Apparatus and method enabling quantitative thyroid SPECT without CT using a neural network

Abstract

An apparatus and method are disclosed that enable quantitative thyroid SPECT without CT using a neural network according to an embodiment. An apparatus that enables quantitative thyroid SPECT without CT using a neural network according to an embodiment includes a generation unit that inputs a first SPECT (Single Photon Emission Computed Tomography) image and a second SPECT image of the thyroid into a first model to generate an attenuation map; a segmentation unit that segments a thyroid map by inputting the attenuation map and the first SPECT image into a second model; and a calculation unit that calculates an uptake amount of a radiopharmaceutical of the thyroid using a first SPECT image sinogram, a second SPECT image sinogram, the attenuation map, and the thyroid map.

Description

Technical Field

[0001] The present disclosure relates to a technique that enables quantitative thyroid SPECT without CT using a neural network.

Background Art

[0002] Thyroid function is evaluated by measuring the amount (%) of a radiopharmaceutical taken up by the thyroid gland after intravenous injection into a thyroid patient. Typically, the measurement of the uptake amount of the radiopharmaceutical is mainly evaluated by a thyroid uptake system, but there is a limitation in that the accuracy is low.

[0003] Therefore, SPECT / CT (Single-Photon Emission Computed Tomography / Computed Tomography), which has been reported to have higher accuracy in measuring the uptake amount than the thyroid uptake system, has attracted attention.

[0004] SPECT / CT is a nuclear medicine imaging method that provides quantitative information using a radiopharmaceutical, for example, Tc-99m pertechnetate. SPECT / CT uses CT for quantitative measurement. Here, CT performs attenuation correction to accurately measure the uptake amount of the thyroid gland and is used for thyroid segmentation.

[0005] However, the method for evaluating thyroid function using SPECT / CT induces a total radiation exposure of about 3.34 mSv, which is the sum of the radiation exposure due to CT of about 1.12 mSv and the radiation exposure due to SPECT of about 2.22 mSv. Furthermore, there is a problem that thyroid segmentation is troublesome because it needs to be segmented one by one by a specialist.

Summary of the Invention

Problems to be Solved by the Invention

[0006] The disclosed embodiments aim to enable quantitative thyroid SPECT without CT using a neural network.

Means for Solving the Problems

[0007] An apparatus for enabling quantitative thyroid SPECT without CT using a neural network according to an embodiment includes: a generation unit that inputs a first SPECT (Single Photon Emission Computed Tomography) image and a second SPECT image of the thyroid into a first model to generate an attenuation map; a segmentation unit that segments a thyroid map by inputting the attenuation map and the first SPECT image into a second model; and a calculation unit that calculates the uptake amount of the radiopharmaceutical of the thyroid using a first SPECT image sinogram, a second SPECT image sinogram, the attenuation map, and the thyroid map.

[0008] The first SPECT image may be a primary emission SPECT image.

[0009] The second SPECT image may be a scattering SPECT image.

[0010] The generation unit may generate a third SPECT image by performing attenuation correction (AC) on the first SPECT image sinogram using the attenuation map, performing scatter correction (SC) on the first SPECT image sinogram using the second SPECT image sinogram, and applying resolution recovery (RR) to the first SPECT image sinogram.

[0011] The third SPECT image may be a quantitative ACSCRR SPECT image.

[0012] The calculation unit may calculate the uptake amount of the radiopharmaceutical in the thyroid by combining the third SPECT image and the thyroid map.

[0013] The calculation unit may calculate the uptake amount of the radiopharmaceutical in the thyroid by combining the third SPECT image and the thyroid map and counting matching voxels.

[0014] The second model may be trained to segment a thyroid map in an input image based on a plurality of thyroid segmentation maps drawn and labeled along the outline of the thyroid in a CT (Computed Tomography) image.

[0015] A method for enabling quantitative thyroid SPECT without CT using a neural network according to an embodiment is a method performed by an apparatus for enabling quantitative thyroid SPECT without CT using a neural network including one or more processors and a memory storing one or more programs executed by the one or more processors, the method including: inputting a first SPECT (Single Photon Emission Computed Tomography) image and a second SPECT image of the thyroid into a first model to generate an attenuation map; inputting the attenuation map and the first SPECT image into a second model to segment a thyroid map; and calculating an uptake amount of a radiopharmaceutical of the thyroid using a first SPECT image sinogram, a second SPECT image sinogram, the attenuation map, and the thyroid map.

