Sum Tree Structure Motion Correction Algorithm for Medical Images Using 3D to 2D Projection

Abstract

New technologies are disclosed herein that use a motion correction algorithm for 3D medical images using two-dimensional projection to address the ambiguity in medical images caused by the motion of a rigid body (such as a patient) relative to a medical scanning device.

Description

Technical Field

[0001] The present invention relates to medical imaging. More particularly, the present disclosure relates to motion correction of three-dimensional (3D) medical images. Background Art

[0002] Medical images are often used for the diagnosis of medical disorders. With advancements in computers and image processing, 3D medical images are increasingly being used in medical investigations. There are various types of 3D medical imaging techniques, including, for example, computed tomography (CT), positron emission tomography (PET), magnetic resonance imaging (MRI or MR), single photon emission computed tomography (SPECT), and ultrasound, among others.

[0003] Each of the above examples generates 3D medical images in a different manner. For example, a CT scan uses multiple X-ray images taken in multiple directions (i.e., using a scanner positioned at different orientations relative to the patient) to generate a three-dimensional image or multiple tomographic image slices. CT scans typically have higher resolution, shorter scan times, and are superior for providing structural data such as the structure of bones, organs, etc.

[0004] As another example, a PET scan uses a gamma-emitting radiopharmaceutical that is either ingested or injected into the patient. Multiple gamma-ray images are taken in multiple directions to generate a three-dimensional PET image or multiple slices. PET scans typically have lower resolution, but provide more useful information about the functional status of body tissues and systems such as the cardiovascular system. For example, PET is superior for indicating the presence of soft tissue tumors or reduced blood flow to certain organs or regions of the body. PET scans require a relatively long duration data acquisition period, which is approximately several minutes (e.g., about 30 minutes) for typical clinically sufficient images. Generally, during this period, a large amount of PET data acquisition is obtained at many different angles.

[0005] Although CT and PET scans use different techniques to generate medical imaging data, each of these specific examples, as well as other medical imaging techniques, can be vulnerable to both expected and unexpected relative motion that occurs between the scanning instrument and the patient being scanned. Many techniques have been developed to correct for or account for such relative motion. For example, gated scanning, discussed in U.S. Patent No. 9,510,800, has addressed some of the causes of motion-induced blurring in medical images by identifying and utilizing a patient's physiological signals (e.g., respiratory or cardiac signals). By measuring such physiological signals, the expected motion of the patient and / or a particular target within the patient (e.g., the lungs or the heart) can be determined during acquisition. This information can be used to detect time intervals (referred to as gates, time gates, or time windows) of relatively small organ motion during which the image can be acquired, or during which the image data can be accepted for reconstruction of the data set (where data from periods of larger motion is discarded).

[0006] While current techniques (such as the gated scanning described above) have produced some improvements in motion-induced blurring in medical images for predictable motion, such techniques are less useful for unexpected motion (such as a patient sneezing). One current technique for addressing unexpected patient motion uses additional devices (such as cameras) to measure the patient's motion and then uses these measurements to correct the received image data. Unfortunately, the patient's movement can occur so rapidly during a short time period that current methods are unable to correct the data collected during that movement. Often, the data will simply be discarded. If the uncorrected data is not discarded, the final image will be compromised, resulting in a blurred image that is less useful or, in some cases, unusable altogether.

[0007] Accordingly, improved methods for correcting motion-induced blurring in medical images are desired. SUMMARY OF THE INVENTION

[0008] Novel techniques are disclosed herein that address blurring in medical images due to motion of a target object (such as a patient) relative to a medical scanning device during an imaging acquisition session by using a two-dimensional (“2D”) projection for a 3D medical image with a sum tree structure motion correction algorithm.

[0009] According to some embodiments, a computer-implemented method for processing data of medical imaging is disclosed. The disclosed method may be performed by a computer within a medical imaging system tasked with processing image data from an imaging acquisition session. Alternatively, the disclosed method may be performed by a computer system external to the medical imaging system. The method may include: receiving a first data set representing a first three-dimensional medical image and generating a first two-dimensional medical image by projecting the first data set onto a first plane. The method may further include: receiving a second data set representing a second three-dimensional medical image and generating a second two-dimensional medical image by projecting the second data set onto the first plane. A spatial registration existing between the first two-dimensional medical image and the second two-dimensional medical image in the first plane may be calculated. Using the calculated spatial registration, a correction vector may be generated. The correction vector may be applied to one or more datum in the second data set, thereby modifying it. A combined data set representing a processed three-dimensional medical image may be generated by combining the first data set and the modified second data set. The method may be iteratively performed in two additional planes, each plane being orthogonal to the other two planes. Additionally, the method may be iteratively performed for additional data sets, each data set representing a different three-dimensional medical image. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 Illustrates a frame sequence of a PET scan.

