GB2463254A

GB2463254A - Radiation detector for determining a direction to a radio-active source

Info

Publication number: GB2463254A
Application number: GB0816159A
Authority: GB
Inventors: David Ramsden; Antony Bird; Gerry Skinner
Original assignee: University of Southampton; Symetrica Ltd
Current assignee: University of Southampton; Symetrica Ltd
Priority date: 2008-09-04
Filing date: 2008-09-04
Publication date: 2010-03-10
Anticipated expiration: 2028-09-04
Also published as: EP2329296A2; WO2010026363A2; GB2463254B; GB0816159D0; WO2010026363A3

Abstract

An apparatus 10 for determining a direction to a source of radiation, e.g. a gamma-ray or neutron source at a range of 100 m or more, is described. The apparatus 10 is rotatable, e.g. about a vertical axis, and comprises a plurality of radiation shields 14 and radiation detectors 16. The radiation shields 14 are separated from one another by gaps 15 to form an alternating series of radiation shields 14 and gaps 15 around a closed path within the plane of rotation for the apparatus 10. The radiation detectors 16 are arranged to detect changes in radiation passing through gaps 15 between the radiation shields 14 as the apparatus 10 is rotated. In providing a rotatable series of radiation shields 14 and gaps 15 with radiation detectors 16 arranged to detect radiation passing through the gaps 15 in this way, a radiation imager 10 is provided that may be seen as being broadly based on coded mask techniques, but which is sensitive to radiation from all azimuths in its plane of rotation.

Description

S

TITLE OF THE INVENTION

RADIATION DETECTOR

BACKGROUND ART

The invention relates to radiation detection, and in particular to determining the direction to a source of radiation.

There is often a need to identify the presence of radiation sources, e.g. sources of neutrons and / or gamma-rays, in an environment, and also to obtain an indication of the location of such sources. For example, there may be a need to determine a direction towards a lost radio-active source in a legitimate environment, such as a nuclear processing plant or power station, or there may be a need to identify and determine a direction towards illegitimate sources of radiation, e.g. in the interest of a nation's security.

There are a number of known schemes for scanning cargoes / people for radioactive materials, e.g. for use at ports, or other border crossings, to provide security against trafficking in radioactive materials. Examples of such schemes include radiation portal detectors, e.g. for lorries to drive through, and handheld scanners, e.g. for an operator to scan a person / cargo. Examples of such devices are described in US 2006/0284094 [1] and by Larssen [2]. These types of detectors have proved effective in scanning individual vehicles / people as they pass a control point, but are less suited to general scanning of larger areas.

Scanning larger areas may be desired in a range of situations. For example, at a port it would be more time-efficient if a detector could be used to identify the presence of, and direction to, a radioactive source within a shipping container in a compound of many other containers, without needing to scan each of the containers individually, either by hand or as they pass a control point. In other examples there may be a desire to identify the presence and location of a source of radiation in an environment where there is no controlled passage of individuals through a control point. For example, a national security force may have reason to believe that an illicit source of radiation is being stored within a particular region, and wish to scan the region to attempt to identif' the presence of, and direction to, any source.

Thus there is therefore a desire for radiation detectors thatare able to determine directions to a potential sources of radiation within relatively large areas, e.g. to be able to detect and indicate a direction to sources of radiation that might be on the order of 100 m away, or further. This type of radiation detection might be referred to generally as "stand-off' detection, as opposed to radiation detection based on close-quarters scanning.

The detection of radioactive materials at large stand-off distances in the terrestrial environment is made difficult because of the high levels of radiation produced by long-lived isotopes in both the natural and the built environment. This-locally-produced background radiation can be reduced by limiting the field of view of a detector through the use of collimation. However, a collimation approach generally leads to the use of a large additional mass in a system, and provides relatively poor angular-resolution. If the collimation angle is small (to increase spatial resolution), then many separate observations are required in order to locate a source because of the

small field of view.

As an alternative to a collimation approach, an imaging approach may be considered. One type of "stand-off" radiation imaging technique, e.g. for imaging gamma-rays having energies from say 100 keV to a few MeV, is the so-called coded mask / aperture imaging technique developed in the field of astronomy.

High energy radiation cannot generally be focussed using lens or mirror systems such as may be used at other energies (e.g. for visible light). At higher energies so-called straight-line optics solutions are employed.

The simplest straight-line optics imager is a pin-hole camera. This type of irnager employs a position sensitive radiation detector to detect radiation passing through a "pin-hole" in a radiation-opaque front panel. Radiation from a point source passes through the pin-hole and is detected at a pixel in the detector plane. Back projection from a pixel showing a high signal through the pin-hole thus indicates a direction to a source of the high signal. A drawback of the pin-hole camera is that it has very limited transmission, and hence low sensitivity. However, if the pin-hole is made larger to collect more radiation, the resolution of the image is degraded.

S

Coded aperture imaging (also known as coded mask imaging) in effect employs an array of pin-holes (the mask) placed above a position-sensitive detector-plane. The use of multiple pin-holes provides greater sensitivity as more radiation passes to the detector plane. Each pin-hole produces its own image at the detector plane, with the images from the different pin-holes being shifted with respect to one another. Mathematical processing techniques, such as spatial deconvolution, are used to generate a single image from the multiple overlapping images.

Thus in coded aperture imaging, the mask consists of both open (transmitting) and closed (opaque) elements, with the opaque elements constructed from a high-Z absorbing material such as lead or tungsten. The overall mask transmission (the fraction of open elements) can be varied. A 50% fill is often considered optimum for use in a low signal-to-noise environment. The mask pattern may be chosen either with a random distribution of open and closed elements, or in accordance with well-established mathematical prescriptions [3, 4J so as to provide optimal imaging (e.g. to provide a point spread function comprising, so far as possible, a delta function with flat side-lobes.

Figure 1 shows in schematic cross-section a conventional astronomical-type coded mask imager 1. The imager I comprises a mask 2, a detector 4 and a surrounding shield 6. Also shown schematically in Figure 1 are two sources of radiation A and B. These are considered sufficiently distant from the imager 1, that radiation from them is taken to be parallel, as schematically indicated by the series of rays from the sources A and B to the imager I shown in the figure.