[0016] The first SPECT image may be a primary emission SPECT image.

[0017] The second SPECT image may be a scattering SPECT image.

[0018] The generating step may include the steps of performing attenuation correction (AC) on the first SPECT image sinogram using the attenuation map, performing scatter correction (SC) on the first SPECT image sinogram using the second SPECT image sinogram, and applying resolution recovery (RR) to the first SPECT image sinogram to generate a third SPECT image. The third SPECT image may be a quantitative ACSCRR SPECT image.

[0019] The calculating step may include the step of combining the third SPECT image and the thyroid map to calculate the uptake amount of the radiopharmaceutical in the thyroid.

[0020] The calculating step may include the step of combining the third SPECT image and the thyroid map to count the matching voxels to calculate the uptake amount of the radiopharmaceutical in the thyroid.

[0021] The second model may be learned to segment the thyroid map in the input image based on a plurality of thyroid segmentation maps drawn and labeled along the outline of the thyroid in a CT (Computed Tomography) image.

Advantages of the Invention

[0022] The disclosed embodiments can reduce radiation exposure by CT based on the ALARA (As Low As Reasonably Achievable) principle by performing attenuation correction using a neural network instead of CT.

[0023] The disclosed embodiments can segment a thyroid map using a neural network and provide the thyroid map without human intervention.

Brief Description of the Drawings

[0024]

Figure 1

Figure 2

Figures 3a - 4c

Figure 5

Modes for Carrying Out the Invention

[0025] The terms used in this specification are selected considering their functions and, as much as possible, general terms that are currently widely used. However, this may change depending on the intentions or customs of those skilled in the art, or the emergence of new technologies. Also, in certain cases, there are terms arbitrarily selected by the applicant, and in such cases, the meaning thereof is described in the corresponding description part of the specification. Therefore, it should be clarified that the terms used in this specification should be interpreted based not only on the name of the terms but also on the substantial meaning they possess and the overall content of this specification.

[0026] Terms such as first and second used in this application may be attached to describe various components, but the components are not limited by the terms. The terms are only for the purpose of distinguishing one component from another. For example, without departing from the scope of rights according to the concept of the present invention, the first component may be referred to as the second component, and similarly, the second component may be referred to as the first component.

[0027] The singular expressions used in this application include plural expressions unless the context clearly indicates a different meaning. In this application, terms such as "including", "comprising", or "having" are used to express the existence of the components described in the specification or combinations thereof, and do not preclude the possibility of the existence or addition of other components or features in advance.

[0028] Also, the embodiments described in this specification may have aspects that are entirely hardware, partially hardware and partially software, or entirely software. In this specification, terms such as "unit", "module", "device", "server", or "system" refer to computer-related entities such as hardware, a combination of hardware and software, or software. For example, a unit, module, device, server, or system may refer to hardware that constitutes part or all of a platform and / or software such as an application for driving the hardware.

[0029] Hereinafter, actual examples will be described in detail with reference to the accompanying drawings and the content shown in the accompanying drawings, but the scope to be claimed is not limited or restricted by the examples.

[0030] FIG. 1 is a block diagram for explaining an apparatus 100 that enables quantitative thyroid SPECT without CT using a neural network according to an embodiment.

[0031] Referring to FIG. 1, an apparatus 100 that enables quantitative thyroid SPECT without CT using a neural network includes a generation unit 110, a segmentation unit 120, and a calculation unit 130.

[0032] The generation unit 110, the segmentation unit 120, and the calculation unit 130 are implemented using one or more physically separated apparatuses, but may also be implemented by one or more processors or a combination of one or more processors and software, and may not be clearly separated in specific operations, different from the illustrated example.

[0033] The generation unit 110 inputs a first SPECT (Single Photon Emission Computed Tomography) image and a second SPECT image of the thyroid into a first model to generate an attenuation map.

[0034] Single Photon Emission Computed Tomography is one of the nuclear medicine imaging techniques. After injecting a radiopharmaceutical that emits single photons (gamma rays) into the living body, it measures the gamma rays transmitted in the living body and reconstructs the in-vivo distribution of the single photon-emitting radionuclide as an image. In particular, Single Photon Emission Computed Tomography can provide a functional image of the living body, which is a biochemical phenomenon of the living body.

