly distributed measurement paths which define desirable measurement points for data input to a convolution-backprojection image reconstruction program. The corresponding points A', B' and C' in FIG. 3 represent equivalent sample times at which optimum data sampling for the paths A, B and C are desired. Typically, the points A', B' and C' do not correspond with the actual sampled data points.

In accordance with the present invention, transmission data values from a group of sample points adjacent the optimum measurement positions A', B' and C' are interpolated to provide an equivalent transmission value at the optimum measurement positions A', B' and C'. For example, the equivalent transmission value Y.sub.A at the point A' is calculated by interpolating measured transmission values at the five nearest sample points 101, 102, 103, 104 and 105. Likewise the equivalent transmission value at the point C', Y.sub.C, is determined by interpolating data from the five nearest measurement points 109, 110, 111, 112 and 113.

Interpolation is most efficiently accomplished in the digital computer. The computer program first searches the file of position encoder values, as corrected for scanner geometry and motion effects, to determine which measurement point is closest in space to the effective position of a desired transmission path. The closest measurement point (for example point 103 for path position A') together with the four adjoining measurement points (101, 102, 104 and 105) are fitted to a curve, typically a quadratic or higher order curve, by means of a leastsquares fitting technique, or any other accepted method, to produce a functional representation of the continuous variation of transmission values in the region of the point A'. The generated function is then interpolated to provide an equivalent transmission value at the point A'. The curve fitting and interpolation algorithm provides smoothing of statistical data variations, in effect providing a low pass filer, and allows optimum spacing of equivalent transmission paths for efficient use of the convolution-backprojection algorithm.

By way of example a quadratic curve having the form a + bx + cx.sup.2 may be generated through the points y.sub.-2... y.sub.2 by a least-squares quadratic fit using the equations

By way of further example and to permit others to more easily practice the invention Appendix A is a representative computer program (written in Digital Electronics Corporation PDP 11 machine language with Fortran subroutines) which compensates for scanner motion and geometry effects and synthesizes optimal spatially sampled data from uniformly time sampled data.

The methods and apparatus of the present invention allow rapid scanning and acquisition of transmission data for computed tomography systems, by permitting use of scanning mechanisms having non-uniform velocity profiles, with subsequent efficient processing using convolution-backprojection algorithms. The method further allows statistical averaging and low-pass filtering which tend to reduce the effects of noise and other statistical variations on image quality and thus contribute to lower radiation levels and patient dose.

The invention has been described in detail herein in accord with certain embodiments thereof, yet many modifications and changes may be effected by those skilled in the art. Accordingly, it is intended by the appended claims to cover all such modifications and changes as fall within the true spirit and scope of the invention. ##SPC1## ##SPC2## ##SPC3##