Radiation measuring device, radiation measuring method, radiation measuring program, internal body tissue movement method, and non-transitory recording medium

Abstract

A radiation measuring device (1) is a forceps-type radiation measuring device (1) for measuring radiation from internal body tissue, and comprises: a pair of radiation detecting units (10) each provided with a plurality of radiation detecting elements; and an adjusting unit (20) for estimating a position of a radiation source from detected data acquired by the pair of radiation detecting units, and adjusting a detection sensitivity of the pair of radiation detecting units with respect to radiation on the basis of the estimated position of the radiation source.

Description

Radiation measurement device, radiation measurement method, radiation measurement program, internal tissue movement method, and non-transitory recording medium

[0001] The present invention relates to a radiation measurement device, a radiation measurement method, a radiation measurement program, a method for moving internal tissue, and a non-transitory recording medium.

[0002] As a conventional technique, a method for diagnosing cancer metastasis to lymph nodes and the like using PET (Positron Emission Tomography) diagnostic agents is known. 18 Examples of such diagnostic methods include F-FluoroDeoxyGlucose (F-FluoroDeoxyGlucose) and PSMA (Prostate Specific Membrane Antigen). Using this method, an insertable radiation detection unit has been developed to confirm the presence or absence of metastasis to lymph nodes. It is preferable that a device inserted into the body during surgery to diagnose metastasis to lymph nodes or the like be insertable into the body via a trocar or the like, in order to reduce the burden on the human body, i.e., to achieve minimal invasiveness.

[0003] For example, Non-Patent Document 1 discloses measuring the radioactivity concentration of lymph nodes using a forceps-type coincidence counter that can be inserted into the body. According to this document, the forceps-type coincidence counter, which is equipped with two radiation detection elements that can hold the lymph nodes, can eliminate background gamma rays and detect metastasis to individual lymph nodes. Furthermore, Non-Patent Document 2 discloses a forceps-type coincidence counter equipped with a pair of small scintillation detectors at the tip, intended for use under endoscopic observation.

[0004] Furthermore, Non-Patent Document 3 discloses a miniaturized radiation measurement device for insertion into the body, which uses a probe-type radiation detection unit in which the detector is divided into multiple parts, i.e., multiple detection elements are arranged, thereby making it possible to measure a simple distribution of a radiation source.

[0005] Takahashi, M., "Intraoperative Lymph Node Nuclear Medicine Diagnosis," Next Generation PET Research Report 2020, pp. 83-90 (2021). Ohashi, R., "Simulation of Sensitivity Distribution of Forceps-Type Mini-PET," Next Generation PET Research Report 2021, pp. 81-84 (2022). Abolaban, F., Alhawsawi, A. "Intraoperative Gamma Probe for Locating Cancerous Lymph Nodes," Arab J Sci Eng 46, 713-719 (2021).

[0006] However, the device disclosed in Non-Patent Document 3 is a device that measures single photons rather than coincidence counting and measures only radiation incident from one direction, and while it can obtain a simple distribution of the radiation source, it is not able to perform a quantitative estimation of radioactivity. Furthermore, the technologies disclosed in Non-Patent Documents 1 and 2 are coincidence counting devices in which two detection elements are arranged opposite each other, and therefore the radiation detection sensitivity varies depending on the position of the radiation source. Therefore, if the measurement values ​​are used as they are, there is a risk that the radioactivity cannot be estimated accurately.

[0007] In view of the above-described problems, one aspect of the present invention provides a technique for obtaining measurement values ​​for estimating radioactivity with high accuracy when a coincidence-type radiation detection unit is used.

[0008] In order to solve the above problems, a radiation measurement device according to one aspect of the present invention is a forceps-type radiation measurement device that measures radiation from internal body tissue, and includes a pair of radiation detection units, each having a plurality of radiation detection elements, and a correction unit that estimates a position of a radiation source from detection data acquired by the pair of radiation detection units, and corrects the detection sensitivity of the pair of radiation detection units for the radiation based on the estimated position of the radiation source.

[0009] In order to solve the above-mentioned problems, a radiation measurement method according to one aspect of the present invention is a radiation measurement method for measuring radiation from body tissue, and includes the steps of: estimating a position of a radiation source from detection data acquired by a pair of radiation detection units each having a plurality of radiation detection elements; and correcting the detection sensitivity of the pair of radiation detection units for the radiation based on the estimated position of the radiation source.

[0010] Furthermore, a radiation measurement program for causing a computer to implement the correction unit of the radiation measurement device, and a computer-readable recording medium on which the program is recorded, also fall within the scope of the present invention.

[0011] According to one aspect of the present invention, when a coincidence-type radiation detection unit is used, it is possible to obtain a measurement value for estimating radioactivity with high accuracy.