[0035] The first SPECT image may be a primary emission SPECT (Primary Emission SPECT) image generated using Single Photon Emission Computed Tomography.

[0036] The primary emission SPECT image may be a SPECT image of the most frequent primary gamma rays in Single Photon Emission Computed Tomography.

[0037] The second SPECT image may be a scattered SPECT (Scattering SPECT) image generated using single photon emission computed tomography. The scattered SPECT image may be a SPECT image of gamma rays scattered during single photon emission computed tomography.

[0038] At least one of the first SPECT image and the second SPECT image may be a SPECT image of NCRR corrected for collimator-detector response, in other words, resolution recovery, for accuracy improvement. Specifically, at least one of the first SPECT image and the second SPECT image may be an image corrected for resolution recovery using a Butterworth low-pass filter for reduction of statistical noise.

[0039] At least one of the first SPECT image and the second SPECT image may be an image normalized at the maximum value of the composite image of two SPECTs.

[0040] The generation unit 110 may generate a third SPECT image by attenuation-correcting (Attenuation Correction; AC) the first SPECT image sinogram using an attenuation map.

[0041] The generation unit 110 may generate a third SPECT image by performing at least one of attenuation correction, scatter correction (SC), and resolution recovery (Resolution Recovery; RR) using an attenuation map.

[0042] At this time, the third SPECT image may be a quantitative ACSCRR SPECT image. The quantitative ACSCRR SPECT image may be a SPECT image to which correction has been applied to the SPECT image sinogram of primary gamma rays.

[0043] The segmentation unit 120 inputs the attenuation map and the first SPECT image into the second model to segment the thyroid map.

[0044] The calculation unit 130 calculates the uptake amount of the radiopharmaceutical in the thyroid gland using the first SPECT image sinogram, the second SPECT image sinogram, the attenuation map, and the thyroid map.

[0045] The calculation unit 130 may calculate the uptake amount of the radiopharmaceutical in the thyroid gland by combining the third SPECT image and the thyroid map.

[0046] The calculation unit 130 may calculate the uptake amount of the radiopharmaceutical in the thyroid gland by combining the third SPECT image and the thyroid map and counting the matching voxels.

[0047] The calculation unit 130 may calculate the uptake amount of the radiopharmaceutical in the thyroid gland by counting the matching voxels between the third SPECT image and the thyroid map, considering negative voxel values as 0.

[0048] FIG. 2 is a block diagram for explaining a learning model of an apparatus 100 that enables quantitative thyroid SPECT without CT using a neural network according to an embodiment.

[0049] Referring to FIG. 2, the first SPECT image 211 and the second SPECT image 212 are input into the first model 220. The second model 230 receives the first SPECT image 211 and the attenuation map output by the first model 220.

[0050] Specifically, when the first SPECT image 211 and the second SPECT image 212 are input, the first model 220 can output, as an output value, the attenuation map provided by CT even without CT using one or more neural networks.

[0051] The first model 220 may include a loss function based on the following mathematical formula 1. [Mathematical formula 1] L(G(X), Y)=L error (G(X), Y)+LGDL (G(X), Y)

[0052] Here, the loss function is L(G(X), Y), Y is the ground truth of the attenuation map, X is the input SPECT, G(X) is the generated attenuation map, and L error may be the previously defined first loss function or second loss function. L GDL may be GDL (Gradient Difference Loss) for the sharpness of the generated attenuation map.

[0053] For example, the first loss function may be a function that calculates the sum of the differences between the absolute values between the ground truth of the attenuation map and the generated attenuation map. In other words, the first loss function (L1) may be set based on the following mathematical formula 2. [Mathematical formula 2] JPEG2025525002000002.jpg6170

[0054] For example, the second loss function may be a function that calculates the sum of the squares of the differences between the ground truth of the attenuation map and the generated attenuation map. In other words, the second loss function (L2) may be set based on the following mathematical formula 3. [Mathematical formula 3] JPEG2025525002000003.jpg6170

[0055] For example, LGDL may be used to compensate for imaging blur due to the loss effect of the second loss function, and LGDL may be set based on the following mathematical formula 4. [Mathematical formula 4] JPEG2025525002000004.jpg7170

[0056] At this time, the first model 220 may be learned based on a hyperfunction attenuation map that can be obtained by photographing the hyperfunctioning thyroid by CT, a hypofunction attenuation map that can be obtained by photographing the hypofunctioning thyroid by CT, and a function standard attenuation map that can be obtained by photographing the thyroid with standard function by CT.