[0011] Figures 2A through 2C illustrate reconstructed images of selected frames from the Figure 1 frame sequence.

[0012] Figure 3 Illustrates a flow chart of a computer-implemented method for correcting motion from an image according to some embodiments.

[0013] Figure 4 Illustrates a 2D image of 3D medical imaging data projected onto a plane according to some embodiments.

[0014] Figure 5 Shows a flow chart of another computer-implemented method for correcting motion from an image according to some embodiments.

[0015] Figure 6 Shows a flow chart of another computer-implemented method for correcting motion from an image according to some embodiments.

[0016] Figure 7 Illustrates the results of an iterative process for correcting motion from image blurriness according to some embodiments.

[0017] Figure 8Illustrates the registration of multiple floating images with a single target image according to some embodiments.

[0018] Figure 9 Illustrates the iterative summation of multiple images.

[0019] Figure 10 Illustrates a diagram for rotational and translational correction for accounting for patient motion using a method according to some embodiments.

[0020] Figure 11 Compares various views of uncorrected medical images with various views of images corrected by rotation and translation using Figure 10 thereof.

[0021] Figure 12 Is an architectural diagram of a system that can be used to implement the methods disclosed herein according to some embodiments.

[0022] This application discloses illustrative (i.e., example) embodiments. The claimed invention is not limited to the illustrative embodiments. Accordingly, many implementations of the claims will differ from the illustrative embodiments. Various modifications can be made to the claimed invention without departing from the spirit and scope of this disclosure. The claims are intended to cover implementations having such modifications. Detailed Description

[0023] This description of the exemplary embodiments is intended to be read in conjunction with the accompanying drawings, which are regarded as a part of the entire written description. To facilitate understanding of the principles of this disclosure, reference will now be made to multiple illustrative embodiments in the accompanying drawings, and these embodiments will be described using specific language.

[0024] Figure 1 Illustrates a sequence 100 of twelve PET medical image frames, each frame having a duration of two seconds in the case of time-of-flight backprojection, which is obtained from a 20-minute high-resolution PET brain study using a Siemens Biograph Vision PET scanner in list mode. The frame sequence 100 is sorted chronologically from left to right and top to bottom. Most of these frames (such as frames 102 and 104) occur during periods when the patient moves little to not at all. Thus, the data from these images provides medically relevant details useful in diagnosis and other analyses. On the other hand, frame 106 shows the blurring effect of the rapid movement caused by the patient sneezing during the PET scan. Through the visual comparison of frame 106 with frames 102 and 104 (the frames immediately before and after frame 106), both the change in the patient's position and the effect on the received data are obvious to a person of ordinary skill in the art ("POSA").

[0025] Figures 2A - 2C illustrate different planar views of the reconstructed non - attenuated images 200A - 200C formed from the data received during frame 106 from Figure 1 . More specifically, FIG. 2A illustrates the reconstructed image 200A from 100 milliseconds of data starting at time zero (i.e., at the start of the 2 - second duration that makes up frame 106), FIG. 2B shows the reconstructed image 200B from 100 milliseconds of data starting at 1 second into frame 106, and Figure 2C FIG. 2C shows the reconstructed image 200C from 100 milliseconds of data starting at 1.5 seconds into frame 106. In each of FIGS. 2A - 2C, the left - most image is the reconstructed image observed in the x - y plane, the central image is the reconstructed image observed in the x - z plane, and the right - most image is the reconstructed image observed in the y - z plane. These images are too noisy and too blurred for conventional correction techniques to provide medically relevant data.

[0026] According to some embodiments, Figure 3 a computer - implemented method 300 for correcting motion from medical images is provided in

[0027] . The method begins at blocks 302 and 304. At block 302, a first data set representing a first 3D medical image is received from an image data acquisition component of a medical imaging instrument / systems (such as a scanning detector in a PET, SPECT, CT, MRI (or MR), ultrasound, x - ray imaging system).

[0028] As used herein, a medical image refers to an image constructed by any type of medical imaging system and can include PET, SPECT, CT, MRI (or MR), ultrasound, x - ray, and other medical imaging techniques. Additionally, while embodiments demonstrating the advantages of the present disclosure utilize human patients as the target, POSA will recognize that the advantages of the methods disclosed herein apply to any target of a medical image that can move relative to the imaging device.