[0012] FIG. 1 is a block diagram showing the configuration of a radiation measurement device according to a first embodiment of the present invention. FIG. 2 is a flow diagram showing the flow of a radiation measurement method according to the first embodiment of the present invention. FIG. 3 is a schematic diagram showing the principle of measuring radiation from a PET diagnostic agent. FIG. 4 is a schematic diagram showing that detection sensitivity varies depending on the arrangement of a pair of radiation detection units and a radiation source. FIG. 5 is a schematic diagram showing another example in which detection sensitivity varies depending on the arrangement of a pair of radiation detection units and a radiation source. FIG. 6 is a schematic diagram showing differences in sensitivity parameters of detection elements due to differences in opening and closing methods of the radiation detection units. FIG. 7 is a schematic diagram showing a case in which each of a pair of radiation detection units is divided into four elements, and a diagram showing combinations of elements in which radiation is counted coincidentally. FIG. 8 is a schematic diagram showing another example of the number of divisions of the elements of the radiation detection unit. FIG. 9 is a graph showing changes in the radiation measurement rate depending on the position of the radiation source. FIG. 10 is a schematic diagram showing another example of the division positions of the elements of the radiation detection unit. FIG. 11 is a schematic diagram showing differences in the number of measurements for each combination of detection elements depending on the position of the radiation source. FIG. 12 is a conceptual diagram showing the configuration of a position estimation model. FIG. 13 is a schematic diagram showing the position of the radiation source output by the position estimation model. FIG. 14 is a schematic diagram showing the concept of correction performed by a correction unit. FIG. 15 is a flow diagram showing the flow of an internal tissue moving method. Fig. 1 is a schematic diagram showing the detection data of the radiation detection unit and the concept of its correction. Fig. 2 is a graph showing the learning results of the position estimation model. Fig. 3 is a graph showing the coincidence value estimation accuracy in each axis direction for all data of the position estimation model. Fig. 4 is a histogram of estimated radioactivity values ​​before and after sensitivity correction.

[0013] First Embodiment A radiation measurement device according to one embodiment of the present invention will be described in detail below, but before that, the necessity of the radiation measurement device according to this embodiment will be described.

[0014] When diagnosing cancer metastasis, false negatives (when a positive result is actually detected but the result is mistakenly diagnosed as negative) must be avoided as much as possible. However, because cancer metastasis is distributed irregularly in body tissues, the location (affected area) where the PET diagnostic agent accumulates also becomes irregular. Furthermore, when using coincidence counting technology, due to its principle, the detection sensitivity varies depending on the part of the radiation detection unit where detection occurs, i.e., depending on the position of the radiation source (hereinafter simply referred to as the "radiation source"). Therefore, interpreting measured values ​​assuming a constant detection sensitivity may increase the number of false negatives. The radiation measurement device according to this embodiment corrects the sensitivity of the radiation detection unit depending on the position of the radiation source.

[0015] (Radiation Measurement Apparatus 1) FIG. 1 is a block diagram showing the configuration of a radiation measurement apparatus 1 according to a first embodiment. The radiation measurement apparatus 1 is a forceps-type radiation measurement apparatus that measures radiation from internal body tissue. The forceps-type refers to a measurement apparatus having the shape of medical grasping forceps, with a radiation detection element disposed in a grasping portion at the tip of the forceps. More specifically, the radiation measurement apparatus 1 is a radiation measurement apparatus in which a radiation detection unit 10 is provided at each of two openable and closable ends of the medical grasping forceps. Such a radiation measurement apparatus 1 can be inserted into the abdominal cavity, thoracic cavity, or the like via a trocar or the like and used. In this case, the radiation measurement apparatus 1 is configured to have a small diameter so that it can be inserted into the trocar.

[0016] In this embodiment, the internal tissue is, for example, a lymph node. When FDG is used as a PET diagnostic agent, cancer cells that have metastasized to lymph nodes or the like generally tend to accumulate FDG. Therefore, by measuring the radiation emitted from FDG, it is possible to determine the extent to which cancer cells are contained in lymph nodes or the like. However, the internal tissue is not limited to lymph nodes as long as it is an organ in which a source of annihilation gamma rays such as FDG accumulates.

[0017] As shown in FIG. 1 , the radiation measurement device 1 includes a radiation detection unit 10 and a correction unit 20. The radiation detection unit 10 is configured as a pair (radiation detection units indicated by reference numerals 10A and 10B), and each of the pair of radiation detection units 10A and 10B includes a plurality of radiation detection elements. The radiation detection units are configured as a pair in order to measure annihilation gamma rays, as will be described later. For example, the radiation detection elements may be scintillation detection elements or semiconductor detection elements. Examples of scintillation detection elements that may be used include CsI(Tl) scintillators and NaI(Tl) scintillators. Examples of semiconductor detection elements that may be used include CdTe(CZT) semiconductor detection elements, Si semiconductor detection elements, and Ge semiconductor detection elements. Hereinafter, the radiation detection elements may also be simply referred to as "detection elements."