[0057] CT may be entirely photographed so as to cover all of the axis field of view of SPECT, or may be photographed so as to cover only 1 / 2 to 2 / 3 of the axis field of view of SPECT in order to reduce unnecessary radiation exposure by CT. It is preferable that SPECT and CT are photographed in the same axis field of view from the mid-skull to the upper-mediastinum.

[0058] In the CT image or the SPECT image, as a clinical diagnosis by a nuclear medicine specialist, for example, a diagnosis result for at least one of Graves' disease / hyperthyroidism, painless / subacute thyroiditis, SNG (Single Nodular Goiter) / MNG (Multi Nodular Goiter), drug-induced thyroiditis, and lingual thyroid may be labeled.

[0059] The second model 230 may be designed to segment a thyroid map as an output value when the first SPECT image 211 and the attenuation map output by the first model 220 are input using one or more neural networks.

[0060] The second model 230 may include a loss function based on categorical cross entropy (CCE).

[0061] Specifically, the second model 230 may include a loss function based on Mathematical Formula 5. [Mathematical Formula 5] JPEG2025525002000005.jpg10170

[0062] At this time, y i is the ground truth, y h is synthetic, and n may be the number of classes of the second model 230.

[0063] For example, y i is the ground truth represented by 0 or 1, y h is a probability value between 0 and 1, and n may be set to three, such as background, left thyroid, and right thyroid.

[0064] At this time, the second model 230 may be learned to segment the thyroid map in the input image based on a plurality of thyroid maps drawn and labeled along the outline of the thyroid in the CT image.

[0065] At least one of the first model 220 and the second model 230 may be a U-net structure, preferably a 3D U-net-based structure. In other words, the first model 220 and the second model 230 may be designed with a structure that skip-connects a contraction path and a segmentation path.

[0066] As a specific example, at least one of the first model 220 and the second model 230 may be designed to include 64 initial neurons and 4 skip connections. At least one of the first model 220 and the second model 230 may be a neural network based on a U-net structure that skip-connects a contraction path including one or more folding blocks including a 3×3 convolution layer, batch normalization, and a ReLU function and a segmentation path including one or more of the folding blocks.

[0067] At this time, for the folding block in the contraction path, max pooling, preferably 2×2×2 stride max pooling, may be applied, and for the folding block in the segmentation path, transposed convolution, preferably 2×2×2 upconvolution, may be applied.

[0068] At least one of the first model 220 and the second model 230 may further include a 1×1×1 convolutional layer as the last output layer.

[0069] The first model 220 and the second model 230 may be designed to be end-to-end learned.

[0070] On the other hand, although the first model 220 and the second model 230 have been described as being of the U-net structure widely used in the medical field, this is exemplary, and they may be designed based on various already-known structures such as Seg-NET, MFU-net, Generative Adversarial Network (GAN), and are not necessarily limited to the U-net structure.

[0071] Figs. 3a to 4c are exemplary diagrams showing the performance of the learning model of the apparatus 100 that enables quantitative thyroid SPECT without CT using a neural network according to an embodiment.

[0072] Fig. 3a is an exemplary diagram shown for comparing the ground truth and synthetic for the attenuation map.

[0073] The ground truth on the left is the attenuation map generated by CT imaging, and the synthetic on the right is the attenuation map generated by the apparatus 100 that enables quantitative thyroid SPECT without CT according to an embodiment.

[0074] Figure 3b shows the correlation of the attenuation coefficient between the ground truth of the attenuation map in Figure 3a and the synthetic attenuation map. Figure 3b shows a high correlation between the ground truth of the attenuation map and the synthetic attenuation map.

[0075] Figure 4a shows the uptake of radiopharmaceuticals in the thyroid measured by the existing SPECT / CT and the device 100 that enables quantitative thyroid SPECT without CT according to one embodiment.

[0076] The horizontal axis in Figure 4a represents the uptake of radiopharmaceuticals in the thyroid measured by SPECT / CT, and the vertical axis represents the uptake of radiopharmaceuticals of the device 100 that enables quantitative thyroid SPECT without CT according to one embodiment.