The mask 2 in the plane of the figure comprises 10 elements, Six elements are opaque to the radiation being imaged (shown shaded in Figure 1), and four are transparent (shown un-shaded in Figure 1). The detector 4 in the plane of the Figure comprises a line of detector pixels. The shield 6 is opaque to the radiation being imaged and prevents radiation that has not passed through the mask from reaching the detector.

Sources within the field-of-view (FOV) of the imager I project respective shadows of the mask pattern Onto the detector plane. If a source is sufficiently close (in angular terms) to a central axis 8 of the imager I that it illuminates the whole detector plane (e.g. such as source A in Figure 1), the source is in the fully-coded field of view.

S

However, if a source only illuminates a part of the detector plane (e.g. such as source B in Figure I) because it is relatively far from the central axis 8, then the source is in the partially-coded field of view, and is imaged with a lower sensitivity.

A coded mask imager will in practice generally have a two dimensional mask (in some respects resembling a crossword puzzle layout) and a two dimensional detector plane.

In use, the detector plane 4 records the superposition of shadows of the mask 2 cast by sources within the field of view (the shadowgram). The source distribution may then be recovered from the shadowgram by any of a number of well-established deconvolution techniques [5], including simple matrix inversion (rarely used in practice), cross-correlation, back-projection or iterative methods such as maximum likelihood analyses.

An example application of coded mask-imaging techniques for searching for sources of gamma-rays in a terrestrial context is described by Ziock [6].

An enhancement to the basic scheme represented in Figure 1 is known as anti-mask imaging. With anti-mask imaging the transparent and opaque elements of the mask are periodically swapped, e.g., by rotation or translation of the mask [7]. This is so that detector pixels that were previously exposed to a source through a transparent mask element, and so observing a source signal plus a background signal, are now shielded from that particular source, and so only observing a background signal, and vice versa. Subtraction of mask and anti-mask images allows the removal of non-uniform detector background, which can be important in a large imager where complete pixel uniformity cannot be assured.

In terms of the key performance parameters, the angular resolution of a coded mask imager is primarily defined by the angular size of the mask elements as seen by the detector plane, although strong sources may be located to a higher precision by determining the centroid of a reconstructed source position.

There are a number of advantages of coded aperture imaging approaches, such as: (i) It is a well established technique used in astronomy since early I 970s.

(ii) It is able to image relatively faint sources in relatively low signal-to-noise scenarios (given sufficient exposure time).

S

(iii) It can image multiple sources simultaneously (the multiplex advantage compared to simple collimatecLsystems).

(iv) It can still provide useful results even with a significant number of detector pixel failures because information for each source is spread over the whole detector plane.

However, the inventors have identifid a number of drawbacks of conventional coded mask imagers in the context of searching generally for radiation sources in a terrestrial environment, for example: (i) The FOV of conventional coded mask imagers is much less than 180 degrees.

(ii) To provide both a reasonably large FOV and good spatial resolution requires a large number of detector pixels, with inherent cost and complexity issues; (iii) Conventional coded mask imagers provide relatively poor sensitivity to extended sources and only sources no larger than the instrument's angular resolution (in effect point sources) are measured with full sensitivity.

(iv) At high energies (> I MeV), the mask needs to be thick (several cm of high-Z material) to remain opaque, and so begins to act as a collimator, which degrades imaging performance.

(v) Reconstruction assumes that background is uniform over the whole detector, or may be reduced to that state by subtraction or scaling prior to deconvolution.

There is therefore a need for radiation detection schemes for determining a direction to a source of radiation in large area / stand-off scanning which are less affected by the above-mentioned drawbacks of conventional coded mask imaging approaches.

S

SUMMARY OF THE INVENTION

According to a first aspect of the invention there is provided an apparatus for determining a direction to a source of radiation, the apparatus comprising a rotatable frame, a plurality of radiation shields mounted to the frame and separated from one another by gaps to form an alternating series of radiation shields and gaps around a closed path, and a plurality of radiation detectors mounted to the frame and arranged to detect radiation passing through gaps between the radiation shields as the frame is rotated.

In providing a rotatable alternating series of radiation shields and gaps around a closed path, and a plurality of radiation detectors arranged to detect radiation passing through the gaps, a radiation imager is provided that may be seen as being broadly based on coded mask techniques, but which is sensitive to radiation from all azimuths in its plane of rotation. In general the plane of interest will be horizontal so that the axis of rotation of the frame will be vertical.

The frame might typically have a characteristic extent (e.g. a diameter in the case of a generally circular frame, or a side in the case of a square frame), on the order of one to ten, or so, metres. For example, the frame may have a characteristic extent of at least a dimension selected from the group comprising 0.3 m, 0.5 m, 1.0 m, 1.5 m, 2.0 in, 2.5 m and 3 m and / or a characteristic extent of up to a dimension selected from the group comprising 10 m, 8 m, 6 m, 5 m, 3 m, 2.5 m, 2 m, 1.5 m and I m. The closed path of the alternating series of radiation shields and gaps may have a similar range of dimensions.

Thus in one embodiment the frame and closed path may be generally circular with a diameter on the order of 2 m, and the apparatus may be mounted on a vehicle, e.g. a lorry or trailer. Thus the apparatus may be driven to a location where a scan is to be performed, and the frame rotated about a vertical axis while signals from the detectors are recorded in order to perform the scan.

The number of radiation shields in the alternating series of radiation shields and gaps may be around 10 or so, e.g. between 5 and 25 gaps. Different ones of the radiation shields and / or different ones of the gaps may have different extents along the path. For example, some shields and or some gaps may be 2, 3, 4, or 5 times larger

S

than other shields or gaps. Radiation shields having different lengths may comprise different numbers of shielding elements of a given size.

The shields and gaps may comprise roughly equal extents around the closed path. Shields and gaps may, for example, extend over characteristic azimuths of between 3 degrees and 30 degrees or more as measured in the plane of rotation of the frame and from its axis of rotation.