[0018] The correction unit 20 estimates the position of the radiation source from the detection data acquired by the pair of radiation detection units 10, and corrects the detection sensitivity of the pair of radiation detection units for radiation based on the estimated position of the radiation source. As shown in the figure, the correction unit 20 may include a position estimation model 21. The position estimation model 21 is a trained model that has been machine-learned to estimate the position of the radiation source of radiation detected by the radiation detection units 10, as will be described later.

[0019] 1 illustrates the radiation detection unit 10, the correction unit 20, and the position estimation model 21 as being included in a single housing, but the configuration of the radiation measurement device 1 is not limited to this. For example, the correction unit 20 and / or the position estimation model 21 may be located in a location separate from the radiation detection unit 10 so as to be able to communicate information. Then, detection data from the radiation detection unit 10 may be transmitted to the correction unit 20, and the correction unit 20 may further transmit the data to the position estimation model 21. Furthermore, the correction unit 20 and / or the position estimation model 21 may be located on the cloud. Then, the corrected detection data may be transmitted to an output unit provided in the radiation measurement device 1 or a display device or the like provided outside the radiation measurement device 1.

[0020] 3 is a schematic diagram showing the principle of measuring radiation from FDG. １８ F decays by emitting a positron,１８ O. The emitted positron quickly combines with an electron and undergoes pair annihilation. At this time, the energy generated is emitted as two annihilation gamma rays, which are measured. Because annihilation gamma rays are generated simultaneously in directions approximately 180° apart due to the law of conservation of momentum, they can be detected using a pair of radiation detection units 10A, 10B and a coincidence counting circuit. Specifically, radiation detected approximately simultaneously (within a predetermined time period) by the radiation detection units 10A, 10B is considered to be annihilation gamma rays and is counted. The number of annihilation gamma rays is proportional to the amount of radiation source (radioactivity), but the radiation detection units 10A, 10B cannot detect all annihilation gamma rays. This is the reason for the detection sensitivity. Hereinafter, "annihilation gamma rays" will also be simply referred to as "radiation."

[0021] Next, the correction of detection sensitivity performed by the correction unit 20 will be described. FIG. 4 is a schematic diagram showing that detection sensitivity differs depending on the arrangement of the pair of radiation detection units (e.g., scintillators) 10A, 10B and the radiation source. FIG. 4 shows an example of parallel opening and closing, in which the pair of radiation detection units 10A, 10B are opened and closed while remaining parallel. The radiation detection unit 10 can detect radiation using a coincidence circuit only when a radiation source is located between the radiation detection units 10A and 10B. However, due to the detection principle, the detection efficiency differs depending on the position of the radiation source. Specifically, as shown in FIG. 4( a), when the radiation source is located near the center of the radiation detection units 10A, 10B, the solid angle over which radiation can be counted simultaneously is large. In other words, the range over which radiation from the radiation source can be counted simultaneously is wide. Conversely, as shown in FIG. 4( b), when the radiation source is located away from the center of the radiation detection units 10A, 10B, the solid angle over which radiation can be counted simultaneously is small. In other words, the range over which radiation from the radiation source can be counted simultaneously is narrow. Therefore, as shown in FIG. 4C, the sensitivity distribution is such that the closer the radiation source position is to the center, the higher the sensitivity is, and the closer the radiation source position is to the edge, the lower the sensitivity is.

[0022] 5A and 5B are schematic diagrams showing another example in which detection sensitivity varies depending on the arrangement of the pair of radiation detection units 10A, 10B and the radiation source. As shown in Fig. 5A, when the opening angle is relatively small (D1), the solid angle in which coincidence counting of radiation from the radiation source can be performed is large. Conversely, as shown in Fig. 5B, when the opening angle is relatively large (D2), the solid angle in which coincidence counting of radiation from the radiation source can be performed is small. Therefore, as shown in Fig. 5C, the sensitivity distribution becomes smaller as the opening angle increases.

[0023] Figure 6 is a schematic diagram showing the difference in sensitivity parameters of the detection elements due to differences in the opening and closing methods of the radiation detection units 10A and 10B. Figure 6(a) shows a parallel opening and closing method in which the distance D changes while the radiation detection units 10A and 10B are parallel, as shown in Figure 5. In this case, the sensitivity parameters are the distance D and the radiation source position. Figure 6(b) shows a method in which the radiation detection units 10A and 10B open and close with their axes like scissors. In this case, the opening degree changes depending on the opening and closing angle θ, so the sensitivity parameters are the opening and closing angle θ and the radiation source position.

[0024] 7A and 7B are schematic diagrams of the radiation detection units 10A and 10B divided into four detection elements, A1, B1, C1, and D1, and A2, B2, C2, and D2, respectively. Fig. 7A is a diagram seen from an oblique side, with the handle side of the radiation measurement device 1 being the "rear" and the tip side of the radiation measurement device 1 being the "front."