[0077] The correlation between the uptake of radiopharmaceuticals in the thyroid measured by the existing SPECT / CT and the uptake of radiopharmaceuticals in the thyroid output by the device 100 that enables quantitative thyroid SPECT without CT according to one embodiment shows r = 0.9980, R^2 = 0.9959, p < 0.0001 in the range of uptake 0 - 30. At this time, r represents the correlation coefficient in the Pearson correlation, and p represents significance.

[0078] Figure 4b is a Bland - altman plot chart showing the difference between the uptake of radiopharmaceuticals in the thyroid measured by SPECT / CT and the uptake of radiopharmaceuticals in the thyroid output by the device 100 that enables quantitative thyroid SPECT without CT according to one embodiment.

[0079] The horizontal axis in Figure 4b represents the average value between the uptake of radiopharmaceuticals in the thyroid measured by SPECT / CT and the uptake of radiopharmaceuticals in the thyroid output by the device 100 that enables quantitative thyroid SPECT without CT according to one embodiment. The vertical axis represents the difference between the uptake of radiopharmaceuticals in the thyroid measured by SPECT / CT and the uptake output by the device 100 that enables quantitative thyroid SPECT without CT according to one embodiment.

[0080] The Brandt - Altman plot chart of FIG. 4b shows a bias of -0.99% and a non - significant systematic deviation, indicating the high accuracy of the apparatus 100 that enables quantitative thyroid SPECT without CT according to one embodiment.

[0081] FIG. 4c is a graph showing the uptake of thyroid radiopharmaceuticals of SPECT / CT by thyroid disease type and the apparatus 100 that enables quantitative thyroid SPECT without CT according to one embodiment.

[0082] At this time, the uptake of thyroid radiopharmaceuticals of SPECT / CT by thyroid disease type and the apparatus 100 that enables quantitative thyroid SPECT without CT according to one embodiment is shown as an average ± standard deviation value.

[0083] Thyroid diseases can be distinguished by the absorption rate of the thyroid. Referring to FIG. 4b, the uptake of thyroid radiopharmaceuticals of the apparatus 100 that enables quantitative thyroid SPECT without CT and SPECT / CT shows, for example, the uptake of radiopharmaceuticals for Graves‘ disease / hyperthyroidism, Painless / subacute thyroiditis, SNG (Single Nodular Goiter) / MNG (Multi Nodular Goiter), and others (drug - induced thyroiditis, lingual thyroid).

[0084] FIG. 5 is a flowchart for explaining a method that enables quantitative thyroid SPECT without CT using a neural network according to one embodiment.

[0085] Referring to FIG. 5, the method that enables quantitative thyroid SPECT without CT using a neural network according to one embodiment may be performed by the apparatus for quantitative thyroid SPECT without CT using the neural network of FIG. 1.

[0086] First, a quantitative thyroid SPECT apparatus without CT using a neural network inputs a first SPECT (Single Photon Emission Computed Tomography) image and a second SPECT image of the thyroid into a first model to generate (510) an attenuation map.

[0087] Next, the quantitative thyroid SPECT apparatus without CT using a neural network inputs the attenuation map and the first SPECT image into a second model to segment (520) a thyroid map.

[0088] Next, the quantitative thyroid SPECT apparatus without CT using a neural network calculates (530) the uptake amount of a thyroid radiopharmaceutical using the first SPECT image sinogram, the second SPECT sinogram, the attenuation map, and the thyroid map.

[0089] The method of FIG. 5 shown is described with reference to the flowchart presented in the same drawing. For the sake of explanation, the method is illustrated and described in a series of blocks, but the present invention is not limited to the order of the blocks, and some blocks may be executed in a different order than or simultaneously with the other blocks as illustrated and described herein, and various other branches, flow paths, and orders of blocks that achieve the same or similar results may be implemented. Also, not all of the blocks illustrated for the implementation of the method described herein may be required.

[0090] Furthermore, the method according to an embodiment of the present invention may be embodied in the form of a computer program for executing a series of processes, and the computer program may be recorded on a computer-readable recording medium. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs, DVDs, magneto-optical media such as floptical disks, and hardware devices specially configured to store and execute program instruction words such as ROMs, RAMs, and flash memories.

[0091] As described above with reference to the embodiments, it should be understood that those skilled in the art can make various modifications and changes to the present invention without departing from the spirit and scope of the present invention described in the following claims.