[0029] Figure 4Illustrated are six examples (400A to 400F) of 2D medical images formed by projecting a 3D medical image dataset onto various orthogonal planes (i.e., the x-y, x-z, and y-z planes). As understood by POSA, the unprocessed dataset represents a 3D image by storing the x-y-z coordinates of each detected emission event during the period of interest. These coordinates can be used to visually represent the location of each event on a display.

[0030] To project the 3D dataset onto a plane, the data along the axis (e.g., the z-axis) along which the image is being viewed is simply ignored, leaving only the x and y data projected onto the x-y plane. For example, image 400A illustrates only the data projected onto the x-y plane, and this view is along the z-axis. Similarly, image 400B illustrates the x-z projection of the data viewed along the y-axis, and image 400C illustrates the y-z data viewed along the x-axis.

[0031] In Figure 4 the example provided, a set of images 400A, 400B, and 400C are labeled as target ("T") images and are labeled as T z 、T y and T x . As used herein, a target image is the image to which another image is corrected. The other images corrected to the target image are referred to as floating ("F") images. Figure 4 Illustrated are three floating images: 400D, 400E, and 400F. These floating images are labeled using one of the F z 、F y or F x labels. Similar to the target images, floating image 400D illustrates the x and y data projected onto the x-y plane, image 400E illustrates the x and z data projected onto the x-z plane, and image 400F illustrates the y and z data projected onto the y-z plane. For each of the floating images 400D, 400E, and 400F, the "0" refers to the number of iterations of the method described herein that have been applied to the floating image dataset.

[0032] Referring Figure 3 to, the first dataset can represent the target image, and the second dataset can represent the floating image, and each of the first 2D image and the second 2D image represents the projection of the corresponding dataset representative of the 3D medical image onto one of the planes (x-y, x-z, or y-z).

[0033] At block 310, the spatial registration existing between a first 2D medical image and a second 2D medical image in a first plane is calculated. The spatial registration is calculated by comparing the first and second 2D medical images and determining the modifications required to correct or move the second 2D image onto the first 2D image, thereby reducing or removing the effects of patient movement. This determination can be made by a trained technician, a computer algorithm, or both. For example, a technician can view the relative positions of perceptible structures or other features in the two images and manipulate the positioning of one or more of these 2D images until a "best fit" is achieved. Similarly, a computer algorithm can perform a similar process or recommend a best fit, which is reviewed and / or edited by the technician. The system on which these images are manipulated can then measure or be used to measure the translations and / or rotations required to correct the second two-dimensional floating medical image relative to the first two-dimensional medical image.

[0034] For effective comparison with other images to determine spatial registration in three dimensions, a three-dimensional image affected by rapid movement may be too noisy. Projecting the 3D dataset onto a single plane advantageously provides a more effective and reliable determination of the spatial registration between the two datasets, thereby providing better correction for patient movement, as well as a clearer and more usable image from which a diagnosis can be made.

[0035] At block 312, a correction vector based on the calculated spatial registration is generated. The correction vector is a set of values that will be applied to the original 3D data to remove the effects of patient movement. For example, if the floating image is determined to have been translated 2 units in the x direction and 3 units in the y direction in the x - y plane and rotated 10 degrees about the z - axis relative to the target image, the correction vector can be set to translate - 2 units in the x direction, translate - 3 units in the y direction, and rotate - 10 degrees about the Z - axis. Then, at block 314, the second dataset is modified by applying the correction vector to one or more fiducials in the second dataset, thereby removing patient movement from the second dataset.

[0036] If the modification of the second data set sufficiently registers the second 2D image to the first 2D image, which can be observed by noting the minimum spatial differences between the images, method 300 proceeds to block 316. However, if the modification of the second data set does not sufficiently register the second 2D image to the first 2D image, method 300 proceeds to block 320, where a modified second 2D image is generated from the modified second data set. At block 322, a new spatial registration between the first 2D image and the modified second 2D image is computed. Then, method 300 returns to block 312. Method 300 may repeat blocks 312, 314, 320, and 322 until the second 2D image is sufficiently registered. This iterative process results in a more efficient and effective correction of the motion of the object.

[0037] Then, at block 316, a combined data set representing a processed three-dimensional medical image is generated by combining the first data set and the modified second data set, the modified second data set having been modified by the correction vector at block 314. In some embodiments, the combined data set may be transmitted to, for example, a remote location such as an external computer system. In some embodiments, the combined data set may be displayed to show the medical image in which the motion effects have been reduced or removed. Method 300 ends at block 318.