[0025] 7(b) is a diagram showing combinations of detection elements that result in radiation coincidence in a configuration in which each radiation detection unit 10A, 10B is divided into two halves in the front-to-back direction and the side (left-to-right) direction, arranging a total of four elements. "Side" refers to a view from the side, perpendicular to the front-to-back direction. When the radiation detection units 10A, 10B are configured with a total of four elements, two vertically and two horizontally, there are a total of 16 combinations, including combinations of four elements, A1, B1, C1, and D1, and four elements, A2, B2, C2, and D2.

[0026] By configuring each of the radiation detection units 10A and 10B with a plurality of detection elements, the accuracy of estimating the position of the radiation source is improved. However, while a larger size (volume) of each detection element is preferable, a smaller overall size of the radiation detection units 10A and 10B is preferable in consideration of the possibility of inserting them into the abdominal cavity, for example, via a trocar. The number of segmented elements of the radiation detection units 10A and 10B is determined taking these constraints into account.

[0027] Figure 8 shows an example in which each of the radiation detection units 10A, 10B is divided into six elements. The front portion is further divided into smaller sections. The volumes of the front and rear detecting elements are not limited to being equal, as in the example shown in Figure 7(a), and the size of the detecting elements may be changed. For example, when forceps are actually used, the affected area is often clamped with the front portion of the forceps. By dividing the front portion into smaller sections, it is possible to improve the ability to discriminate the position of the radiation source in the front portion, where the affected area is most frequently clamped.

[0028] FIG. 9 is a graph showing the change in radiation measurement rate depending on the radiation source position for the radiation detection units 10A and 10B shown in FIG. 7( a). It is assumed, however, that the four detection elements are of equal size. The horizontal axis represents the radiation source position, with larger numbers indicating more forward positions. The vertical axis represents the ratio of measured coincidence counts. The "front-front" line in the graph represents the radiation count ratio measured between elements at the front of the radiation detection units 10A and 10B, such as A1-A2 and A1-B2. The "back-back" line represents the radiation count ratio measured between elements at the rear, such as C1-C2 and C1-D2. The "front-back" line represents the radiation count ratio measured between elements at the front and rear, such as A1-C2 and A1-D2. Except when the radiation source is located at both ends, the radiation count ratio measured between elements at the front and rear is higher on average. Furthermore, if there is a radiation source around the division position of the element, the count number will be widely distributed among the 16 combinations of detecting elements, and the estimation accuracy will be high.

[0029] Figure 10 is a schematic diagram showing examples of different division positions of the detector elements of the radiation detection units 10A, 10B as viewed from the side. Figure 10(a) shows an example in which the detector elements are divided equally in the front and rear, while Figure 10(b) shows an example in which the detector elements are divided closer to the front. In other words, the detector elements in the front of the radiation detection units 10A, 10B are small in volume, while the detector elements in the rear are large in volume. This division position increases the accuracy of estimating the radiation source position near the aforementioned division position, thereby improving the position discrimination ability of the front, which is where tissue is often grasped.

[0030] FIG. 11 is a schematic diagram showing that the number of measurements for each combination of detector elements differs depending on the position of the radiation source. The proportion Pk of radiation detected by the detector elements with hatching at the top and bottom in FIG. 11 is expressed by the following formula (1): Pk = Ωk / Ωall (1) Here, Ωk is the solid angle seen from the radiation source to the detector elements with hatching at the top and bottom in FIG. 11, and Ωall is the total solid angle seen from the radiation source to the detector elements at the top and bottom. Pk is set for each combination of detector elements. In other words, k = 1 to 16. Because the proportion Ωk changes depending on the position of the radiation source, the position can be estimated using Pk.

[0031] 4 to 6, the detection efficiency varies depending on the opening and closing methods of the radiation detection units 10A, 10B and the position of the radiation source. Therefore, the correction unit 20 estimates the position of the radiation source from the radiation detection data and further corrects the detection sensitivity. In this case, the correction unit 20 may estimate the position based on actual measurement values ​​obtained by changing the spatial positions of the radiation source and the radiation detection units 10A, 10B.

[0032] However, estimating the position of the radiation source based on actually measured data requires a large amount of actually measured data, which may be difficult in practice. Therefore, the correction unit 20 may estimate the position of the radiation source from the detection data using a position estimation model 21 that has been trained in advance by machine learning. The position estimation model 21 is, for example, a regression model using a neural network, and is a position estimation model that receives detection data for each combination of detection elements as input and outputs the position of the radiation source.