Industrial Applicability

[0092] An apparatus and method that enable quantitative thyroid SPECT without CT using a neural network according to an embodiment can be used in the medical diagnosis industry in the field of nuclear medicine by performing attenuation correction using a neural network instead of CT.

Claims

1. A generation unit that inputs a first SPECT (Single Photon Emission Computed Tomography) image and a second SPECT image of the thyroid gland into a first model to generate an attenuation map; A segmentation unit that inputs the attenuation map and the first SPECT image into a second model to segment a thyroid map; and An apparatus enabling CT-free quantitative thyroid SPECT using a neural network, including a calculation unit that calculates the uptake amount of a radiopharmaceutical of the thyroid gland using a first SPECT image sinogram, a second SPECT image sinogram, the attenuation map, and the thyroid map.

2. The apparatus enabling CT-free quantitative thyroid SPECT using a neural network according to Claim 1, wherein the first SPECT image is a primary emission SPECT image, and the second SPECT image is a scattering SPECT image.

3. The generation unit according to Claim 1 generates a third SPECT image by performing attenuation correction (AC) on the first SPECT image sinogram using the attenuation map, performing scatter correction (SC) on the first SPECT image sinogram using the second SPECT image sinogram, and applying resolution recovery (RR) to the first SPECT image sinogram.

4. The apparatus enabling CT-free quantitative thyroid SPECT using a neural network according to Claim 3, wherein the third SPECT image is a quantitative ACSCRR SPECT image.

5. The apparatus enabling CT-free quantitative thyroid SPECT using a neural network according to Claim 3, wherein the calculation unit calculates the uptake amount of the radiopharmaceutical of the thyroid gland by combining the third SPECT image and the thyroid map.

6. The calculation unit calculates the uptake amount of the radiopharmaceutical of the thyroid gland by combining the third SPECT image and the thyroid map and counting the matching voxels, and the apparatus enables CT-free quantitative thyroid SPECT using the neural network according to claim 5.

7. The second model is learned to segment the thyroid map in the input image based on a plurality of thyroid segmentation maps drawn and labeled along the outline of the thyroid gland in a CT (Computed Tomography) image, and the apparatus enables CT-free quantitative thyroid SPECT using the neural network according to claim 1.

8. One or more processors, and A method performed by an apparatus enabling CT-free quantitative thyroid SPECT using a neural network, comprising a memory storing one or more programs executed by the one or more processors, Inputting a first SPECT (Single Photon Emission Computed Tomography) image and a second SPECT image of the thyroid gland into a first model to generate an attenuation map; Inputting the attenuation map and the first SPECT image into a second model to segment the thyroid map; A method enabling CT-free quantitative thyroid SPECT using a neural network, including calculating the uptake amount of the radiopharmaceutical of the thyroid gland using a first SPECT image sinogram, a second SPECT image sinogram, the attenuation map, and the thyroid map.

9. The first SPECT image is a primary emission SPECT image, and the second SPECT image is a scattering SPECT image, and the method enables CT-free quantitative thyroid SPECT using the neural network according to claim 8.

10. The generating step includes attenuating and correcting (Attenuation Correction; AC) the first SPECT image sinogram using the attenuation map, scatter-correcting (Scatter Correction; SC) the first SPECT image sinogram using the second SPECT image sinogram, and applying resolution recovery (Resolution Recovery; RR) to the first SPECT image sinogram to generate a third SPECT image, the method for enabling quantitative thyroid SPECT without CT using the neural network according to claim 8.

11. The method for enabling quantitative thyroid SPECT without CT using the neural network according to claim 10, wherein the third SPECT image is a quantitative ACSCRR SPECT image.

12. The calculating step of the method for enabling quantitative thyroid SPECT without CT using the neural network according to claim 11 includes combining the third SPECT image and the thyroid map to calculate the uptake amount of the radiopharmaceutical in the thyroid.

13. The calculating step of the method for enabling quantitative thyroid SPECT without CT using the neural network according to claim 12 includes combining the third SPECT image and the thyroid map and counting the matching voxels to calculate the uptake amount of the radiopharmaceutical in the thyroid.

14. The second model is trained to segment the thyroid map in the input image based on a plurality of thyroid segmentation maps drawn and labeled along the outline of the thyroid in a CT (Computed Tomography) image, the method for enabling quantitative thyroid SPECT without CT using the neural network according to claim 8.

Images