The radiation detectors -may be sensitive to radiation from all directions, but be shielded from radiation not passing through the gaps. For example, the radiation detectors may be shielded from radiation not passing through the gaps by respective ones of the radiation shields, e.g. by being placed adjacent to them. The radiation shields may comprise shielding element placed around three sides of a volume of a detector for example.

The radiation detectors may be gamma-ray radiation detectors or neutron radiation detectors, for example. In one gamma-ray imager example the detectors may comprise a scintillation body coupled to a photo-detector.

The apparatus may further comprise a processor operable to record detector signals from the radiation detectors as the frame is rotated. Thus the processor may be operable to determine a direction and / or distance to a source of radiation based on an analysis of differences in the radiation detected by the radiation detectors for different angles of rotation of the frame. The processor may further be operable to generate an energy-loss spectrum for radiation detected by the radiation detectors. This can allow the nature of the source of the radiation to be determined.

The apparatus may comprise a further plurality of radiation shields mounted to the frame, wherein the further plurality of radiation shields are separated from one another by gaps to form a further alternating series of radiation shields and gaps around a further closed path, and wherein the further alternating series of radiation shields and gaps is arranged at a different height relative to the frame compared to the first-mentioned alternating series of radiation shields and gaps. This can provide sensitivity to radiation sources at different elevations with respect to the plane of rotation of the frame.

S

Similarly, elevational sensitivity may also be provided where the plurality of radiation detectors comprises at least two groups of radiation detectors arranged at different heights relative to the frame.

According to a second aspect of the invention there is provided a method for determining a direction to a source of radiation comprising: providing a plurality of radiation shields separated from one another by gaps to form an alternating series of radiation shields and gaps around a closed path; providing a plurality of radiation detectors arranged to detect radiation passing through gaps between the radiation shields; and rotating the radiation shields and radiation detectors.

Thus in accordance with embodiments of the invention, a direction to a source of radiation, perhaps up to several hundred metres away, may readily be determined.

In some cases a source of radiation may be too weak to be reliably detected at such a range (although long integration times (e.g. a slow frame rotation, or stacked data from multiple frame rotations) may be used to improve statistical accuracy). In general it may be expected that radiation levels from special nuclear materials are likely to be relatively weak and so harder to detect at larger ranges. However, intense radioactive sources generally (i.e. not just special nuclear materials) can also present a serious risk, particularly if they can be stripped of their shielding and placed in areas where the general public can be exposed to serious, and even fatal, radiation doses. Of particular interest in this context might be those strong gamma-ray sources that are widely used in radio-therapy, food and blood irradiation, and in radio-thermal power generators (RTG5), for example. The latter may consist of intense alpha or beta emitters which may give rise to associated gamma-radiation. Embodiments of the present invention may be ideally suited to detecting this type of radiation source, for example. This detection and location process could, for example, be achieved by mounting the apparatus on a mobile system, such as a truck, aircraft (e.g. helicopter) or coastal vessel (e.g. boat or submarine).

S

BRIEF DESCRIPTION OF THE DRAWINGS

For a better understanding of the invention and to show how the same may be carried into effect reference is now made by way of example to the accompanying drawings inwhich: Figure 1 schematically shows in schematic cross-section a conventional coded mask imager; Figure 2 schematically shows in perspective view a radiation imager according to an embodiment of the invention; Figure 3 schematically shows in plan view the radiation imager of Figure 2; Figure 4 schematically shows in plan view features of the radiation imager of Figures 2 and 3 when in use; Figure 5 shows a variation in signal detected by a detector of the imager of Figures 2 to 4 as the imager is rotated; Figure 6 shows a point spread function of the imager of Figures 2 to 4; Figure 7 shows a point spread function of the imager of Figures 2 to 4 with data filtering; and Figure 8 schematically shows a plot of modelled signal-to-noise ratios for an imager in accordance with embodiments of the invention as a function of range for different radiation source strengths.

DETAILED DESCRIPTION

Figure 2 schematically shows in perspective view a gamma-ray imager 10 for determining the direction to a gamma-ray source according to an embodiment of the invention. Figure 3 schematically shows the gamma-ray imager ro in plan view from above. In this example it is assumed the imager 10 is to be used for detecting a source of gamma-ray radiation, e.g. Bremsstrahlung gamma-rays from a radio-thermal power generator.

The imager 10 comprises a platform 12, a plurality of radiation shields 14 mounted to the platform and separated from one another by gaps 15 to form an alternating series of radiation shields and gaps arounda circular path, and a plurality of radiation detectors 16 also mounted to the platform and arranged to detect radiation passing through gaps between the radiation shields. The platform in this example is generally horizontal and the imager 10 further comprises a motor (not shown) operable to rotate the platform 12 about a vertical axis through its centre, as schematically indicated by arrow 13.

In this example there are 23 radiation shields and the same number of detectors (this need not always be the case, for example, if one of the radiation detectors were missing, or broken, the imager 10 would continue to function but with a corresponding reduction in sensitivity). The radiation shield and detectors are located at the periphery of the platform, which in this example has a diameter of 1.5 m.

Each radiation detector 16 in this example is a conventional scintillator-based detector, e.g. comprising a 3 inch (7.5 cm) by 2 inch (5 cm) by 6 inch (15 cm) volume of thallium doped sodium iodide scintillator (schematically indicated in Figure 1 as a cuboids) optically coupled to a photo-multiplier tube (schematically indicated in Figure 1 as a tubes atop the cuboids). The detectors may, for example be oriented with their 3 inch (7.5 cm) by 2 inch (5 cm) faces aligned horizontally, and their 3 inch (7.5 cm) by 6 inch (15 cm) faces aligned azimuthally with respect to the rotational axis of the frame.

Each radiation shield 14 comprises a 0.5 inch (-1cm) layer of lead arranged around three sides of a corresponding scintillation body of one of the radiation detectors 16. Thus only a side of the respective radiation detectors 16 facing towards

S

the inside of the ring of radiation detectors / shields is exposed. In some embodiments the top and/or bottom of the radiation detectors may also be shielded. However, in this example it is assumed the application at hand is such that any radiation sources are expected to be only at ground level (the level of the imager), and so shielding from above and or below is not used. In some other situations it may be considered more likely that there will be radiation sources below the elevatiorial field of view of the imager, than above it. In which:.case it may be considered appropriate to shield the bottoms of the detectors to reduce the amount of radiation the received from sources that are not within the imager's filed of view (and hence un-modulated by the mask).