[0033] FIG. 12 is a conceptual diagram showing the configuration of the position estimation model 21. When the radiation detection units 10A and 10B shown in FIG. 10 are used, the position estimation model 21 includes, for example, an input layer I to which 16 pieces of detection data for each combination of detection elements are input as input X, multiple hidden layers W, and an output layer S. The position estimation model 21 outputs, for example, the probability (likelihood) of a radiation source's existence in 24 sections on the xy plane. The position estimation model 21 may also be a model that inputs parameters representing the spatial arrangement of a pair of radiation detection units 10 in addition to the detection data. The spatial arrangement of the radiation detection units 10 may be, as described with reference to FIG. 6 , an arrangement in which the radiation detection units 10A and 10B are opened and closed in parallel, or an arrangement in which the radiation detection units 10A and 10B are opened and closed axially. The parameter may be, for example, a distance D in the case of opening and closing in parallel, or an angle θ in the case of opening and closing in an axis. The distance D or angle θ of the radiation detection units 10A and 10B can be measured by installing a sensor or the like in the opening and closing mechanism of the radiation detection units 10A and 10B. The correction unit 20 inputs the measured parameters into the position estimation model 21, thereby making it possible to output more accurate estimation results.

[0034] FIG. 13 is a schematic diagram showing the radiation source position output by the position estimation model 21. FIG. 13( a) is a side view of the axis-opening / closing type radiation detection units 10A and 10B. As shown in FIG. 13( a), an xy plane is assumed to exist midway between the radiation detection units 10A and 10B, with the front-to-back direction being the x-axis and the left-to-right direction being the y-axis. FIG. 13( b) is a view of the xy plane as seen from the z-axis direction. As shown in FIG. 13( b), the xy plane may be divided into 4 × 6 = 24 sections, and the position estimation model 21 may estimate the radiation source presence probability in each section. In FIG. 13( b), the radiation source presence probability in each section estimated by the position estimation model 21 is represented by density. Furthermore, as shown in the figure, the position estimation model 21 may estimate the center position (x, y, z) of the radiation source in the internal tissue, or may estimate the distribution (area) of the radiation source positions in the internal tissue, based on the radiation source presence probability. Alternatively, the correction unit 20 may estimate the central position (x, y, z) of the radiation source in the internal tissue or the distribution (area) of the positions of the radiation source from the existence probability of the radiation source estimated by the position estimation model 21. The central position is the central position if the radiation source is a single sphere, and may also be referred to as the center of gravity of the distribution of the radiation source if the radiation source is not a sphere. A method for training the position estimation model 21 will be described later.

[0035] 14 is a schematic diagram showing the concept of the correction performed by the correction unit 20. The correction unit 20 acquires coincidence count values ​​Ci (i=1, 2, ... N), which are detection data of the radiation detection units 10A and 10B, and an angle value θ. In the above example, N=16. The coincidence count value Ci is acquired for each combination of detection elements. The angle value θ is an example of a parameter that represents the spatial arrangement of the pair of radiation detection units 10A and 10B.

[0036] The correction unit 20 estimates the position of the radiation source from the coincidence count value Ci and the angle value θ. Then, the correction unit 20 determines a detection sensitivity correction coefficient α using the estimated radiation source position, a sensitivity correction table which is the reciprocal of a value normalized so that the radiation source intensity is uniform from the detection sensitivity distribution according to the radiation source position obtained in advance, and the angle value θ (a parameter representing the spatial arrangement of a pair of radiation detection units). The method for determining the correction coefficient α will be described later.

[0037] Furthermore, the correction unit 20 derives a sensitivity-corrected coincidence count value by multiplying the sum (ΣCi) of the coincidence count values ​​Ci by a correction coefficient α. The radiation measurement device 1 may be equipped with an output unit that outputs the derived coincidence count value. By referring to the output coincidence count value, the operator of the radiation measurement device 1 can accurately estimate the amount of the radiation source in the body tissue of the measurement target, i.e., the amount of the accumulated PET diagnostic agent, without underestimating it. FDG, PSMA, and the like are known to accumulate in many cancer tissues, and the amount of this accumulation can be used to diagnose the presence or absence of cancer tissue.

[0038] As described above, the radiation measurement device 1 according to this embodiment can acquire measurement values ​​for estimating radioactivity with high accuracy when using a coincidence-type radiation detection unit. Therefore, measurement data measured during surgery using the radiation measurement device 1 can be used directly to diagnose the presence or absence of cancer tissue.

[0039] (Radiation Measurement Method) Next, a radiation measurement method S1 according to this embodiment will be described. FIG. 2 is a flow diagram showing the flow of the radiation measurement method S1. As shown in the figure, the radiation measurement method S1 is a radiation measurement method for measuring radiation from internal body tissue, and includes steps S11 and S12. In step S11, the correction unit 20 estimates the position of the radiation source from detection data acquired by a pair of radiation detection units 10A, 10B each having a plurality of radiation detection elements. Next, in step S12, the correction unit 20 corrects the detection sensitivity of the pair of radiation detection units for radiation based on the estimated position of the radiation source.

[0040] The radiation measurement method S1 may be executed by a program that causes a computer to execute, for example, the steps of estimating the position of a radiation source from detection data acquired by a pair of radiation detection units each including a plurality of radiation detection elements, and correcting the detection sensitivity of the pair of radiation detection units for radiation based on the estimated position of the radiation source.