[0038] Figure 5 A computer-implemented method 500 for correcting motion from medical images is illustrated in accordance with some embodiments. Method 500 begins at block 502, where blocks 302 through 310 of Figure 3 are performed. At block 504, a third two-dimensional medical image is generated by projecting the first data set onto a second plane as described above. The second plane is orthogonal to the first plane. At block 506, a fourth two-dimensional medical image is generated by projecting the second data set onto the second plane as described above. At block 508, a second spatial registration existing between the third two-dimensional medical image and the fourth two-dimensional medical image in the second plane is computed as described above. Similarly, blocks 510 and 512 generate fifth and sixth 2D images by projecting the respective data sets onto a third plane that is orthogonal to both the first and second planes, and at block 514, a third spatial registration is computed.

[0039] Method 500 continues at block 516, where, in each of only three planes (e.g., the other two planes as in Figure 4 ), rather than with respect to Figure 3Generate correction vectors based on first, second, and third spatial registrations in only one plane as described above. As POSA will recognize, data in one dimension (e.g., the x dimension) will be affected by the (one or more) multiple correction vectors (and thus the (one or more) computed spatial registrations) generated from multiple planar projection views. For example, data in the "x" dimension will appear in both the x-y and x-z planes. Thus, both spatial registration values computed for each plane will provide inputs into the correction vectors, e.g., by summing these two values together. In some embodiments, these spatial registration values for a given dimension can be averaged together.

[0040] By computing the spatial registration (and thus the patient's movement) in each plane separately from the other planes, more effective and reliable motion correction can be performed because the excessive noise that would appear in 3D medical images is reduced by using 2D projections.

[0041] At block 518, perform Figure 3 blocks 314 to 318 of. Method 500 ends at block 520.

[0042] Figure 6 Illustrates another computer-implemented method 600 for correcting motion from medical images. Figure 5 and Figure 6 The main difference between is that in Figure 5 each of the various generated 2D images and spatial registrations is generated based on the original second 3D dataset, and as shown below, while Figure 6 uses an iterative process where some additional 2D projections are not based on the original second dataset, but rather on the second dataset modified by the (one or more) first (or earlier) correction vectors.

[0043] Method 600 begins at block 602, where at block 602, perform Figure 3 blocks 302 to 314 of. At block 314, the second dataset is modified by the correction vector generated at block 312. At block 604, a third two-dimensional medical image is generated by projecting the first dataset onto a second plane orthogonal to the first plane. At block 606, a fourth two-dimensional medical image is generated by projecting the second dataset modified at block 314 onto the second plane. At block 608, a second spatial registration existing between the third two-dimensional medical image and the fourth two-dimensional medical image in the second plane is computed. At block 610, the second correction vector is created using the second spatial registration. At block 612, the modified second dataset (see block 314) is further modified by applying the second correction vector to one or more fiducials in the modified (block 314) second dataset.

[0044] At block 614, a fifth two-dimensional medical image is generated by projecting the first data set onto a third plane that is orthogonal to the first and second planes. Similarly, at block 616, a sixth two-dimensional medical image is generated by projecting the further modified (block 612) second data set onto a third plane that is orthogonal to the first and second planes. At block 618, a third spatial registration existing between the fifth two-dimensional medical image and the sixth two-dimensional medical image in the third plane is computed. At block 620, a third correction vector based on the third spatial registration is generated. At block 622, the further modified (block 612) second data set is further modified using the third correction vector.

[0045] This iterative process can be further continued in method 600. After the further modified (block 612) second data set is further modified at block 622, if the images are sufficiently registered - which can be observed by noting the minimum spatial differences between the images, method 600 can proceed to block 624, or if the images are not sufficiently registered, method 600 can return to block 602 via line 628. However, block 602 is modified in such a way that the further modified (block 622) second data set is used to generate a second 2D image in the first plane, thus allowing the modification of the image in the first plane to be iterated after the image has been modified through all three planes.

[0046] At block 624, blocks 316 to 318 are executed and method 600 ends at block 626.

[0047] As described above with respect to Figure 6 By iteratively modifying the second data set, more effective and reliable motion correction can be performed because the excessive noise that would appear in the 3D medical image is reduced even more, as noise and movement are removed when modifying the (modified) second data set - which has already been modified with respect to an earlier determined spatial registration.

