MEDICAL IMAGE PROCESSING DEVICE, COMPUTER PROGRAM, AND NUCLEAR MEDICINE DEVICE

Abstract

An image is reconstituted by iterative approximation, a PET event updated image is produced by updating a current image using a PET event, a Compton event updated image is produced by updating the current image using a Compton event, the PET event updated image and the Compton event updated image that have been independently produced are weighted and added together, and the current image is updated using an image obtained by addition processing. In this way, PET events and Compton events, which have different properties, can be used in combination to efficiently and stably reconstitute images, improving image quality.

Claims

1. A medical image processing device for reconstructing an image by iterative approximation using a PET event where a coincidence signal of a pair of annihilation radiations is obtained and a Compton event obtained by Compton scattering, the medical image processing device comprising: a PET event updated image production unit configured to produce a PET event updated image by updating a current image using a PET event; a Compton event updated image production unit configured to produce a Compton event updated image by updating the current image using a Compton event; an addition unit configured to weight and add the PET event updated image and the Compton event updated image produced independently of each other; an update unit configured to update the current image using an image obtained by the addition unit; and an iteration unit configured to iterate processing of the PET event updated image production unit, the Compton event updated image production unit, the addition unit, and the update unit.

2. The medical image processing device according to claim 1, wherein at least either the PET event updated image production unit or the Compton event updated image production unit is configured to set a number of subsets as an update parameter.

3. The medical image processing device according to claim 2, wherein at least either the PET event updated image production unit or the Compton event updated image production unit is configured to make a sub iteration of image update using the subsets.

4. The medical image processing device according to claim 2, wherein the number of subsets that is the update parameter, a number of times of sub iterations, and timing of weighted addition can be set for the PET event and the Compton event independently.

5. The medical image processing device according to claim 1, wherein the Compton event includes an annihilation radiation Compton event and a single gamma-ray Compton event.

6. The medical image processing device according to claim 1, wherein the PET event includes a PET event between absorber detectors, a PET event between a scatterer detector and an absorber detector, and a PET event between scatterer detectors.

7. The medical image processing device according to claim 1, wherein the PET event includes a PET event with time of flight information and a PET event without time of flight information.

8. The medical image processing device according to claim 1, wherein, if a nuclide is a 3-gamma-ray nuclide, a 3-gamma event is further included.

9. A non-transitory computer readable recording medium for recording a computer program for causing a computer to implement the medical image processing device according to claim 1.

10. A nuclear medicine device comprising: a PET-Compton simultaneous measurement device including scatterer detectors and absorber detectors; and the medical image processing device according to claim 5.

11. The nuclear medicine device according to claim 10, wherein at least either the scatterer detectors or the absorber detectors are arranged in a ring shape, a partial ring shape, or an opposed shape.

12. The nuclear medicine device according to claim 10, wherein the scatterer detectors are arranged in a multi-ring shape.

13. The nuclear medicine device according to claim 10, wherein the scatterer detectors are located inside a measurement field of view of the absorber detectors.

14. The nuclear medicine device according to claim 10, wherein the scatterer detectors are located outside a measurement field of view of the absorber detectors.

15. The nuclear medicine device according to claim 10, wherein a measurement field of view of the Compton event is made greater than a measurement field of view of the PET event.

16. The nuclear medicine device according to claim 15, wherein a pixel size of the measurement field of view of the Compton event is made greater than that of the measurement field of view including the PET event.

17. A nuclear medicine device comprising: a PET-Compton event simultaneous measurement device configured to simultaneously measure a PET event and a Compton event; a single event collection device configured to collect single events from output of the PET-Compton event simultaneous measurement device; a software coincidence device configured to collect PET events from output of the PET-Compton event simultaneous measurement device; a hybrid image reconstruction device configured to reconstruct a hybrid image of the PET events and the Compton events on a basis of output of the single event collection device and the software coincidence device, the hybrid image reconstruction device including the medical image processing device according to claim 1; a control device configured to control the PET-Compton event simultaneous measurement device, the single event collection device, the software coincidence device, and the hybrid image reconstruction device; and a display and operation controller.

Description

BRIEF DESCRIPTION OF DRAWINGS

[0033] FIG. 1 is a diagram showing a first technique for hybrid image reconstruction according to an embodiment of the present invention.