[0041] According to the above-described radiation measurement method S1, similar to the effect of the radiation measurement device 1, when a coincidence type radiation detection unit is used, it is possible to acquire measurement values ​​for estimating radioactivity with high accuracy.

[0042] [Embodiment 2] Another embodiment of the present invention will be described below. For ease of explanation, the same reference numerals will be used to designate components having the same functions as those described in the above embodiment, and the description thereof will not be repeated.

[0043] (Internal tissue moving method) By referring to the position of the radiation source estimated by the correction unit 20 in embodiment 1, the internal tissue containing the radiation source can be moved. Specifically, in surgery using a medical robot, the position of the internal tissue (i.e., the radiation source) is estimated from the results of measuring radiation in the internal tissue, and if the internal tissue is not located near the center of the radiation detection unit 10, the radiation source is moved to near the center of the radiation detection unit 10. Internal tissue moving method S2 will be described below.

[0044] The internal tissue moving method S2 includes the steps of: estimating the position of the internal tissue from detection data of radiation from the internal tissue detected by a pair of radiation detection units each having a plurality of radiation detection elements; and moving the internal tissue to a position where the detection sensitivity of the radiation detection units is relatively high. The internal tissue moving method will be described in more detail below with reference to the drawings.

[0045] 15 is a flow chart showing a detailed flow of the internal tissue moving method S2. As shown in the figure, the internal tissue moving method S2 includes steps S21 to S24. In step S21, the correction unit 20 estimates the position of the internal tissue from detection data of radiation from the internal tissue detected by a pair of radiation detection units equipped with a plurality of radiation detection elements. The radiation detection unit 10 includes a pair of radiation detection units 10A and 10B, as described in embodiment 1.

[0046] In step S22, the correction unit 20 determines whether the position of the internal tissue is in a position where the detection sensitivity of the radiation detection unit 10 is relatively high. A position where the detection sensitivity is relatively high is near the center of the radiation detection unit. In step S22, if it is determined that the position of the internal tissue is in a position where the detection sensitivity of the radiation detection unit 10 is relatively high (step S: YES), the process proceeds to step S24, where the correction unit 20 corrects the measured radiation.

[0047] On the other hand, if it is determined in step S22 that the position of the internal tissue is not at a position where the detection sensitivity of the radiation detection unit 10 is relatively high (step S: NO), the process proceeds to step S23, where the internal tissue is moved to a position where the detection sensitivity of the radiation detection unit is relatively high. Specifically, the correction unit 20 controls the medical robot to move the internal tissue to a position where the detection sensitivity is relatively high. The process then proceeds to step S24, where the correction unit 20 measures the radiation again and corrects the measured radiation.

[0048] This method allows the body tissue to be moved using information on the estimated position of the body tissue (radiation source), and the body tissue can be moved to a position with higher detection sensitivity. Then, by performing radiation measurement again, highly accurate radiation measurement can be performed. Instead of moving the body tissue, the radiation detection unit may be moved so that the body tissue is positioned at a position with relatively high detection sensitivity.

[0049] [Embodiment 3] Another embodiment of the present invention will be described below. For the sake of convenience, the same reference numerals will be used to designate components having the same functions as those described in the above embodiment, and the description thereof will not be repeated.

[0050] (Learning Method of Position Estimation Model 21) In this embodiment, a learning method of the position estimation model 21 described in embodiment 1 will be described. In this embodiment, an example of a learning method of the position estimation model 21 that estimates the position of a radiation source from detection data measured by a pair of radiation detection units 10A and 10B, each divided into four detection elements, described in embodiment 1 will be described.

[0051] FIG. 16 is a schematic diagram illustrating the detection data of the radiation detection units 10A and 10B and the concept of correction. FIG. 16(a) shows the coordinates of the radiation source relative to the radiation detection units 10A and 10B. R is the distance from the detector center O to the radiation source (x, y, z). FIG. 16(b) is an example of a detection value measured at the radiation source position R. As shown in FIG. 16(b), the detection value generally tends to decrease as the distance R from the center O of the radiation detection units 10A and 10B to the radiation source increases. Therefore, it is necessary to correct the detection value so that it remains constant regardless of the radiation source position. If the desired measurement value is 200, a correction coefficient is set so that the correction value becomes 200. For example, multiplying the detection data for R=5 by correction coefficient C5, the detection data for R=6 by correction coefficient C6, and the detection data for R=7 by correction coefficient C7 results in 200. By setting this correction coefficient for all radiation source positions, the correction coefficient α is obtained. In this embodiment, for simplicity, the distance R, which is obtained by reducing the number of dimensions of the radiation source (three-dimensional directional coordinates) information, is used as the correction factor. However, it is also possible to investigate the characteristics of the detected values ​​in each of the x, y, and z directions and use three independent correction factors (or four if the distance D and the opening / closing angle θ are included).