[0048] In some embodiments, the iterative process of method 600 can further include generating additional 2D images on the first plane from the second data set modified by the methods listed in blocks 602 to 622. This further allows the floating image of the modified second data set to be further registered to the 2D image of the target first data set as projected on the first plane, thus allowing the motion-induced blurring in the second data set to be further reduced. In some embodiments, the iterative process is performed multiple times through each plane until the floating image is sufficiently registered - which can be observed by noting the minimum change in the modification of the data set. Then, the floating image is summed with the target image.

[0049] Referring to Figure 7, an example 700 of iteratively correcting motion-induced blurriness in a medical image is illustrated. In example 700, spatial registration, generation of correction vectors, and modification of the data set are performed through multiple iterations of the above process. Each image represents a 2D projection of the modified data set after the indicated number of iterations. For example, image 702 (i.e., the x-y projection of the original data set), image 704 (x-z projection) after 4 iterations, image 706 (y-z projection) after 8 iterations, image 708 (x-y projection) after 12 iterations, image 710 (x-z projection) after 16 iterations, and image 712 (y-z projection) after 20 iterations, each projection using the data set modified during the previous iteration. As can be seen, this iterative process produces clearer and more useful images because as the number of iterations increases, the resulting floating image aligns better with the target image.

[0050] Although each of the methods described above with respect to Figure 3 , 5 and 6 describes correcting a single floating image (represented by a second data set) to a target image (represented by a first data set), these methods are not limited to using a target image to correct a single floating image. For example, Figure 8 illustrates that multiple floating images 804 to 812 (images Float-1 to Float-5 respectively) are each individually corrected to a target image 802. Each of these floating images 804 to 812 can be corrected to the target image 802 using one of the methods 300, 500, and 600 described above. However, each of the methods 300, 500, and 600 will include generating a combined data set that combines the modified data sets of each of the floating images 804 to 812 with the data set of the target image 802.

[0051] Figure 9 illustrates a correction sequence 900 between multiple images 902 to 914 generated from an original uncorrected data set representing a three-dimensional medical image, and the summed images ("Σ") 918 to 928. Each of these corrections between any two of these images can be performed using one of the methods 300, 500, or 600 described above. For example, the floating image 916 ("Float-7") can be corrected to the floating image 914 using any one of the above methods. In this correction, the floating image 914 ("Float-6") can be considered the target for the correction of the image 916. As Figure 9As shown, the "target" image to which the floating image is corrected is the image pointed to by the arrow, such as image 914. Once image 916 has been corrected to image 914 using the method described above, a new image 918 is generated, which is represented by a set of combinations (or sums) of the data of image 914 and the modified data of image 916. A similar correction process can occur between images 910 and 912 to form image 920, between images 906 and 908 to form image 922, and between images 902 and 904 to form image 924. Then, image 918 can be corrected to image 920 to form image 926, and image 922 can be corrected to image 924 to generate image 928. Finally, image 926 can be corrected to image 928, thus forming the final combined data set.

[0052] Image 902, which has never been corrected to another image, can be considered the Figure 9 overall target image in.

[0053] In some embodiments, for example, with respect to the disclosure of method 300, image 902 can be formed from a third data set representing a third three-dimensional medical image, which is received from an image data acquisition component of a medical imaging instrument / system (such as a scanning detector in a PET, SPECT, CT, MRI (or MR), ultrasound, x-ray imaging system). A two-dimensional medical image can be generated by projecting the third data set onto a first plane. Another two-dimensional medical image can be generated by projecting the combined data set, which is formed by, for example, the correction of images 908 to 906, onto the first plane. The spatial registration can be calculated as described above and used to generate another correction vector, which is applied to one or more fiducials of the combined data set, thereby creating a second modified combined data set including the third data set and the modified combined data set.

[0054] The advantage of aligning and combining datasets of two images (e.g., 914 and 916, neither of which is the final target image 902) is that alignment is achievable even if one or both of the two images have insufficient data or too much noise to be directly summed with image 902. However, if images 914 and 916 are combined, the resulting aligned image (918) can be combined with the final target image (902) either directly or after further alignment, thereby allowing motion-induced blurring caused by patient movement to be corrected from the data. Multiple motion events can also complicate image alignment. Each of these events can produce images that are aligned (or alignable) to other images in the same motion event using the methods disclosed herein, thereby providing sufficient data such that the aligned combined datasets for each motion event can be combined with the aligned combined datasets for other motion events. These alignments result in a more efficient and reliable alignment of medical images compared to conventional methods that struggle with multiple motion events.

[0055] While Figure 9 FIG. illustrates one final target image (902) and seven floating images (904 - 916) formed from an original (i.e., unaligned) dataset representing a three-dimensional medical image, but the methods disclosed herein are not limited to these specific numbers and can be applied to any number of datasets. Any of the methods disclosed above can be used iteratively to modify each dataset to align the motion within the dataset to the target image and add the aligned dataset to a combined dataset that includes multiple aligned datasets. These methods can be used iteratively until all datasets are aligned or sufficiently registered to the "final" target image.

[0056] Referring Figure 1 to FIGS. 2A - 2C, a brain PET scan was performed during which the patient sneezed, resulting in a blurred image in frame 106 and scattered data in FIGS. 200A - 200C. Using the techniques described above, the two-second frame 106 was divided into a sequence of 100-millisecond frames, each frame providing a dataset that forms a 3D medical image. The images were compared and aligned with each other by projecting the data of each image onto a series of planes, calculating spatial registration, generating alignment vectors using the methods described above, modifying the datasets, and generating a combined dataset. The process was iterated to obtain Figure 10 values. Diagram 1002 represents the final rotation through the axis (i.e., R y that is a rotation about the y-axis, R z that is a rotation about the z-axis, and Rx (is a rotation about the x-axis). For example, the values in box 1006 show the rotations about the y, z, and x axes of the data set for the third 100 millisecond frame from top to bottom. As seen in 1002, no rotation values are applied to the first 100 millisecond frame because this frame is the target image.

[0057] Similarly, diagram 1004 shows the x, y, and z translations (S x 、S y and S z ) for each of these frames determined by the method disclosed herein. Diagrams 1002 and 1004 together represent an effective correction vector for modifying the data set of the corresponding frame to remove the influence of patient motion.

[0058] Figure 11 demonstrates the effectiveness of the method disclosed herein in correcting data from frame 106 to remove the influence of motion. The set of images in box 1102 represents the 2D projections of the original (i.e., without motion correction) data that forms frame 106 on three orthogonal planes. The images in box 1104 are produced by applying the corresponding rotation and translation corrections (see Figure 10 ) to each frame, generating a combined data set that includes the target data set within frame 106 and the other 19 frames, and then projecting this combined data set onto the same three orthogonal planes as the images in box 1104. As can be seen, there is a significant improvement in image clarity, resulting in medically meaningful information for analysis and diagnosis. Existing methods for correcting these images would not be able to correct for this motion, either resulting in unclear images (such as the images in box 1102) or the discarding of the data.

[0059] Figure 12 is a block diagram of a system 1200 that can be used in some embodiments, for example, to implement the method disclosed herein. The computer system 1200 can include one or more processors 1202. Each processor 1202 is connected to a communication infrastructure 1206 (e.g., a communication bus, a cross-over bar, or a network), which provides an interface for information communication between the various directly and indirectly connected components of the system 1200. The computer system 1200 can include a display interface 1222 that forwards graphics (e.g., 2D and 3D medical images), text, and other data from the communication infrastructure 1206 (or from a frame buffer, not shown) for display on a display unit 1224 to the user.

[0060] The computer system 1200 may also include a main memory 1204 (such as, a random access memory (RAM)) and an auxiliary memory 1208. The auxiliary memory 1208 may include, for example, a hard disk drive (HDD) 1210 and / or a removable storage drive 1212, which may represent a floppy disk drive, a tape drive, an optical disk drive, a memory stick, etc., as known in the art. The removable storage drive 1212 reads from and / or writes to a removable storage unit 1216. The removable storage unit 1216 may be a floppy disk, a tape, an optical disk, a memory stick, etc. As will be appreciated, the removable storage unit 1216 may include a computer-readable storage medium having data and / or computer software instructions tangibly stored therein (embodied thereon), for example, for causing the (one or more) processors to perform various operations, including the methods disclosed herein.

[0061] In an alternative embodiment, the auxiliary memory 1208 may include other similar devices for allowing computer programs or other instructions to be loaded into the computer system 1200. The auxiliary memory 1208 may include: a removable storage unit 1218 and a corresponding removable storage interface 1214, which may be similar to the removable storage drive 1212 having its own removable storage unit 1216. Examples of such removable storage units include, but are not limited to, USB or flash drives, which allow software and data to be transferred from the removable storage units 1216, 1218 to the computer system 1200.

[0062] The memory systems described above are configured to store, on either a more temporary or a more permanent basis, various data sets representing 3D medical images, which are both data sets in their original form from medical imaging instruments and data sets modified by the (one or more) processors 1202. Additionally, these memories may store data and / or computer software instructions, for example, for causing the (one or more) processors to perform the methods disclosed herein.