[0034] FIG. 2 is a diagram showing a second technique for the hybrid image reconstruction according to the embodiment of the present invention.

[0035] FIG. 3 is a diagram showing equations for implementing the technique of FIG. 2.

[0036] FIG. 4 is a diagram showing a third technique for the hybrid image reconstruction according to the embodiment of the present invention.

[0037] FIG. 5 is a diagram showing a fourth technique for the hybrid image reconstruction according to the embodiment of the present invention.

[0038] FIG. 6 is a diagram showing equations Los implementing the technique of FIG. 5.

[0039] FIG. 7 is a diagram showing a fifth technique for the hybrid image reconstruction according to the embodiment of the present invention.

[0040] FIG. 8 is a diagram showing a sixth technique for the hybrid image reconstruction according to the embodiment of the present invention.

[0041] FIG. 9 is a diagram showing a seventh technique for the hybrid image reconstruction according to the embodiment of the present invention.

[0042] FIG. 10 is a diagram showing an eighth technique for the same.

[0043] FIG. 11 is a diagram showing classification of measurement events in a PET-Compton event simultaneous measurement device used in the embodiment of the present invention.

[0044] FIG. 12 is a diagram showing classification of annihilation radiation Compton events that can be handled as PET events used in the same device.

[0045] FIG. 13 is a flowchart of an algorithm for extracting a PET event, an annihilation radiation Compton event, and a single gamma-ray Compton event on the basis of time information and energy information about single events measured by scatterer detectors and absorber detectors used in the same device.

[0046] FIG. 14A is a cross-sectional view showing an example of a double-ring detector layout for carrying out the present invention.

[0047] FIG. 14B is a cross-sectional view showing an example of a multi-ring detector layout for the same purpose.

[0048] FIG. 15A is a longitudinal sectional view showing an example where a scatterer detector ring is disposed inside an absorber detector ring in a body axis direction for the same purpose.

[0049] FIG. 15B is a longitudinal sectional view showing an example where scatterer detector rings are disposed outside an absorber detector ring for the same purpose.

[0050] FIG. 16A is a cross-sectional view showing an example of detector layout with a partial scatterer detector ring for the same purpose.

[0051] FIG. 16B is a cross-sectional view showing an example of detector layout with partial detector rings for the same purpose.

[0052] FIG. 16C is a cross-sectional view showing an example of opposed detector layout for the same purpose.

[0053] FIG. 17A is a sectional view showing as example of a scintillation detector used as a detector.

[0054] FIG. 17B is a sectional view showing an example of a DOI detector used as a detector.

[0055] FIG. 17C is a sectional view showing an example of a semiconductor detector used as a detector.

[0056] FIG. 18A is a sectional view showing a configuration example of multi-ring detectors using semiconductor detectors for carrying out the present invention.

[0057] FIG. 18B is a sectional view showing a configuration example of multi-ring detectors using scintillation detectors for the same purpose.

[0058] FIG. 19 is a horizontal sectional view showing an example where pixels outside a range where PFT events can be measured are increased in size to reduce the amount of calculation for image reconstruction according to the embodiment of the present invention.

[0059] FIG. 20 is a diagram showing an overall configuration of an embodiment of a nuclear medicine device according to the embodiment of the present invention.

[0060] FIG. 21 is a flowchart showing a processing procedure of the same.

[0061] FIG. 22 is a diagram showing a demonstrative experiment setup for the nuclear medicine device according to the embodiment of the present invention.

[0062] FIG. 23A is a diagram showing a result of reconstruction using only PET events in measurement data obtained by the demonstrative experiment.

[0063] FIG. 23B is a diagram showing a result of reconstruction using only Compton events therein.

[0064] FIG. 23C is a diagram showing a result of application of simple simultaneous reconstruct on (conventional method) thereto.

[0065] FIG. 23D is a diagram showing a result of application or the hybrid reconstruction (method of the present invention) thereto.

DESCRIPTION OF EMBODIMENTS

[0066] An embodiment of the present invention will be described in detail below with reference to the drawings. Note that the present invention is riot limited to the description of the following embodiment or practical examples. The components of the embodiment and practical examples described below include ones easily conceivable by those skilled in the art, substantially identical ones, and ones within the so-called range of equivalency. The components of the embodiment and practical examples described below may be combined as appropriate, or selected and used as appropriate.

[0067] The embodiment of the present invention is a hybrid image reconstruction technique as an iterative approximation image reconstruction method, and includes production of updated images using a PET event and a Compton event separately in each iteration, and combination of the two updated images by weighted averaging to obtain an updated image. FIGS. 1 to 10 show the concept of this technique.

[0068] According to a first technique for hybrid image reconstruction according to the embodiment of the present invention, as shown in FIG. 1, a count value k of an iteration counter is reset to zero in step 1000. A current image 1020 obtained by image initialization in step 1010 is updated by a PET event to obtain a PET event updated image in step 1030, and updated by a Compton event to obtain a Compton event updated image in step 1040. An adder 1050 performs weighted addition by multiplication of predetermined factors β (0<β<1) and (1−β) to obtain an updated image 1060. In step 1070, the count value k of the counter is counted up by one. In step 1080, whether the count value k has reached a setting value K is determined. If the count value k is less than the setting value K, the processing returns to step 1020 to repeat the update 1030 by a PET event and the update 1040 by a Compton event independently of each other. If the count value k of The counter has reached the setting value K, the processing proceeds to step 1090 to output the image.

[0069] In a second technique using subsets for the hybrid image reconstruction according to the embodiment of the present invention, as shown in FIG. 2, a count value l of a subset counter and the count value k of the iteration counter are reset to zero in step 1100. Using PET event subsets 1132 and Compton event subsets 1142, a current image 1120 obtained by image initialization in step 1110 is updated into iterative approximation images by respective predetermined subsets 1 in steps 1130 and 1140. An adder 1150 then performs weighted addition to obtain an updated image 1160. In step 1170, the count value I of the subset counter is then counted up. The update of the iterative approximation image using the PET event subsets and the update of the iterative approximation image using the Compton event subsets are repeated until the count value l reaches a setting value L in step 1180. If the count value l of the subset counter is determined to have reached the setting value L in step 1180, the processing proceeds to step 1190 to reset the count value l of the subset counter to zero and count up the count value k of the iteration counter. Steps 1200 to 1190 are then repeated until the count value k of the iteration counter is determined to have reached the setting value K in step 1200. If the count value k of the iteration counter is determined to have reached the setting value K in step 1200, the image is outputted in step 1210.

[0070] FIG. 3 shows equations for an implementation example corresponding to FIG. 2. Using PET events and Compton events divided into respective L subsets, images updated using the respective events are produced, weighted using a predetermined factor (parameter) β, and then added for image update. Here, images which have the same dimensions and quantitative properties can be added since variations in the sensitivity of the pixels are corrected with respect to each event type before the addition into an overall sensitivity image.

[0071] Next, a third technique for making a sub iteration multiple times in the hybrid image reconstruction according to the embodiment of the present invention will be described with reference to FIG. 4.

[0072] In this technique, the count value k of the counter is reset in step 1300. An image is initialized in step 1310 to obtain a current image in step 1320. The processing then proceeds to step 1330, where a sub iteration is made to update an iterative approximation image by a subset of the PET event, using subsets 1332 of the PET event. The subsets are used for respective sub iterations in order, and return to the zeroth subset after the last one.

[0073] Update 1340 of an iterative approximation image by a subset of the Compton event is similarly performed to update the iterative approximation image by a subset, using subsets 1342 of the Compton event. The subsets are used for respective sub iterations in order, and return to the zeroth subset after the last one.

[0074] When both the sub iterations for the update 1330 of an iterative approximation image using the subsets of the PET event and the sub iterations for the update 1340 of an iterative approximation image using the subsets of the Compton event end, an adder 1350 performs weighted addition by multiplication of predetermined factors β (0<β<1) and (1−β) to obtain an updated image 1360.

[0075] The processing then proceeds to step 1370 to count up the count value k of the counter by one. In step 1380, whether the count value k is less than the setting value K is determined. If the determination in step 1380 is positive, the processing returns to step 1320 to repeat the main iteration.

[0076] If the count value k is determined to have reached the setting value K in step 1380, the processing proceeds to step 1390 to output the image.

[0077] Next, a fourth technique for setting the numbers of times of sub iterations with a PET event and a Compton event separately in the hybrid image reconstruction according to the embodiment of the present invention will be described with reference to FIG. 5.

[0078] In this technique, count values k, l.sub.p, and l.sub.c of counters are each initially reset to zero in step 1400.

[0079] The processing then proceeds to step 1410, where an image is initialized to obtain a current image 1420.

[0080] The processing then proceeds to step 1430. The update 1430 of an iterative approximate image by a PET event uses subsets 1432 of the PET event. The count value l.sub.p of the counter is counted Up in step 1440, and the sub iteration is repeated until the count value l.sub.p reaches a set number of times N.sub.p.

[0081] Meanwhile, the update 1450 of an iterative approximate image by a Compton event uses subsets 1452 of the Compton event. The count value l.sub.c of the counter is counted up in step 1460, and the sub iteration is repeated until the count value l.sub.c reaches a predetermined number of times L.sub.c.

[0082] When the update of the iterative approximation image based on the count value l.sub.p of the PET event subset counter and The update of the iterative approximation image based on the count value l.sub.c of the Compton event subset counter end, an adder 1470 performs weighted addition to obtain an updated image 1480. The processing then proceeds to step 1490 to count up the count value k of the counter by one. In step 1500, whether the count value k has reached the setting value K is determined. If the count value k has not reached the predetermined value, the processing returns to step 1420 to repeat the main iteration.

[0083] If, in step 1500, the count value k of the counter is determined to have reached the setting value K, the processing proceeds to step 1510 to output the image.

[0084] FIG. 6 shows equations for an implementation example corresponding to FIG. 5. In each main iteration, a sub iteration using the PET event is performed N.sub.P times and a sub iteration using the Compton event is performed N.sub.C times. The respective images after the end of the sub iterations are weighted using the predetermined factor (parameter) β and added for image update at the end of the main iteration. As with the equations in FIG. 3, images which have the same dimensions and quantitative properties can be added since variations in the sensitivity of the pixels are corrected with respect to each event type before the addition into an overall sensitivity image.

[0085] Next, a fifth technique for the hybrid image reconstruction according to the embodiment of the present invention will be described. with reference to FIG. 7.

[0086] In this technique, the count value k of the counter is initially reset to zero in step 1600. In step 1610, an image is initialized. In step 1620, a current image is obtained.

[0087] Next, the image is updated by a PET event in step 1630, updated by an annihilation radiation Compton event in step 1640, and updated by a single gamma-ray event in step 1650.

[0088] The processing then proceeds to an adder 1660, which weights and adds respective predetermined factors β.sub.1, β.sub.2, and β.sub.3 (β.sub.1+β.sub.2+β.sub.3=1) to obtain an updated image 1670.

[0089] Next, in step 1680, the count value k of the counter is counted up by one. In step 1690, whether the count value k of the counter is less than the setting value K is determined. If the determination is positive, the processing proceeds to step 1620 to repeat the image update in steps 1630, 1640, and 1650.

[0090] If the determination in step 1690 is negative, the processing proceeds to step 1700 to output the image.

[0091] Next, a sixth technique for the hybrid image reconstruction according to the embodiment of the present invention will be described with reference to FIG. 8. According to this technique, the count value k of the counter is reset to zero in step 1800. In step 1810, an image is initialized. In step 1820, a current image is obtained. The image is then updated by a PET event between absorber detectors of an absorber detector ring 80 in step 1830, updated by a PET event between a scatterer detector of a scatterer detector ring 70 and an absorber detector in step 1840, updated by a PET event between scatterer detectors of the scatterer detector ring 70 in step 1850, updated by an annihilation radiation Compton event in step 1860, and updated by a single gamma-ray Compton event in step 1870.

[0092] After the image update, an adder 1880 performs weighted addition using factors β.sub.1, β.sub.2, β.sub.3, β.sub.4, and β.sub.5 (β.sub.1+β.sub.2+β.sub.3+β.sub.4+β.sub.5=1) to obtain an updated image 1890. The processing then proceeds to step 1900 to count up the count value k of the counter by one. In step 1910, whether the count value k of the counter is less than, the setting value K is determined. If the determination is positive, the processing proceeds to step 1820 to repeat the image update.

[0093] On the other hand, if the determination in step 1910 is negative, the image is outputted in step 1920.

[0094] Next, a seventh technique for the hybrid image reconstruction according to the embodiment of the present invention will be described with reference to FIG. 9.

[0095] In this technique, the count value k of the counter is reset to zero in step 2000. In step 2010, an image is initialized. In step 2020, a current image is obtained.

[0096] Next, an image is updated by a PET event with time of flight (TOF) information in step 2030, updated by a normal PET event without TOF information in step 2040, and updated by a Compton event in step 2050.

[0097] After the end of the respective independent image updates, an adder 2060 weights and adds the respective outputs using factors β.sub.1, β.sub.2, and β.sub.3 (β.sub.1+β.sub.2+β.sub.3=1) to obtain an updated image of step 2070.

[0098] The processing then proceeds to step 2080 to count up the count value k of the counter by one. In step 2090, whether the count value k of the counter is less than the setting value K is determined. If the determination is positive, the processing returns to step 2020 to repeat the image update in steps 2030, 2040, and 2050.

[0099] On the other hand, if the determination in step 2090 is negative, the processing proceeds to step 2100 to output the image.

[0100] Next, an eighth technique for the hybrid image reconstruction according to the embodiment of the present invention will be described with reference to FIG. 10.

[0101] In this eighth technique, the count value k of the counter is reset to zero in step 2200. After image initialization in step 2210, a current image is obtained in step 2220.

[0102] Next, the image is updated by a PET event in step 2230, updated by a Compton event in step 2400, and updated by a 3-gamma event in step 2500. After the image update, an adder 2260 performs weighted addition using factors β.sub.1, β.sub.2, and (β.sub.3+β.sub.2β.sub.3=1) to obtain an updated image 2270.

[0103] Next, in step 2280, the count value k of the counter is counted up by one. The processing then proceeds to step 2290, and whether the count value k of the counter is less than the setting value K is determined. If the determination is positive, the processing returns to step 2220.

[0104] On the other hand, if the determination in step 2290 is negative, the processing proceeds to step 2300 to output the image.

[0105] While all of the foregoing techniques use the number of times of iterations as the condition to end the iteration. However, the condition to end iteration is not limited thereto.

[0106] The techniques according to the embodiment of the present invention can correct properties specific to respective events separately like sensitivity correction. This enables stable image update since the information to be combined has the same dimensions and quantitative properties. Moreover, as shown in FIG. 2, subsets can easily be used for speedup. Image reconstruction can be performed using two events of different properties effectively by simply combining the updated images using the subsets of the respective events. As shown in FIG. 11, the Compton event to be used here is directed to both where either one of a pair of annihilation radiations emitted from a nuclide (here, PET nuclide) 60 is detected (annihilation radiation Compton event) and where a single gamma-ray occurring from a nuclide that emits single gamma-rays beside positrons is detected (single gamma-ray Compton event). In using the latter nuclide, as shown in FIG. 7, the updated image can be a weighted average of three updated images produced using a PET event, an annihilation radiation Compton event, and a single gamma-ray Compton event.

[0107] Moreover, if scatterer detectors and absorber detectors differ greatly in spatial resolution, PET event updated images can be produced for respective detector combinations and averaged with weights.

[0108] Since convergence properties vary depending on the event type, the final convergence properties can be improved by devising the numbers of subsets and the timing to combine the updated images. For example, after several iterations for obtaining a Compton event updated image, the Compton event updated image may be combined with a PET event image obtained by one iteration.

[0109] Furthermore, the total value of the weighting factors is not limited to 1. For example, the factors can be changed to make the total value smaller than 1 depending on the iterations, so that both improvement of the convergence properties by the use of the subsets and improvement of the image quality due to reduced noise propagation are achieved.

[0110] FIG. 11 shows the classification of PET events and Compton events in measuring a nuclide 60 emitting single gamma rays in addition positrons using a double-ring device. Since PET events can be linearly identified compared to Compton events, an annihilation radiation Compton event falling within the same time window as another annihilation radiation Compton event or an absorption event as in FIG. 12 is desirably handled as a PET event. In view of this, PET events and Compton events can be classified using energy information and time information. FIG. 13 shows the flowchart thereof.

[0111] In the procedure shown in FIG. 13, a time window is initially set in step 3000. Next, in step 3010, whether there is a pair of unprocessed single events in the time window is determined.

[0112] If the determination is positive, the processing proceeds to step 3020 to select the pair of single events.

[0113] Next, in step 3030, the combination of the detectors is determined.

[0114] If the detectors are either scatterer detectors or absorber detectors, the processing proceeds to step 3040 to determine whether each event falls within the energy window of annihilation radiation (for example, 400 to 600 keV). If the determination is negative, the processing returns to step 3010.

[0115] On the other hand, if the detectors are a combination of a scatterer detector and an absorber detector, the processing proceeds to step 3050 to determine whether each event falls within the energy window of annihilation radiation (for example, 400 to 600 keV). If the determination in step 3050 is negative, the processing proceeds to step 3060 to determine whether the total energy falls within the energy window of annihilation radiation (for example, 400 to 600 keV).

[0116] If the determination in step 3060 is positive, the processing proceeds to step 3070 to determine whether the scatterer detector falls within the scattering angle-limited energy window of annihilation radiation (for example, 10 to 120 keV). If the determination is positive, the processing proceeds to step 3080 to determine whether there is an event (total of a scatterer detector and an absorber detector, or singly) in the energy window of annihilation radiation within the same time window.

[0117] On the other hand, if the determination in step 3060 or 3070 is negative, the processing proceeds to step 3090 to determine whether the total energy falls within the energy window of a single gamma ray (for example, 800 to 1000 keV) and the scatterer detector falls within the angle-limited energy window of a single gamma-ray (for example, 10 to 350 keV).

[0118] If the determination in step 3090 is positive, the processing proceeds to step 3120 to extract the events as a single gamma-ray Compton event. If the determination is negative, the processing returns to step 3010.

[0119] If the determination in the foregoing step 3040, 3050, or 3080 is positive, the processing proceeds to step 3100 to extract the events as a PET event.

[0120] If the determination in step 3080 is negative, the processing proceeds to step 3110 to extract the events as an annihilation radiation Compton event.

[0121] On the other hand, if the determination in step 3010 is negative, then in step 3200, the processing proceeds to the next time window.

[0122] High quality images can be stably reconstructed by applying the hybrid image reconstruction techniques to the respective extracted events.

[0123] To apply the techniques according to the embodiment of the present invention, PET events and Compton events need to be measurable from the same distribution. FIG. 14A shows an example of the most effective detector layout, where a scatterer detector ring 70 constituted by scatterer detector blocks 72 and an absorber detector ring 80 constituted by absorber detector blocks 82 are both disposed in an annular configuration, with the absorber detector ring 80 outside and the scatterer detector ring 70 inside. A patient 62 to be measured is located further inside. Here, as shown in FIG. 14B, scatterer detector block arrays 74 can be used to configure a multiple scatterer detector ring 70 for a multi-ring system. The latter multi-ring system not only facilitates scattering for improved sensitivity but enables layer-by-layer identification of scattering positions and is superior to simply increasing the ring thickness.

[0124] Instead of the scatterer detector ring 70 being located inside the absorber detector ring 80 with their positions in the body axis direction of the patient 62 matched as shown in FIG. 15A, the scatterer detector ring 70 and the absorber detector 80 may be disposed so that the scatterer detector ring 70 is divided into two rings 70A and 70B and located outside the absorber detector ring 80 in the body axis direction of the patient as shown in FIG. 15B. Such a configuration prevents the absorber detector ring 80 from being interfered with by the scatterer detector rings 70A and 70B, and the sensitivity of the absorber detector ring 80 for a PET event can be maximized.

[0125] Instead of the annular configuration, only the scatterer detectors may be configured as a partial ring 76 as shown in FIG. 16A or both the scatterer detectors and the absorption detects may be configured as partial rings 76 and 86 as shown in FIG. 16B, for example, and disposed only in a location of easy installation. As shown in FIG. 16C, scatterer detectors 78 and absorber detectors 88 of non-annular shapes may be disposed in an opposed configuration. The detectors disposed in an opposed configuration may be rotated relative to the patient 62 (the patient 62 may be rotated) during measurement. FIG. 16B shows a bed 64 and a pillow 66.

[0126] Examples of the detectors to be used include: a scintillation detector 100 constituted by a scintillator array 102 and a photodetector 104, which is commonly used as a PET detector, as shown in FIG. 17A; a three-dimensional DOI (Depth-of-Interaction) detector 110 including a three-dimensional scintillator array 112 as shown in FIG. 17B; and a semiconductor radiation detector 120 as shown in FIG. 17C.

[0127] The multi-ring system can be implemented by arranging semiconductor radiation detectors 120 at appropriate distances as shown in FIG. 18A, or similarly arranging scintillation detectors 100 at appropriate distances as shown in FIG. 18B. Liquid xenon or gas scintillation detectors can also be used. The detectors can be configured as scatterer detectors or absorber detectors. In any configuration, parameters such as the thicknesses and widths of the detectors, the distances between the detectors, and the distances from the measurement target are desirably adjusted to optimize the sensitivities of a Compton event and a PET event.

[0128] Measurable field of view of a Compton event is wider than that of a PET event. However, the spatial resolution deteriorates in proportion to the distance from the detectors. So, the pixel size can be increased to reduce the amount of calculation in performing image reconstruction outside the range where PET events can be measured as shown in FIG. 19.

[0129] Both PET events and Compton events need coincidence measurement. However, a hardware coincidence circuit is considered to be difficult to implement. The reason is that complicate processing is needed, like a Compton event can desirably be extracted as a PET event depending on the energy as illustrated in the flowchart of FIG. 13. For implementation, all measurements are therefore collected as single events, and the coincidence processing is performed by postprocessing or on-the-fly software coincidence calculation. A hybrid image reconstruction method for reconstructing an image by efficiently combining PET events and Compton events extracted by the processing of the flowchart of FIG. 13 and the like can be implemented using the equations shown in FIG. 3 or 6, for example. Here, the hybrid image reconstruction method is implemented on the basis of list-mode OSEM.

[0130] The hybrid image reconstruction technique is characterized in that updated images are produced using a PET event and a Compton event independently of each other in each sub iteration and added to produce a new updated image by weighted averaging. Here, various correction methods such as sensitivity correction, absorption correction, and scattering correction can be independently applied. Since the updated images have the same dimensions and quantitative properties, a stable solution can be easily obtained by combining the updated images. Aside from the list-mode OSEM, various iterative approximation methods such as MAP-EM (Maximum a Posterior-Expectation Maximization) using a priori information about images can be applied.

[0131] Detectors capable of obtaining TOF (Time Of Flight) information can also be used. As shown in FIG. 9, this enables an application such that PET events are classified into ones with TOF information and ones without TOF information, and updated images are produced using the respective events and averaged with weights along with an image updated by a Compton event. In particular, the separate application of the hybrid image reconstruction techniques depending on the presence or absence of TOF information is considered to be effective if high energy resolution is required of the scatterer detectors and is difficult to be achieved in a compatible mariner with time resolution.

[0132] If a nuclide that simultaneously emits single gamma-rays and positrons, such as .sup.44Sc, is used and a PET event and a single gamma-ray Compton event coincide in the same time window, the position of the radiation source can be limited to an intersection of the LOR and the Compton cone. As shown in FIG. 10, a 3-gamma event can thus be extracted and included into the weighted averaging as another updated image.

[0133] FIG. 20 shows an overall configuration of an embodiment of a nuclear medicine device using medical image processing according to the embodiment of the present invention. This embodiment includes: a PET-Compton event simultaneous measurement device 200 such as described above; a single event collection device 210 that collects single events on the basis of output of the PET-Compton event simultaneous measurement device 200; a single event storage 220; a software coincidence device 230 that detects coincidence by software on the basis of output of the single event storage 220; an extracted event storage 240; a hybrid image reconstruction device 250 that performs the hybrid image reconstruction according to the embodiment of the present invention on the basis of output of the storages 220 and 240; a control device 260 that controls the PET-Compton event simultaneous measurement device 200, the single event collection device 210, the software coincidence device 230, and the hybrid image reconstruction device 250; and a display and operation console 270 connected to the control device 260.

[0134] This nuclear medicine device performs on-the-fly software calculation as shown in FIG. 21. Initially, in step 4000, the PET-Compton event simultaneous measurement device 200 performs measurement. In step 4010, the detection signal is transmitted to the single event collection device 210 to collect single events.

[0135] In step 4020, the single event collection device 210 writes single events into the single event storage 220.

[0136] Meanwhile, the software coincidence device 230 starts software coincidence processing at the start of measurement. In step 4030, the software coincidence device 230 extracts PET events and Compton events based on the single events read from the single event storage 220. In step 4040, the software coincidence device 230 writes the extracted events into the extracted event storage 240.

[0137] The processing then proceeds to step 4050, where the hybrid image reconstruction device 250 performs the hybrid image reconstruction according to the embodiment of the present invention using the PET events and Compton events.

[0138] FIG. 22 shows a schematic diagram of a demonstrative experiment setup for hybrid reconstruction where a small animal is measured using a double-ring PET-Compton event simultaneous measurement device 200. A mouse 68 to which 9.8 MBq of .sup.89Zr-oxalate was administered was placed in a scattering detection ring 70, which had a width covering only the upper half of the body in the body axis direction, 22 hours after the administration with an absorption ring 80 disposed outside as a double-ring embodiment, and measured for five minutes. Since .sup.89Zr has a sufficient time difference between decay including positron decay and subsequent decay accompanied by emission of 909-key single gamma-rays, PET events and Compton events occurring from the same distribution can be independently measured.

[0139] To demonstrate the efficiency and safety of the hybrid reconstruction technique, PET events only between the scatterer detectors inside were extracted. For Compton events, only 909-keV single gamma-rays were extracted.

[0140] FIG. 23A shows images obtained in the case of using PET events. FIG. 23B shows images obtained in the case of using Compton events. FIG. 23C shows images obtained in the case of performing image reconstruction (simple simultaneous reconstruction) according to a conventional method by simply combining two events into one system. FIG. 23D shows images obtained in the case of performing reconstruction by the hybrid image reconstruction method according to the embodiment of the present invention. From the result of the reconstruction using PET events in FIG. 23A, up to which part of the upper body of the mouse is in the scatterer detector ring can be seen. In the case of using Compton events shown in in FIG. 23B, it can be seen that the images of the parts outside the scatterer detector ring can also be obtained. By the conventional simple simultaneous reconstruction shown in FIG. 23C, the image quality inside the ring was somewhat improved but strong artifacts appeared at the ends of the scatterer detector ring. By contrast, in the images obtained by the hybrid image reconstruction technique according to the embodiment of the present invention shown in FIG. 23D, artifacts at the ring ends were suppressed. It can also be seen that the image inside the scatterer detector ring was reconstructed with higher image quality than by the simple simultaneous reconstruction of FIG. 23G. It can be seen that the images outside the scatterer detector ring were also successfully obtained without much artifact. From the result of this demonstrative experiment, it is confirmed that PET events and Compton events can be efficiently and stably combined by the technique according to the embodiment of the present invention.

[0141] For the iterative approximation image reconstruction method, a list-mode method can be used, for example. However, the image reconstruction method is not limited thereto.

INDUSTRIAL APPLICABILITY

[0142] A new nuclear medicine device combining the principles of PET and a Compton camera can be implemented by using the hybrid image reconstruction technique of the present invention. This device can improve the sensitivity of inspections using a normal PET nuclide and stably improve image quality. In addition, when a nuclide that emits single gamma-rays from the same distribution, such as .sup.89Zr, is used, the device is expected to significantly improve image quality since components that only cause noise in an ordinary PET device can be effectively used for imaging. In particular, long half-life nuclides such as .sup.89Zr having a half-life of 3.3 days enable follow-up over a long period where commonly-used .sup.18F-FDG with a half-life of 110 minutes is unable to be measured due to attenuation. Demand for such a device is expected to grow in the future for various pharmacokinetic analyses and inspections. The present technique absolutely essential for the implementation of the device can be expected to become industrially important.

REFERENCE SIGNS LIST

[0143] 60 . . . nuclide

[0144] 62. . . patient

[0145] 64 . . . bed

[0146] 68 . . . mouse

[0147] 70, 70A, 70B . . . scatterer detector ring

[0148] 72 . . . scatterer detector block

[0149] 74 . . . scatterer detector block array

[0150] 76 . . . partial scatterer detector ring

[0151] 78 . . . scatterer detector

[0152] 80 . . . absorber detector ring

[0153] 82 . . . absorber detector block

[0154] 86 . . . partial absorber detector ring

[0155] 88 . . . absorber detector

[0156] 100 . . . scintillation detector

[0157] 102 . . . scintillator array

[0158] 104 . . . photodetector

[0159] 110 . . . DOI detector

[0160] 120 . . . semiconductor radiation detector

[0161] 200 . . . PET-Compton event simultaneous measurement device

[0162] 210 . . . single event collection device

[0163] 220 . . . single event storage

[0164] 230 . . . software coincidence device

[0165] 240 . . . extracted event storage

[0166] 250 . . . hybrid image reconstruction device

[0167] 260 . . . control device