[0052] The learning data for learning the position estimation model 21 is created by computer simulation. １８ Radiation (annihilation gamma rays) is generated randomly according to the half-life of F, and if the radiation passes through a detection element, it is evaluated that the radiation has been detected by the detection element. The simulation conditions are, for example, as follows: First, the shape of a lymph node as internal tissue is assumed to be spherical and its composition to be water. The diameter of the sphere is assumed to be 5 mm. The lymph node is randomly positioned between 7.5 mm and 12.5 mm from the innermost part of the radiation detection unit. The position of the 5 mm diameter lymph node determines the opening angle θ of the radiation detection units 10A and 10B that sandwich the lymph node.

[0053] The radioactivity of the source distributed inside the lymph nodes was 600 Bq. １８The spherical radiation source is a spherical radiation source containing F. The size of the sphere is set randomly within the range of 0.2 mm to 4.8 mm in diameter. The spherical radiation source is positioned randomly inside the lymph node, but is set so that the sphere does not protrude beyond the lymph node.

[0054] The measurement time by the radiation detection unit 10 is set to 30 seconds, the coincidence time window (when radiation is detected by opposing detection elements within this time, it is counted as one coincidence) is set to 12 ns, the coincidence energy window (when two radiations have radiation energies within this range, they are evaluated as an annihilation gamma ray) is set to 400 keV-600 keV, and the dead time (when a detection element has detected radiation once, it is not able to make a next detection during this dead time) is set to 256 ns.

[0055] Under the above simulation conditions, 15,000 measurement data sets are generated under various radiation source and position conditions. Of the generated data sets, 3,000 sets are used as training data for the position estimation model 21, and an optimal transformation function F(P1, P2, ..., P16) = (x, y, z) that fits 3,000 (P1, P2, ..., P16) → (x, y, z) is learned. Note that P1 to P16 are Pk described in FIG. 11 . Of the remaining data sets, 3,000 sets are used as validation data, and 9,000 sets are used as test data (evaluation data). First, the position estimation model 21 is trained using the training data to obtain parameters. The validation data is used to evaluate the position estimation model in the learning phase (the generalization performance of the trained model can be verified). During learning, if the loss value does not improve for 30 epochs using the validation data, learning is stopped. Next, the position estimation accuracy and coincidence value estimation accuracy of the position estimation model 21 are evaluated using the test data.

[0056] The position estimation model 21 is a machine learning model that takes the coincidence counts of 16 combinations of detector elements as input and outputs the coordinates (x, y, z) of the center position of the radiation source distribution. The network structure of the position estimation model 21 is a neural network structure including an input layer, three hidden layers (with 128, 128, and 64 neurons, respectively), and an output layer. The loss function is the mean square error, and optimization is performed using an RMSprop optimizer.

[0057] The position estimation accuracy of the position estimation model 21 is evaluated using the error between the true value (simulation condition) of the central position coordinate of the radiation source distribution and the estimated value of the position estimation model 21. Specifically, the true value is set to wtl = [xtl, ytl, ztl], the estimated value is set to wpl = [xpl, ypl, zpl], and the error εl = wpl - wtl = [xpl - xtl, ypl - ytl, zpl - ztl] is calculated. Furthermore, the evaluation of the sensitivity correction uses the average value and standard deviation of the radioactivity values ​​before and after the sensitivity correction as evaluation indices. In this case, as with the evaluation of the position estimation model 21, the average value and standard deviation of the radioactivity values ​​calculated for each specified range are used as evaluation indices.

[0058] 17 is a graph showing the learning results of the localization model 21 trained under the above conditions. The graph shows the relationship between the number of learning epochs on the horizontal axis and the loss on the vertical axis, and it can be seen that the loss decreases to a sufficiently small value around 20 epochs in both the training stage and the validation stage, indicating that learning is progressing.

[0059] Figure 18 is a graph showing the accuracy of coincidence value estimation in each axis direction for all data from the position estimation model 21. Figure 18(a) is a histogram showing the frequency of errors from the true value of estimated values ​​in the x-axis direction, Figure 18(b) is a histogram showing the frequency of errors from the true value of estimated values ​​in the y-axis direction, and Figure 18(c) is a histogram showing the frequency of errors from the true value in the z-axis direction. The curves in the graphs are approximate curves for the normal distribution of frequencies. The right side of the graphs shows the average value and standard deviation calculated by fitting the approximate curve of the normal distribution. For each axis, the average error was -0.03 to -0.06 mm, and the standard deviation was 0.24 to 0.39 mm, indicating that highly accurate estimates were obtained.

[0060] Figure 19 shows histograms of estimated radioactivity values ​​before and after sensitivity correction (the vertical axis represents frequency). The ideal value is 600 Bq. The estimated value before correction was 330±132 [Bq], while after correction it was 571±143 [Bq], showing improvements in both the variability and underestimation of radioactivity.

[0061] The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention.