The radiation shields 14 provide two shielding functions.

Firstly, each radiation shield acts to shield its respective radiation detector from radiation in the horizontal plane in such a way that the detector is only sensitive to radiation that has first entered the ring of radiation detectors / shields through the opposing side of ring. That is to say, the radiation shields act to shield the detectors from radiation coming from "behind" them. (For the purposes of explanation, reference will sometimes be made to directions which are "in-front" of or "behind" a detector (and similar / related terms). It will be understood that "in-front" of is used to indicate the direction from the detector towards the insider of the ring of shields, and "behind" is used to indicate the opposite direction.) Secondly, the radiation shields act to provide a one-dimensional coded mask for detectors on an opposing side of the ring. In a coded-mask analogy, the radiation shields 14 thus provide a one-dimensional mask arranged around a closed (in this case circular) path, and each detector may be seen as a pixel of the detector, the pixels being sensitive only to radiation that has passed through gaps in the mask in-front of them. The radiation shields will not necessarily absorb all the energy of an incident radiation event. Some radiation may traverse a radiation shield and deposit energy in the associated detector. Thus the detectors themselves also contributed to the opacity of the mask provided primarily by the shields. What is more, it is possible that an energetic radiation event may penetrate a radiation shield, pass through the associated detector, and then on into another detector. Such an event contributes to the count rate in the detectors, but is not indicative of radiation that has passed through a gap in the mask, and so would preferably be discarded. Coincidence techniques can be used to do this. For example, if a count is registered in two detectors at the same time (i.e. within a certain time window), it may be assumed that a-single radiation event is responsible, and so the radiation event is not representative of radiation that has passed through a gap, and so should be discarded, e.g. in accordance with the general principles of coincidence-based rejection schemes.

In the example shown in Figure 1, the circular "mask" comprises II opaque sections (two of an extent corresponding to the width of four radiation shield / radiation detector units, one corresponding to three radiation shield / radiation detector units, and four each corresponding to two and one radiation shield / radiation detector units). The gaps between the shields similarly vary in size. The design of the placement of radiation shield / radiation detector units and the gaps between them around the closed path may be in accordance with the general and well understood principles of coded mask designs for ID masks, except in embodiments of the invention such a 1D mask is folded around on itself to form a closed path. For example, the mask may be based on design considerations corresponding to those described by Fenimore [4] and Fenimore & Cannon [3, 5].

A purpose of the apparatus shown in Figure 1 is to determine the directions towards sources of radiation within the environment surrounding the imager 10. To assist in explaining the operation of the imager 10 in this regard, a spherical polar coordinate will be defined. This co-ordinate system is fixed to the environment, that is to say, it does not rotate with the imager 10. The co-ordinate system has an origin at the centre of the ring of radiation shields! detector units 14;16. This is on the axis of rotation of the imager 10, and a short distance above the centre of the platform 12.

Locations in the environment within the horizontal plane (i.e. the plane containing the plane of rotation of the imager) are identified by a range R and an azimuth 0, measured clockwise when viewed from above. Azimuth 0 requires an arbitrary zero to be defined. Here it is assumed that zero is taken to be in a northerly direction, e.g. as measured at the imager location using a compass. In other examples, a different arbitrary zero direction may be defined. For example, the imager 10 may be mounted horizontally on the rear of a lorry, and zero azimuth may be defined as a direction from the origin of the coordinate system toward the front of the vehicle. The azimuth of the platform 12 (and hence the shields 14 and detectors 16 mounted thereto)

S

changes as the platform is rotated during use. The changing azimuth of the platform as it rotates (Opiaffo) may be defined by the azimuthal angle of anarbitrar.ily located rotation mark 20 on the platform.

The positions of locations outside the-R, 0 plane are defined by an elevation angle 4, taken to be positive for locations above the R, 0 plane. The imager 10 is operable to determine the azimuth °source of a source of radiation within an operating.

elevational range +1-ttrange (the imager in this example is not sensitive to the elevation angle source of a source within this range). The elevational range is primarily defined by the angular height of the radiation shields on one side of the ring as seen by a detector on the other side. If a source is outside this range, radiation from the source is not modulated by the shields (i.e. the radiation does not pass through the mask). In the example shown in Figure 2, the elevational range is around +7-3 degrees. Thus the imager of Figure 2 mounted horizontally to the back of a lorry will generally be sensitive to sources of radiation near ground level. It will be appreciated that this elevational sensitivity range is a broad guide only. In practice the elevational sensitivity range at each azimuth will not be a simple angular cone from the origin of the coordinate system. This is because the respective detectors have finite extent, and are also not at the origin of the coordinate system themselves. However, given the relatively small size of the imager 10 compared to the range of distances scanned, these geometric effects are unlikely to be significant in practical terms.

An example of how the imager 10 may be used to determine the azimuth of a source is now described.

Figure 4 schematically shows the imager 10 of Figures 2 and 3 in an environment in which a single gamma-ray source 24 is located. The direction of North (parallel to zero azimuth) is indicated to the left of the figure. The imager is shown in an initial configuration with the platform stationary at azimuth Oplatform 0 degrees (i.e. rotation mark 20 at zero azimuth / directly North of platform centre). The gamma-ray source 24 is at an azimuth °source of around 30 degrees, and furthermore is taken to be within the elevational sensitivity range of the imager, e.g. it is on the ground. In this example the gamma-ray source is taken to be sufficiently far away from the imager (compared to the characteristic size of the imager) that gamma-ray radiation from the source 24 is quasi-parallel at the imager. For example, the source may be at a range Rsoure = 50 metres. Also schematically shown in Figure 4 are five gamma-rays from the source 24, labelled ?l.5.

The various detectors 16 of the imager 10 are only sensitive to radiation sources within various specific cones defined by the gaps between the shields in-front of them (i.e. towards the opposite side of the ring). Each detector thus has a different range of "sensitivity cones", and furthermore, these rotate with the imager. The sensitivity cones for the southernmost detector (labelled 26) in Figure 4 in the initial platform orientation are schematically shown by shading. There are in total seven gaps in the mask provided by the ring of radiation shields 14 that detector 26 can "see" through. These gaps subtend different azimuthal extents at the detector 26, and so the detector 26 will be differently sensitive to sources visible through different gaps. In the exampe shown in Figure 4, the detector 26 will detect radiation from the source because the source is within the angular extent of one of its sensitivity cones for this platform azimuth (the rightmost sensitivity cone shown shaded in the figure). E.g.

schematic gamma-ray y4 will pass through the gap labelled 27, and be detected by the detector 26.

The detection of radiation by detector 26 indicates that in this initial configuration there is a source within one of the sensitivity cones of detector 26, but it is not possible with only this information to say which one. Other detectors may also detect radiation in this configuration, e.g. it can be seen that the detectors labelled 28 and 30 are likely to detect some radiation passing through the gap near to the rotation mark 20 in the figure. The set of detectors receiving radiation in this initial static configuration may thus by themselves be used to provide an indication of the source direction since only certain potential source directions will give rise to certain combinations of detectors receiving radiation simultaneously, the combinations being defined by the mask pattern provided by the shields and gaps. However, improved results may be obtained by rotating the imager and recording the signals detected by the detector 24 as a function of the rotation angle of the platform Oplatfo. For example, the platform may be rotated once in a 36 minute period, and detector signals binned into 30 integration periods (corresponding to an angular binning of 5 degrees). The same degree of signal integration may of course be achieved by rotating the platform faster multiple times (e.g. 36 rotations of one minute period) with appropriate stacking

S

of data from corresponding azimuth bins. In practice the amount of integration time will be determined by the strength of source expected, and Ihe desired angular resolution (e.g. longer integration times mean data can be binned into smaller angular ranges while maintaining sufficient levels of signal-to-noise).

Figure 5 shows an example plot of detected signal D at an arbitrary one of the detectors (the detector indicated in Figure 4 by reference numeral 29) for an arbitrarily-selected source direction (source at azimuth -285 degrees) as the imager 10 shown in Figure 4 is rotated. When the platform is at azimuth 0 degrees (as in Figure 4), the detector 29 associated with the data in Figure 5 is at an azimuth of around 135 degrees. Thus when the platform is at azimuth 0 degrees the detector 29 is more or less directly facing the source, and so this is the platform azimuth at which the detector sees the highest signals. The detected signal is plotted on a scale normalised to the maximum signal seen during the rotation as a function of the azimuth Opiatfo of the platform. Similar plots may also be generated for each of the other detectors, e.g. providing a total 23 similar plots in this example.

The variation in detector signal with platform rotation seen in Figure 5 has three main component, namely a relatively high frequency / chopping component C, a relatively low frequency background component B, and a relatively low frequency projection component V. The relatively high frequency / chopping component C is caused by the source going in and out of the sensitivity cones associated with the detector as the platform rotates. This is the component of primary interest as it indicates at what stage of the rotation the source can be seen by the detector through the various ones of the gaps.

Furthermore, the specific pattern of the chopping peaks can be matched to the pattern in the variations in gap sizes seen by the detector to determine which of the gaps is exposing the detector to the source at which stages of rotation, thereby providing an indication of the direction to the source. The high frequency / chopping component C is most apparent over the range broadly indicated in Figure 4 by the arrow marked "C". The pattern occurs predominantly at azimuth angles of the platform for which the detector for which data are plotted is an azimuth angle of 180 degrees away from the source direction (i.e. when the detector is "facing" the source). The chopping pattern is not apparent when the detector is at broadly the same azimuth as the source because at this azimuth the detector has its "back" to the source, and so does not see the source through any of the gaps-(i.e. the sensitivity cones for the detector are all pointing away from the source).

The low frequency projection component V occurs for platform azimuths at which the source is detectable by the detector through a gap, but where the gap is to one side of the detector (e.g. a gap separated from the detector by an azimuth difference of less than 180 degrees), as opposed to being diametrically opposite. This low-frequency signal modulation arises as the effective sensitive surface of the detector (i.e. its front face) is in effect reduced in size by projection effects. Projection effects are primarily apparent in Figure 5 over the ranges indicated in the figure by the arrows V. A further effect in this range is vignetting caused by the finite thickness of the detector / shield units, and the oblique presentation of the gaps to the detector, reducing the effective extent of the gaps that are off to one side. For example, referring to Figure 4, the sensitivity cone that is most directly in-front of the detector 26 and passing through the gap in the vicinity of the rotation mark 20 and the left-most sensitivity cone are both defined by broadly the same gap extent. However, vignetting associated with left-most cone of sensitivity means this cone is narrower than the forward facing cone, and so is associated with a thinner peak in the chopped signal.

The low frequency background component B peaks around platform azimuths for which the detector for which data are being plotted is at the same azimuth as the source. (In this example when the platform is at azimuth of 180 degrees in its rotation, at which point the detector 29 is at an azimuth of around 285 degrees). This background component arises because the radiation shields are not 100% effective in stopping radiation when the detector has its back to the source, and so a background signal is detected. The background signal is primarily significant over the ranges indicated in the figure by the arrows B. The background signal peaks when the source is directly behind the detector because the shielding at the back of the detector here is normal to the source and so has its smallest effective thickness. At low source energies, this effect is reduced as shield penetration becomes less significant.

A number of image reconstruction algorithms for determining a direction to the source from data obtained during a scan, e.g. such as shown in Figure 5, from the various detectors have been investigated.

S

For example, it can be envisaged that during a complete rotation of the platform, each of the detectors produces a count rate profile similar to that shown in Figure 5, but with a different detailed structure reflecting the different arrangements of shields and gaps that face the respective detectors (i.e. the different-parts of the mask pattern provided by the shields that are opposite the respective detectors). Any one of a number of different analysis methods could be used in accordance with the know principles of coded mask image reconstruction. So far as the image reconstruction processing is concerned, the data may be treated largely as if it had come from a conventional ID mask using a conventional flat detector array (i.e. the fact that the detector is in fact curved around on itself can in large part be ignored so far as the basic principles of the image reconstruction processing algorithms are concerned. Thus largely conventional processing techniques may be employed, e.g. using the detector's respective azimuths in the ring as proxies for positions in an "unwrapped" flat detector plane.

For example, in a basic analysis, a count rate history for each detector may be cross-correlated with an expected count rate history for that detector from a single point source. The expected count rate profile may be obtained using Monte-Carlo modelling of the instrument, for example. The phase of the correlation peak provides a measure of the location of the source(s) in azimuthal angle, and the strength of the correlation provides a measure of the intensity of the source(s). Each correlation output may then be shifted (rotated) into the same angular co-ordinate system with reference to the platform (i.e. shift each correlation plot by an amount based on the azimuths of the respective detectors relative to the zero rotation mark 20), and the correlation outputs from each detector may be summed to produce a final image. An example of such an image for a single point source at an azimuth 180 degrees which is imaged in this way is schematically shown in Figure 6. Figure 6 plots integrated correlation strength (converted to radiation flux in arbitrary units) versus azimuth. The peak at azimuth 180 degrees associated with the point source is very clear. The relatively broad shoulders I side-lobes to the peak are associated with effects caused by the low-frequency projection and background components described above with particular reference to Figure 5.

Thus in a more sophisticated analysis scheme, the recorded and expected count rate histories may be convolved with a high-pass filter function before cross- correlation. Thus the filter function acts as a band-pass filter, allowing only the high-frequency chopping component introduced by the mask (component C indicated in Figure 5) to contribute to the final image. Such filtering can assist in reducing the shoulders / side-lobes from the imager's point spread function (seen in Figure 6), and so produce a sharper image. This can help in allowing weaker sources to be detected.

An example of an image of a single point source imaged in this way is given in Figure 7. Figure 7 is similar to an will be understood from Figure 6, but shows how the high-pass filtering of the count histories for the detectors versus platform azimuth (i.e. filtering data such as are plotted in Figure 5), can provide a still sharper image.

In some embodiments, spectral analysis, energy reconstruction and filtering may be applied to the recorded detector count rate profiles, e.g., prior to the application of a filter function. In this way, clean images can be reconstructed for a single energy of interest, or over a restricted range of energies, e.g., to help avoid contamination in the data from known sources, or specific background isotopes.

It is noted that a consequence of the rotating imager design with the mask and detector arrays wrapped around on themselves is that the whole field of view is in effect only partially coded (i.e. none of the detectors can see through all of the gaps), and rear shield leakage and vignetting can add complications to the response.

Convolution-based image reconstruction approaches in particular are relatively robust to these issues.

An imager design according embodiments of the invention has been evaluated using Monte-Carlo simulations that include the effects of increased attenuation by air and the geometric reduction in flux seen at the imager as a source is considered to be further and further away. This is to investigate the range over which the direction to a source can be determined, The limiting distance at which a source can be detected occurs when the modulated mask pattern (high frequency chopping C in Figure 5) is not significantly visible above the background. The ability to see weaker sources can be improved by analysing only a small energy range around the source energy of interest.

S

Figure 8 shows a series of plots of predicted signal-to-noise ratio (SNR) as a function of source range R for different source strengths for the imager 10 shown in Figure 2. SNR is defined here as ratio of the peak intensity ina modelled plot of the kind shown in Figure 7 to the root mean square variation in the background away from the peak.

The example source is a 137 Caesium source (emitting primarily at 662 keV), and curves are shown-for source strengths of 1, 5, 10, 20, 40, 60 and 80 mCi. A data integration time of six minutes is assumed with a single rotation of the imager (i.e. rotation rate of I degree / second). The dashed horizontal line indicates a signal-to-noise ratio of three sigma.

As seen in Figure 8, detection of a lmCi source is still possible at a range of around lOOm with SNR> 3. Stronger sources are still more readily detectable at this range. It may be noted from Figure 8 that the modelled SNR at the shortest ranges is less than at higher ranges (e.g. compare the SNR values for ranges R around 20m or so with values for ranges R around 40m), which is perhaps counter-intuitive. However, this is simply an artefact of the modelling. The image reconstruction process used here relies on comparing / correlating observed data with expected data. The expected data are all modelled here for a source at a range of 100 m. Thus where the source is in reality much closer, the expected data are a less good representation of what is truly to be expected, and as a consequence the SNR is reduced. Indeed, in some embodiments data from a scan may be processed by correlating with different expected results for different source ranges. Whichever comparison provides the greatest apparent SNR gives an indication of an estimate to the range to the source. Thus a single scan can be used to provide an estimate of source range, as well as azimuth.

It can also be seen from Figure 8 that the curves for the stronger sources tend to flatten out at signal-to-noise ratios of around 85. This indicates that source detection in this range is dominated by systematic effects (i.e. in the imaging system), rather than photon statistics (since for count-rate dominated noise the SNR would fall with increasing range for a given source strength). This in turn indicates that even though the imager shown in Figure 2 can provide excellent performance, further improvements may still be achievable in aspects of the design. For example, performance might be improved further still by increasing this characterising size of

S

the imager (e.g. to increase angular resolution), including more and / or larger detectors (to improve sensitivity), and applying more advanced image reconstruction techniques, e.g. with optimised filtering characteristics.

In embodiments of the invention, the arrangement of the shields and detectors around the circumference of the system may be governed by the desired angular-resolution and the characteristic scale D of the imaging system (e.g. diameter of platform). If the characteristic width -of each detector / shield unit in an azimuthal direction is w, then the intrinsic angular resolution of the system, will be on the order w/D radians. In practice, for detection levels exceeding the 3 limit (i.e. SNR> 3), the source localisation that can be achieved improves linearly with the significance of the detection.

As noted above, the coding pattern used (i.e. the specific series of shield and gap sizes around the imager) may be selected in accordance with the general principles of coded aperture design, e.g. for uniformly-redundant coded-aperture masks, such as described by Gottesman [8]. For example, the use of a system based on 23 detectors each 7.6 cm wide and mounted on the circumference of a circle having a diameter of 1.1 m, would provide an intrinsic resolution of -0.076 radians. This corresponds to an uncertainty in the location of a weak source (e.g. at the limit of detection) at I OOm of metres.

Thus a detector system / imager in accordance with some embodiments of the invention comprises a number of combined detector and shield units arranged in a cylindrical geometry, the cylinder being rotatable about its axis. The arrangement of the pixels (radiation detectors) provides spatial encoding of a source position, and this is further enhanced by temporal coding provided by rotating the imager. A single observation provided by rotating the imager and recording detector signals for the different angles of rotation may be processed to provide an azimuthal map of the source distribution. Combinations of multiple observations / scans from different locations may be used to provide a full reconstruction of source locations within a 360° field of view in a horizontal plane using well known triangulation techniques. In some embodiments of the invention, the imager may be capable, for example, of locating a mCi caesium 137 gamma-ray source to a precision of around 20cm at a distance of

S

lOOm, and also to detect and identify the presence of a source perhaps as weak as I mCi at that distance.

Whilst the above examples employ a single ring of shields / detectors, another embodiment may include a second ring of detectors and I or a second ring of shields at a different height above the platform to provide sensitivity to different elevational ranges. For example an embodiment similar to that show in Figure 2 may comprise a broadly similar second layer of shields and detectors having a different mask pattern (i.e. different shield and gap lengths). If detectors in the first layer detect chopping signals as the imager is rotated which are characteristic of the mask encoding of the first layer, this indicates the source is broadly within the horizontal plane (it can be expected in this case the detectors in the second layer will detect chopping signals as the imager is rotated which are characteristic of the mask encoding of the second layer). However, if detectors in the first layer detect chopping signals as the imager is rotated which are characteristic of the mask encoding pattern of the second layer, this indicates the source is above the horizontal plane. If, on the other hand, detectors in the second layer detect chopping signals as the imager is rotated which are characteristic of the mask encoding pattern of the first layer, this indicates the source is below the horizontal plane. This feature might be particularly useful in a marine environment Thus providing two (or more) layers of detectors and / or shields can provide a degree of sensitivity to a sources elevation.

In addition to identifying the azimuthal direction to a source and I or range (e.g. using triangulation based on azimuth measurements from two locations), the nature of the source may be determined from a spectral analysis of the energy loss in the detectors. The spectral information in the energy loss spectra can be used to identify the nature of the source in the usual way (e.g., by identifying features in the spectra which are characteristic of a given radioactive material). This can be done using conventional techniques, for example spectrum deconvolution such as described in WO 02/03 1536 [9]. The energy loss spectra from the various radiation detectors may be processed separately, or after summing. Summing can be useful if the overall count rates are low because summing can help to reduce the statistical noise. Summing will be most effective where the responses of each of the scintillation bodies are first normalised. Normalisation can help to ensure the summed energy loss spectrum is close to that which would be obtained from a spectrometer comprising a single scintillation body with a volume comparable to that of the separate radiation-detectors taken together. This can help to optimise the ability to identi1' isotopes from the spectrum. However, some processing techniques, e.g. those described in WO 02/031536 [9], take account of the modelled responses of the individual scintillation bodies, and so in these cases it may be preferable-for the spectral processing to be performed separately for each-detector.

It will also be understood that while the above example has focussed on a gamma-ray imager based on scintillation body gamma-ray detectors, other kinds of gamma-ray detector could equally be used, Furthermore, the same principles could be used for detecting other sources of radiation, e.g. sources of neutrons using neutron radiation detectors and neutron shields as appropriate.

It will be appreciated that although a disk-like platform / frame with shields and detectors mounted thereon is shown in the examples described above, the platform may have a different structure. For example, the platform may be above the shields and detectors with the shields and detectors depending therefrom. In examples having more than one layer of shields and or detectors (e.g. to provide elevational sensitivity out of the plane of rotation), different layers may be mounted above and below the platform. In other examples, the platform may be provided by a framework connecting the various shields and detectors together such that the shields and / or detectors themselves play a structural role in providing the platform. For example, the platform may comprise the shielding elements held together by a structural framework bridging the gaps between them.

Thus an apparatus for determining a direction to a source of radiation, e.g. a gamma-ray or neutron source at a range of 100 m or more, is described. The apparatus is rotatable, e.g. about a vertical axis, and comprises a plurality of radiation shields and radiation detectors. The radiation shields are separated from one another by gaps to form an alternating series of radiation shields and gaps around a path within the plane of rotation for the apparatus. The radiation detectors are arranged to detect changes in radiation passing through gaps between the radiation shields as the apparatus is rotated. In providing a rotatable series of radiation shields and gaps with radiation detectors arranged to detect radiation passing through the gaps in this way, a radiation imager is provided that may be seen as being broadly based on coded mask techniques, but which is sensitive to radiation from allazimuths in its-plane of rotation.

S

REFERENCES

[1] US 2006/0284094 (Inbar) [2] Larssen, C., L., & Djeffal, S., Development of a Directional Gamma Ray Probe,. Nuclear Science Symposium Conference Record 2005, II2EEE, Volume1,pages 16-18 [3] E. Fenimore & T. M. Cannon, Applied Optics, Vol 17 No 2, p337, 1978 [4] E. Fenimore, Applied Optics, Vol 17 No 22, p 3562, 1978 [5] E. Fenimore & T. M. Cannon, Applied Optics, Volume 20 No 10, p1858, [6] Ziock, K. P., Collins, J. W., Fabris, L., Gallagher, S., Horn, B. K. P., Lanza, R. C. & Madden, N. W., Source-Search Sensitivity of a Large-Area, Coded-Aperture, Gamma-Ray Imager, IEEE Transactions on Nuclear Science, Volume 53, No. 3, June 2006.

[7] A. Goldwurm et at, Astronomy & Astrophysics, 227, p 640, 1990 [8] Gottesman, S., & Pen imore, E., Applied Optics 28, pp 4344-43 52, 1989 [9] W002/031536

Claims

CLAIMSI. Apparatus for determining a direction to a source of radiation comprising:a rotatable frame;a plurality of radiation shields mounted to the frame and separated from one S another by gaps to form an alternating series of radiation shields and gaps around a-closed path; and a plurality of radiation detectors mounted to the frame and arranged to detect radiation passing through gaps between the radiation shields as the frame is rotated.
2. An apparatus according to claim 1, wherein at least some of the radiation shields have different extents around the path.
3. An apparatus according to claim 2, wherein radiation shields having different extents comprise different numbers of shielding elements.
4. An apparatus according to any of claims I to 3, wherein at least some of the gaps have different extents around the path.
5. An apparatus according to any preceding claim, wherein the radiation detectors are sensitive to radiation from all directions and are shielded from radiation not passing through the gaps.
6. An apparatus according to claim 5, wherein the radiation detectors are shielded from radiation not passing through the gaps by respective ones of the radiation shields.
7. An apparatus according to claim 6, wherein respective ones of the radiation detectors are adjacent respective ones of the radiation shields
8. An apparatus according to any preceding claim, wherein the radiation detectors are gamma-ray detectors.
9. An apparatus according to claim 8, wherein_the radiation detectors-comprise a scintillation body coupled to a photo-detector.
10. An apparatus according to any preceding claim, wherein the radiation detectors are neutron detectors.
11. An apparatus according to any preceding claim, further comprising a processor operable to record detector signals from the radiation detectors as the frame is rotated.
12. An apparatus according to claim 11, wherein the processor is further operable to determine a direction to a source of radiation based on an analysis of differences in the radiation detected by the radiation detectors for different angles of rotation of the frame.
13. An apparatus according to claim 11 or 12, wherein the processor is further operable to determine a distance to a source of radiation based on an analysis of differences in the radiation detected by the radiation detectors for different angles of rotation of the frame.
14. An apparatus according to any of claims 11 to 13, wherein the processor is further operable to generate an energy-loss spectrum for radiation detected by the radiation detectors.
15. An apparatus according to any preceding claim, further comprising a further plurality of radiation shields mounted to the frame, wherein the further plurality of radiation shields are separated from one another by gaps to form a further alternating series of radiation shields and gaps around a further closed path, and wherein the further alternating series of radiation shields and gaps is arranged at a different height relative to the frame compared to the first-mentioned alternating series of radiation shields and gaps.S16. An apparatus according to any preceding claim, wherein the plurality of radiation detectors comprise at least two groups of radiation detectors arranged at different heights relative to the frame.17. A method for determining a direction-to a source of radiation comprising: providing a plurality of radiation shields separated from one another by gaps to form an alternating series of radiation shields and gaps around a closed path.; providing a plurality of radiation detectors arranged to detect radiation passing through gaps between the radiation shields; and rotating the radiation shields and radiation detectors.18. Apparatus for determining a direction to a source of radiation substantially as hereinbefore described with reference to Figures 2 to 9 of the accompanying drawings.19. A method for determining a direction to a source of radiation substantially as hereinbefore described with reference to Figures 2 to 9 of the accompanying drawings.Amendments to the Claims have been filed as follows CLAIMS 28 1. Apparatus for determining a direction to a source of radiation comprising:a rotatable frame;a plurality of radiation shields mounted to the frame and separated from one another by gaps to form an alternating series of radiation shields and gaps around a closed path; and a plurality of radiation detectors mounted to the frame and arranged to detect radiation passing through gaps between the radiation shields as the frame is rotated.2. An apparatus according to claim 1, wherein at least some of the radiation shields have different extents around the path.3. An apparatus according to claim 2, wherein radiation shields having different extents comprise different numbers of shielding elements.4. An apparatus according to any of claims 1 to 3, wherein at least some of the gaps have different extents around the path.* : 5. An apparatus according to any preceding claim, wherein the radiation detectors are sensitive to radiation from all directions and are shielded from radiation not * :* . passing through the gaps.* 6. An apparatus according to claim 5, wherein the radiation detectors are shielded : *" from radiation not passing through the gaps by respective ones of the radiation shields. **** * 25 *a* S* 7. An apparatus according to claim 6, wherein respective ones of the radiation detectors are adjacent respective ones of the radiation shields 8. An apparatus according to any preceding claim, wherein the radiation detectors are gamma-ray detectors.S9. An apparatus according to claim 8, wherein the radiation detectors comprise a scintillation body coupled to a photo-detector.10. An apparatus according to any preceding claim, wherein the radiation detectors are neutron detectors.11. An apparatus according to any preceding claim, further comprising a processor operable to record detector signals from the radiation detectors as the frame is rotated.12. An apparatus according to claim 11, wherein the processor is further operable to determine a direction to a source of radiation based on an analysis of differences in the radiation detected by the radiation detectors for different angles of rotation of the frame.13. An apparatus according to claim 11 or 12, wherein the processor is further operable to generate an energy-loss spectrum for radiation detected by the radiation detectors.14. An apparatus according to any preceding claim, further comprising a further ::::. 20 plurality of radiation shields mounted to the frame, wherein the further plurality of radiation shields are separated from one another by gaps to form a further alternating series of radiation shields and gaps around a further closed path, and wherein the *:. further alternating series of radiation shields and gaps is arranged at a different height * relative to the frame compared to the first-mentioned alternating series of radiation shieldsandgaps.****** * * 15. An apparatus according to any preceding claim, wherein the plurality of radiation detectors comprise at least two groups of radiation detectors arranged at different heights relative to the frame. S 3016. A method for determining a direction to a source of radiation comprising: providing a plurality of radiation shields mounted to a frame and separated from one another by gaps to form an alternating series of radiation shields and gaps around a closed path; providing a plurality of radiation detectors arranged to detect radiation passing through gaps between the radiation shields; rotating the radiation shields and radiation detectors, and determining a direction to the source of radiation based on an analysis of differences in radiation detected by the radiation detectors for different angles of rotation of the frame.17. Apparatus for determining a direction to a source of radiation substantially as hereinbefore described with reference to Figures 2 to 8 of the accompanying drawings.18. A method for determining a direction to a source of radiation substantially as hereinbefore described with reference to Figures 2 to 8 of the accompanying drawings. *.S. * * S *S I S... * * I...IS S * S S * *S *55I * S. * S S 55SIS 1.555 5