[0063] The computer system 1200 may also include a communication interface (e.g., a networking interface) 1220. The communication interface 1220 allows software and data to be transferred between the computer system 1200 and external devices, such as to other remote and / or external computer systems and medical imaging instruments. In some embodiments, the computer system 1200 may be part of a medical imaging system that includes a medical imaging instrument / system that includes an image data acquisition component, such as a scanning detector in a PET, SPECT, CT, MRI (or MR), ultrasound, x-ray imaging system. Examples of the communication interface 1220 may include a modem, an Ethernet card, a wireless network card, a Personal Computer Memory Card International Association (PCMCIA) slot and card, etc. The software and data transferred via the communication interface 1220 may be in the form of signals, which may be electronic, electromagnetic, optical signals, etc. that can be received by the communication interface 1220. These signals may be provided to the communication interface 1220 via a communication path (e.g., a channel), which may be implemented using wires, cables, optical fibers, telephone lines, cellular links, radio frequency (RF) links, and other communication channels.

[0064] In some embodiments, the methods disclosed herein may be stored as instructions on a non-transitory computer-readable storage medium (e.g., the removable storage unit 1216). When the stored instructions are executed by the processor 1202 (or processors) in the system 1200, it causes the system 1200 to execute methods 300, 500, and / or 600. For example, the processor 1202 is configured to: receive various raw data sets representing three-dimensional medical images from an image data acquisition component of a medical imaging instrument / system, such as a scanning detector in a PET, SPECT, CT, MRI (or MR), ultrasound, x-ray imaging system, generate two-dimensional medical images by projecting the raw data sets onto one or more planes, calculate the spatial registration existing between two images, generate a correction vector based on the spatial registration, modify the data sets by applying the correction vector, and generate a combined data set representing the processed three-dimensional medical image by combining the target data set and the modified second data set.

[0065] In addition, the display interface 1222 may cause the display 1224 to display medical images according to the methods disclosed herein, and the communication interface 1220 may be used to receive data sets from an imaging device / apparatus and transmit the combined data set externally, such as transmitting the combined data set to an external computer system.

[0066] However, a non-transitory computer-readable medium (e.g., removable storage unit 1216) is not limited to use only in computer system 1200 and can be used in many other systems or devices such that instructions tangibly embodied thereon, when executed, are configured to cause one or more processors of those systems to execute methods 300, 500, and / or 600 described herein.

[0067] In any of the foregoing embodiments, the disclosed method may further comprise: receiving a plurality of data sets from a medical imaging instrument, each data set representing an additional three-dimensional image, wherein each data set of the plurality of data sets is iteratively used to generate an additional 2D image in a first plane, calculating an additional spatial registration present between the additional 2D image of the corresponding data set and the 2D image generated from any previously derived combined data set, generating an additional correction vector based on the additional spatial registration, modifying the corresponding data set of the plurality of data sets by applying the additional correction vector, and generating a further modified combined data set by combining the corresponding modified data set with the previously derived combined data set.

[0068] In any of the foregoing embodiments, the method may further comprise: performing the steps of the immediately preceding paragraph in two additional planes, each additional plane being orthogonal to the first plane and the other additional plane.

[0069] In any of the foregoing embodiments, the described data sets may be produced from a medical imaging instrument, which may be one of a PET, SPECT, CT, MR, x-ray, and ultrasound imaging system.

[0070] In any of the foregoing embodiments, the generation of the correction vector may be based on one or more of any calculated spatial registrations.

[0071] In any of the foregoing embodiments, calculating the spatial registration may comprise: measuring a translation of a second two-dimensional medical image relative to a first two-dimensional medical image, and measuring a rotation of the second two-dimensional medical image relative to the first two-dimensional medical image.

[0072] In any of the foregoing embodiments, a combined data set comprising one or more fiducials corrected to reduce and / or remove the effects caused by target motion may be transmitted to an external computer system.

[0073] In any of the foregoing embodiments, the combined data set may be displayed as a 3D rendering of the data sets.

[0074] The above describes methods and systems for correcting motion-induced blurring in medical images. These methods and systems are applicable to all types of medical imaging and, in particular, produce clearer, more effective, and more efficiently generated medical images in situations of extreme and / or rapid motion occurring within a short period of time for low count rate scans such as low-dose brain PET scans. Such methods and systems can further be used in conjunction with data-driven motion detection and artificial intelligence motion correction.

[0075] Although the subject matter has been described in accordance with exemplary embodiments, the subject matter is not limited thereto. Rather, the appended claims should be construed broadly to include other variations and embodiments made by those skilled in the art within the scope and ambit of equivalents of the claims.

Claims

1. A computer-implemented method for correcting motion from medical images by processing medical image data, the method comprising: Receiving a first data set from an imaging acquisition session representing a first three-dimensional medical image generated by a medical imaging instrument; Receiving a second data set from the imaging acquisition session representing a second three-dimensional medical image generated by the medical imaging instrument; generating a first two-dimensional medical image by projecting the first data set onto a first plane; Generating a second two-dimensional medical image by projecting the second data set onto the first plane; Calculating a spatial registration existing between the first two-dimensional medical image and the second two-dimensional medical image in the first plane; Generating a correction vector based on the spatial registration; Modifying the second data set by applying the correction vector to one or more fiducials in the second data set; Generating a combined data set representing a processed three-dimensional medical image by combining the first data set and the modified second data set; Receiving a third data set from the imaging acquisition session representing a third three-dimensional medical image generated by the medical imaging instrument; generating a third two-dimensional medical image by projecting the third data set onto the first plane; Generating a fourth two-dimensional medical image by projecting the combined data set onto the first plane; Calculating a second spatial registration existing between the third two-dimensional medical image and the fourth two-dimensional medical image in the first plane; Generating a second correction vector based on the second spatial registration; Modifying the combined data set by applying the second correction vector to one or more fiducials in the combined data set; And Generating a second combined data set representing a second processed three-dimensional medical image by combining the third data set and the modified combined data set.

2. The computer-implemented method according to claim 1, wherein the medical imaging instrument is one of a PET, SPECT, CT, MR, x-ray, and ultrasound imaging system.

3. The computer-implemented method according to claim 1, further comprising: Generating a third two-dimensional medical image by projecting the first data set onto a second plane orthogonal to the first plane; Generating a fourth two-dimensional medical image by projecting the second data set onto the second plane; Calculating a second spatial registration existing between the third two-dimensional medical image and the fourth two-dimensional medical image in the second plane; Generating a fifth two-dimensional medical image by projecting the first data set onto a third plane orthogonal to both the first and second planes; Generating a sixth two-dimensional medical image by projecting the second data set onto the third plane; Calculating a third spatial registration existing between the fifth two-dimensional medical image and the sixth two-dimensional medical image in the third plane; and Wherein generating the correction vector is further based on the second and third spatial registrations.

4. The computer-implemented method according to claim 3, wherein generating the correction vector based on the first, second, and third spatial registrations comprises: Combining the vectors of the first, second, and third spatial registrations with each other.

5. The computer-implemented method according to claim 1, further comprising: Receiving a plurality of data sets from a medical imaging instrument, each data set representing an additional three-dimensional image, wherein each data set of the plurality of data sets is iteratively used to generate an additional 2D image on a first plane, calculating an additional spatial registration existing between the additional 2D image of the corresponding data set and the 2D image generated from any previously derived combined data set, generating an additional correction vector based on the additional spatial registration, modifying the corresponding data set of the plurality of data sets by applying the additional correction vector, and generating a further modified combined data set by combining the corresponding modified data set with the previously derived combined data set.

6. The computer-implemented method according to claim 1, wherein calculating the spatial registration comprises: Measuring a translation of a second two-dimensional medical image relative to a first two-dimensional medical image; And Measuring a rotation of a second two-dimensional medical image relative to a first two-dimensional medical image.

7. The computer-implemented method according to claim 1, further comprising: Transmitting the combined data set to an external computer system.

8. The computer-implemented method according to claim 1, further comprising: Displaying an image defined by the combined data set.

9. A system for processing data of medical images, the system comprising: A processor configured to execute the computer-implemented method according to any one of claims 1 to 8; And A memory configured to store data sets representing three-dimensional medical images generated by a medical imaging instrument.

10. The system according to claim 9, wherein the system further comprises a communication interface configured to: Transmit the combined data set to a location external to the system; and Receive data sets representing three-dimensional medical images generated by a medical imaging instrument.

11. The system according to claim 9, wherein the system further comprises a display configured to display two-dimensional and three-dimensional medical images.

12. A non-transitory computer-readable medium comprising instructions tangibly embodied therein, the instructions being configured to cause one or more processors to execute the computer-implemented method according to any one of claims 1 to 8 when executed.

Images