[0062] [Summary] (Aspect 1) A forceps-type radiation measurement device that measures radiation from internal body tissue, the radiation measurement device comprising: a pair of radiation detection units each having a plurality of radiation detection elements; and a correction unit that estimates a position of a radiation source from detection data acquired by the pair of radiation detection units, and corrects the detection sensitivity of the pair of radiation detection units for the radiation based on the estimated position of the radiation source.

[0063] (Aspect 2) The radiation measurement device according to aspect 1, wherein the correction unit estimates the position of the radiation source from the detection data using a position estimation model that has been trained in advance by machine learning.

[0064] (Aspect 3) The radiation measurement device according to aspect 2, wherein the position estimation model is a model that receives as input, in addition to the detection data, parameters that represent a spatial arrangement of the pair of radiation detection units.

[0065] (Aspect 4) The radiation measurement device according to Aspect 3, wherein the correction unit determines a correction coefficient for the detection sensitivity using, in addition to the estimated position of the radiation source, a sensitivity correction table which is the reciprocal of a value normalized so that radiation source intensity is uniform from a detection sensitivity distribution according to the position of the radiation source obtained in advance, and parameters which represent a spatial arrangement of the pair of radiation detection units.

[0066] (Aspect 5) The radiation measurement device according to any one of Aspects 1 to 4, wherein the correction unit estimates a center position of the radiation source in the body tissue.

[0067] (Aspect 6) The radiation measurement device according to any one of Aspects 1 to 4, wherein the correction unit estimates a distribution of positions of the radiation source in the body tissue.

[0068] (Aspect 7) The radiation measurement device according to any one of Aspects 1 to 6, wherein the volume of the radiation detection element in the front portion of the pair of radiation detection units is smaller than the volume of the radiation detection element in the rear portion.

[0069] (Aspect 8) A radiation measurement method for measuring radiation from body tissue, the radiation measurement method including: a step of estimating a position of a radiation source from detection data acquired by a pair of radiation detection units each including a plurality of radiation detection elements; and a step of correcting detection sensitivity of the pair of radiation detection units for the radiation based on the estimated position of the radiation source.

[0070] (Aspect 9) A radiation measurement program causing a computer to execute the steps of: estimating a position of a radiation source from detection data acquired by a pair of radiation detection units each including a plurality of radiation detection elements; and correcting the detection sensitivity of the pair of radiation detection units for radiation based on the estimated position of the radiation source.

[0071] (Aspect 10) A method for moving internal tissue, comprising: a step of estimating a position of the internal tissue from detection data of radiation from the internal tissue detected by a pair of radiation detection units each having a plurality of radiation detection elements; and a step of moving the internal tissue to a position where the detection sensitivity of the radiation detection units is relatively high.

[0072] (Aspect 11) A computer-readable non-transitory recording medium having the radiation measurement program according to aspect 9 recorded thereon.

[0073] REFERENCE SIGNS LIST 1... Radiation measurement device 10... Radiation detection unit 20... Correction unit 21... Position estimation model

Claims

1. A forceps-type radiation measurement device for measuring radiation from internal body tissue, comprising: A pair of radiation detection units each including a plurality of radiation detection elements; a correction unit that estimates a position of a radiation source or a distribution of the positions, or both, from detection data acquired by the pair of radiation detection units, and corrects the detection sensitivity of the pair of radiation detection units for radiation from the radiation source by using a sensitivity correction table that is the reciprocal of a value normalized so that a radiation source intensity becomes uniform from a detection sensitivity distribution according to the position of the radiation source that has been acquired in advance in addition to the estimated position of the radiation source or the distribution of the positions, or both; A radiation measurement device comprising:

2. 2. The radiation measurement device according to claim 1, wherein the correction unit estimates the positions of the radiation sources, or a distribution of the positions, or both, by using a position estimation model that has been trained in advance by machine learning, and by using parameters that represent a spatial arrangement of the pair of radiation detection units in addition to the detection data.

3. The radiation measurement device according to claim 1 , wherein the correction unit determines a correction coefficient for the detection sensitivity by using the sensitivity correction table and a parameter that represents a spatial arrangement of the pair of radiation detection units.

4. A radiation measurement method for measuring radiation from body tissue, comprising: estimating positions of radiation sources or a distribution of the positions, or both, from detection data acquired by a pair of radiation detection units each including a plurality of radiation detection elements; correcting the detection sensitivity of the pair of radiation detection units for radiation from the radiation source by using a sensitivity correction table that is the reciprocal of a standardized value obtained by normalizing a radiation source intensity to a uniform value based on a detection sensitivity distribution according to the position of the radiation source that has been previously acquired in addition to the estimated position of the radiation source or a distribution of the positions, or both; A radiation measurement method comprising:

5. estimating a position of the internal tissue from detection data of radiation from the internal tissue detected by a pair of radiation detection units each including a plurality of radiation detection elements; moving the body tissue to a position where the detection sensitivity of the radiation detection unit is relatively high; A method for moving tissue in an organism, comprising: