Control method and X-ray CT device
By using machine learning models in an X-ray CT device to detect anatomical landmarks and automatically adjust the tilt angle of the stand, the lens protection problem in head scanning is solved, and efficient radiation reduction and throughput improvement is achieved.
Patent Information
- Application Number
- CN202410652618.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-05-24
- Filing Date
- 2024-05-24
- Publication Date
- 2024-11-26
AI Technical Summary
The existing X-ray CT devices are difficult to effectively protect the lens of the eye during head scans, resulting in increased radiation radiation, and the positioning technology is complex and time-consuming, affecting the throughput of the subject.
By using a rotating CT stand in the X-ray CT device, performing 3-dimensional area positioning CT scan of the subject, using machine learning models to detect anatomical landmarks, determine the inclination angle of the stand, and automatically adjust the stand to avoid lens irradiation, realizing automation CT scan for diagnosis.
It reduces radiation exposure to organs with high radiation sensitivity, improves the throughput of the subject, simplifies the lens protection process, and reduces the risk of human error.
Smart Images

Figure CN119014885A_ABST
Abstract
Description
[0001] Related Application:
[0002] This application enjoys the benefit of priority to U.S. patent application No. 18 / 323,016 filed on May 24, 2023, the entire contents of which are incorporated herein by reference. Technical Field
[0003] The embodiments disclosed in this specification and the drawings relate to a control method and an X-ray CT apparatus.
[0004] The embodiments described in this specification generally relate to an X-ray CT (Computed Tomography) apparatus and method, and more specifically to an X-ray CT apparatus and method having automatic gantry tilting for eye protection. Background Art
[0005] An X-ray CT device is a medical imaging diagnostic device that obtains projection data of a subject from multiple angles by rotating an X-ray source at high speed relative to the subject placed on a table. The X-ray CT device reconstructs the multiple projection data obtained from various angles to generate a 2D CT image and / or a 3D CT image.
[0006] In a scanning mode called spiral scanning, the X-ray source is continuously rotated while the relative positional relationship between the X-ray source and the subject is changed, so that projection data of the entire body of the subject can be obtained in one scan. When an X-ray CT apparatus is used to image a subject, the subject is generally placed on a table so that the body axis direction of the subject (i.e., the direction from the head to the feet) is along the length direction of the table (i.e., the z-axis direction of the device coordinate system described later), and scanning is performed while the table is moved in the z-axis direction.
[0007] In an examination using an X-ray CT apparatus, a pre-scan (scanogram CT scan) is performed to obtain projection data of a scout image or a scout image before the main scan (diagnostic CT scan) to obtain projection data of a medical image for image diagnosis. The scout image or scout image is an image obtained at a low radiation dose before the main scan in order to determine imaging conditions of the main scan such as the radiation dose and imaging range. It has the function of assisting the centering and positioning of the subject and selecting the anatomical target for the next CT scan.
[0008] However, conventional 2D positioning images usually only provide the LR and AP directions of radiographs. Therefore, positioning images have the limitation of being difficult to depict soft tissue organs and not being optimal for tube current modulation. On the other hand, in recent years, ultra-low-dose 3D positioning images have been seen to overcome the limitations of 2D positioning images (such as better soft tissue resolution and organ-based modulation) by providing complete 3D anatomical information while maintaining a sufficiently low dose (<1mGy).
[0009] The radiation exposure to the lens of the eye during head and neck CT scans may increase the risk of cataracts and opacities in the patient. Please refer to Nikupaavo, U. et al., Lens Dose in Routine Head CT: Comparison of Different Optimization Methods With Anthropomorphic Phantoms. AJR 2015; 204: 117-123, and Mosher, E. et al., Lens Dose Reduction by Patient Posture Modification During Neck CT. AJR 2018; 210: 1111-1117. The radiation exposure to the lens and radiation-sensitive tissues can be reduced by using technical solutions such as organ-based tube current modulation (OBTCM), local external shielding, and optimization of the gantry tilt angle. According to research, gantry tilt is an effective method for reducing direct radiation exposure to the lens of the eye in brain CT scans without affecting image quality. See also Mussmann, B. et al., Organ-based tubecurrent modulation in chest CT. A comparison of three vendors. Radiography 2021 Feb; 27(1): 1-7. X-ray CT devices sometimes include a gantry tilt function in the system to reduce direct radiation exposure to the lens of the subject during routine brain scans.
[0010] Gantry tilt or subject positioning (head tilt) is part of basic CT technician training, and tilt has been shown to help reduce radiation dose to the lens, but compliance has long been observed to be low in most clinical settings. See Filice, R., Lens Identification to Prevent Radiation-Induced Cataracts Using Convolutional Neural Networks. Journal of Digital Imaging (2019) 32: 644-650; and Deshpande, H. et al., Deep Learning for the Detection of Landmarks in Head CT Images and Automatic Quality Assessment. Proc. SPIE 11596, Medical Imaging 2021: Image Processing, 115960N (15 February 2021); doi: 10.1117 / 12.2581810. According to a traceability study by Georgetown University Hospital, only approximately 10% of head CT scans follow the guidance for lens protection. This may be because the positioning technique requires manual work based on laser marking (and / or 2D positioning images) to exclude the lens. This is difficult and may be slightly subjective. In addition, the process of adjusting the gantry tilt is relatively time-consuming compared to the total CT acquisition time. In addition, the movement of the subject after positioning may cause the lens to be exposed to unwanted radiation.
[0011] Therefore, it is desired to provide a technology for automatically setting imaging conditions for automating CT head examination based on positioning of the subject and tilting of the gantry in order to appropriately protect the lens and improve the throughput of the entire subject. Summary of the invention
[0012] An object of the present invention is to provide a control method and an X-ray CT apparatus capable of reducing radiation exposure to an organ with high radiation sensitivity.
[0013] The control method of the embodiment includes the following steps: controlling a CT (Computed Tomography) gantry that supports an X-ray source and an X-ray detector so as to be rotatable, performing a positioning CT scan of a three-dimensional region of the head of a subject, thereby generating image data; inputting cross-sectional image data of the three-dimensional region generated based on the image data into a learning completion model, thereby detecting anatomical landmarks, which are used to determine an inclined surface for avoiding irradiation of the lens of the subject in a diagnostic CT scan; determining an inclination angle of the CT gantry based on the anatomical landmarks and the inclined surface; and controlling the CT gantry to be tilted at the determined inclination angle to perform the diagnostic CT scan of the subject.
[0014] Effect of the invention:
[0015] According to the control method and the X-ray CT apparatus of the embodiment, it is possible to reduce the radiation exposure to organs with high radiation sensitivity. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 It is an overall block diagram showing the configuration of an X-ray CT apparatus 1A in the first embodiment.
[0017] Figure 2 It is a schematic side view for explaining the tilt mechanism of the X-ray CT apparatus 1A according to the first embodiment.
[0018] Figure 3 This is a block diagram of a processing circuit in the CT apparatus according to the first embodiment.
[0019] Figure 4A It is a schematic diagram showing the biomarkers and slope lines in the first to fourth embodiments of the present method.
[0020] Figure 4B It is a schematic diagram showing the biomarkers and slope lines in the first to fourth embodiments of the present method.
[0021] Figure 4C The shift of the subject in the first to fourth embodiments of the present method is shown.
[0022] Figure 4D The following shows the displacement of the subject in the first to fourth embodiments of the present method.
[0023] Figure 5A This is a flowchart involved in this embodiment.
[0024] Figure 5B This is a flowchart involved in this embodiment.
[0025] Figure 5CThis is a flowchart involved in this embodiment.
[0026] Figure 5D This is a flowchart involved in this embodiment.
[0027] Figure 6 It is a diagram showing a neural network in the first embodiment.
[0028] Figure 7 It is a diagram showing the training of the neural network in the first embodiment.
[0029] Figure 8 This is a flowchart of a modification of the method for controlling the CT scan according to the first embodiment.
[0030] Fig. 9 The diagram shows three head X-ray CT images in which labels are added to the biomarkers, the inclined surfaces, and the estimated inclined surfaces.
[0031] Fig.10 It is a diagram showing the inclined surface to which a label is added and the estimated inclined surface.
[0032] Fig.11 Is getting Figure 9-10 The image shown is a diagram of the phantom used. DETAILED DESCRIPTION
[0033] The embodiment of the present disclosure relates to a method for controlling a computed tomography (CT) scan, comprising the steps of: using a CT gantry that supports both an X-ray source and an X-ray detector so as to be rotatable, performing a scout CT scan (a positioning CT scan for obtaining projection data of a positioning image or a scout image) of a 3D region of the head of an object to be examined, thereby generating image data. The cross-sectional image data of the 3D region generated based on the image data is input into a trained machine learning model, thereby detecting anatomical landmarks in order to determine an inclined surface positioned to avoid irradiation of the lens of the object in a subsequent diagnostic CT scan. Based on the detected anatomical landmarks, the inclination angle of the CT gantry is determined. Using the CT gantry tilted at the determined inclination angle, a diagnostic CT scan of the object is performed.
[0034] The scout CT scan and the diagnostic CT scan may be performed continuously in one scanning operation. In another embodiment, the step of performing the scout CT scan includes the step of increasing the amount of X-rays in order to obtain a specific view with a specific rotational position in order to determine the OMBL.
[0035] The inclined plane may be defined by the auditory canthal line (Orbitomeatal Base Line: OMBL) of the subject, or may be defined by the supraorbital ridge and the opisthion of the subject.
[0036] Another embodiment of a method for controlling a CT scan includes the following steps: using a CT gantry that supports both an X-ray source and an X-ray detector so as to be rotatable, performing a scout CT scan of a three-dimensional (3D) region of an object to be examined, thereby generating image data. Inputting cross-sectional image data of the 3D region generated from the image data into a trained machine learning model, thereby determining anatomical landmarks in the image data. Based on the determined anatomical landmarks, determining a tilt angle of the CT gantry, and performing a diagnostic CT scan of the object using the CT gantry tilted at the determined tilt angle.
[0037] Still another embodiment of the present disclosure relates to an X-ray CT apparatus having a gantry, an X-ray source and an X-ray detector mounted on the gantry, and a memory storing a trained model configured to detect anatomical landmarks. In addition, the apparatus includes a processing circuit that instructs the apparatus to perform a scout CT scan (spot CT scan) of a three-dimensional (3D) region of an object to be examined to generate image data, inputs the image data to the trained model for detecting anatomical landmarks, determines an inclination angle of the CT gantry based on the detected anatomical landmarks, and performs a diagnostic CT scan of the object using the CT gantry tilted at the determined inclination angle.
[0038] Hereinafter, an X-ray CT apparatus according to each embodiment will be described with reference to the drawings.
[0039] Since the imaging range differs depending on the gender and / or body shape of the subject, the imaging range is preferably set appropriately for each subject. In conventional methods of setting imaging conditions, various template images are used from which anatomical landmarks (hereinafter referred to as AL) stored in a database are extracted.
[0040] In each template image, an imaging range corresponding to an anatomical part is defined. Therefore, by performing coordinate transformation on the imaging range defined in the template image based on the AL extracted from the positioning image and the AL of the template image, the imaging range of the subject can be determined.
[0041] The X-ray CT apparatus can image the subject in one of various scanning modes such as helical scanning and rotation scanning. For example, variable helical pitch scanning (hereinafter referred to as vHP scanning) is known as one of the scanning modes of the X-ray CT apparatus. In the vHP scanning, the imaging conditions are switched for each predetermined imaging range, and a wide range is imaged.
[0042] When the whole body is imaged using vHP scanning, an electrocardiographic synchronized scan is performed in an imaging range including the heart. In electrocardiographic synchronized scan, for example, imaging conditions are set so that projection data is acquired in a cardiac phase that is less affected by the heartbeat during one cardiac cycle. In other words, projection data of a predetermined cardiac phase required for image reconstruction is acquired during multiple cardiac beats. In this way, in the imaging range including the heart in the vHP scan, the moving speed of the table is set to be slower than that of other imaging ranges that do not include the heart.
[0043] If the table moves slowly, the time the subject is exposed to X-rays becomes longer, and the X-ray radiation received by the subject increases. However, if the table moves more than the X-ray beam width in the slice direction during one heartbeat cycle, the reconstructed cardiac CT image contains missing parts in the Z-axis direction.
[0044] Based on the above reasons, the inventors have come up with a revolutionary structure that automatically sets the imaging range as small as possible within the range of maintaining image quality that meets diagnostic requirements. According to this structure, the imaging range exposed to a large dose of X-rays is minimized while maintaining the required image quality for the acquired CT image, thereby reducing the total radiation dose in the entire imaging.
[0045] Hereinafter, an embodiment of an X-ray CT apparatus having a configuration for setting an imaging range using the above-described method will be described.
[0046] (First embodiment)
[0047] Figure 1 It is an overall block diagram showing the configuration of an X-ray CT apparatus 1 in the first embodiment. Figure 1 The X-ray CT apparatus 1 shown includes a scanner 11 and a console 12 .
[0048] The scanner 11 is installed in the examination room, and acquires projection data of the subject P. The console 12 is installed in an operation room (control room) adjacent to the examination room, and reconstructs the projection data input from the scanner 11 to generate a CT image.
[0049] The scanner 11 includes a rotating body 25, a stage driving device 31, a high voltage power supply 41, an aperture driving device 42, a rotation driving device 43, a tilt controller 44, a scanning controller 45, and a communication device 46. For example, the communication device 46 may be a speaker / microphone for communicating with the subject.
[0050] The rotating body 25 is a frame (i.e., a housing) configured to support the X-ray tube 21, the X-ray detector 23, and a cooler (not shown) described later as being rotatable, and is fixed to the base 27 by a fixed base (not shown). Under the control of the scanning controller 45, the rotation drive device 43 rotates the rotating body 25 around a rotation axis parallel to the z-axis of the device coordinate system.
[0051] As an example, the x-axis, y-axis, and z-axis of the device coordinate system are defined as follows. The vertical direction is defined as the y-axis direction, the z-axis direction is defined as a direction orthogonal to the y-axis and parallel to the horizontal movement direction (i.e., the length direction) of the stage, and the x-axis direction is defined as a direction orthogonal to both the y-axis and the z-axis. In this embodiment, unless otherwise specified, it is assumed that the subject P is placed on the stage in such a manner that the body axis direction thereof coincides with the z-axis direction.
[0052] The rotating body 25 and the fixed base are tilted in the z-axis direction of the device coordinate system by the tilt mechanism 26 provided between the fixed base and the base 27. The tilt mechanism 26 includes a tilt bearing (not shown). The rotating body 25 and the fixed base rotate integrally around the tilt bearing. The tilt mechanism 26 tilts the rotating body 25 according to the control signal input from the tilt controller 44 under the control of the scanning controller 45. Figure 2 The details of the tilting operation of the rotating body 25 will be described.
[0053] The rotating body 25 includes an opening. The table driving device 31 moves the table 30 on which the subject P is placed into the opening of the rotating body 25 during imaging, and moves the table 30 to the outside of the opening after imaging is completed. Here, as an example, the negative direction of the z-axis is defined as the direction in which the table 30 moves from the outside of the opening to the inside of the opening. In other words, the positive direction of the z-axis is defined as the direction in which the table 30 moves from the inside of the opening to the outside of the opening. The support table 32 is set on the ground to support the table 30.
[0054] The stage driving device 31 includes a stage moving motor and a stage moving mechanism (not shown). Under the control of the scanning controller 45, the stage driving device 31 moves the stage 30 up and down in the y-axis direction and moves the stage 30 horizontally in the z-axis direction.
[0055] The stage moving mechanism is composed of, for example, a rack and pinion mechanism. The stage 30 moves along the rack meshing with the pinion by the rotational force of the pinion (gear) generated by the stage moving motor. Furthermore, the stage driving device 31 includes a stepping motor (not shown) and transmits position control information such as the current position of the stage 30 and the movement amount of the stage 30 detected by the stepping motor to the scanning controller 45.
[0056] The rotating body 25 includes an X-ray tube 21 , an X-ray optical system 22 , an X-ray detector 23 , and a data acquisition system (DAS) 24 .
[0057] The X-ray tube 21 generates electron beams from its cathode in accordance with the tube voltage supplied from the high voltage power supply 41, and causes the electron beams to collide with the metal target (i.e., its anode), thereby generating X-rays. The X-rays generated by the X-ray tube 21 are radiated toward the X-ray detector 23. Under the control of the scan controller 45, the X-ray tube 21 is supplied with power required to radiate X-rays.
[0058] The X-ray optical system 22 controls the irradiation range of X-rays radiated to the subject P based on the control signal from the aperture driving device 42 under the control of the scanning controller 45. The X-ray optical system 22 is equipped with various control devices configured to control the irradiation conditions such as the dose, irradiation range, shape and line quality of the X-ray beam. Specifically, the X-ray optical system 22 is equipped with a wedge filter and a collimator, for example. The wedge filter is a filter formed of a convex lens and a light metal such as aluminum, and adjusts the dose of X-rays generated by the X-ray tube 21. The collimator is a slit for reducing the irradiation range of X-rays, and the dose of X-rays is controlled and adjusted by the wedge filter.
[0059] As an example of the X-ray detector 23, the X-ray detector 23 may also be a two-dimensional array type detector in which a plurality of detection elements are arranged in a matrix along the X-axis direction (i.e., the channel direction) and the slice direction. The X-ray detector 23 detects X-rays radiated from the X-ray tube 21. In addition, the slice direction is orthogonal to the channel direction. Here, as an example, the slice direction is the rotation axis direction of the rotating body 25, and is consistent with the Z-axis direction when the rotating body 25 is not tilted, i.e., when the tilt angle described later is 0 degrees. Furthermore, regardless of the tilt state of the rotating body 25, the channel direction is consistent with the X-axis direction.
[0060] A 2D array detector is also called a multi-slice detector. When the X-ray detector 23 is a multi-slice detector, by rotating the rotating body 25 360 degrees or 180+α degrees, a 3D region having a width in the slice direction can be scanned. Using a 2D array detector, a 3D positioning image or a scout image can be obtained.
[0061] The DAS 24 includes an amplifier and an A / D (analog / digital) converter (not shown), amplifies the signal of the transmission data of the X-rays detected by each detection element of the X-ray detector 23, and converts the amplified transmission data into a digital signal. The DAS 24 transmits the projection data generated based on the transmission data detected by the X-ray detector 23 to the scan controller 45 via a data forwarding device (not shown).
[0062] The console 12 of the X-ray CT apparatus 1 is computer-based and can communicate with external devices via a network such as a LAN (Local Area Network). The console 12 is composed of hardware such as a processing circuit 51 , a memory 52 , an input interface 53 , and a display 55 .
[0063] The processing circuit 51 is mutually connected to each hardware component constituting the console 12 via a bus which is a transmission path of a common signal. In addition, the console 12 includes a storage medium drive as needed.
[0064] The processing circuit 51 may be configured by dedicated hardware or may be configured to realize various functions by software processing of a built-in processor. Here, as an example, a case where the processing circuit 51 realizes various functions by software processing of its processor will be described.
[0065] The term "processor" mentioned above means, for example, a dedicated or general-purpose CPU, a dedicated or general-purpose GPU (Graphics Processing Unit), an Application Specific Integrated Circuit (ASIC), a programmable logic device, and a Field Programmable Gate Array (FPGA) and other circuits. As the above-mentioned programmable logic device, for example, a simple programmable logic device (SPLD) and a complex programmable logic device (CPLD) can be cited. The processing circuit 51 reads the program stored in the memory 52 and implements various functions by executing the program. As an additional or alternative method, the processing circuit 51 reads the program stored in its own processor and implements various functions by executing the program.
[0066] Furthermore, the processing circuit 51 can be composed of a single processor or a combination of multiple independent processors. In the latter case, multiple memories 52 corresponding to each processor can also be set, and the program executed by each processor is stored in the memory 52 corresponding to the processor. As a further change, one memory 52 can also store all the programs corresponding to the functions of multiple processors. The memory 52 is a storage medium including external storage devices such as HDD (Hard Disk Drive) and optical disk devices in addition to RAM (Random Access Memory) and ROM (Read Only Memory). The memory 52 stores image data, data required for executing the program, and various programs (including application programs and operating systems) executed by the processing circuit 51. Furthermore, the memory 52 can store programs for controlling the operating system. Furthermore, the memory 52 can also store programs for GUI (Graphical User Interface) that assists input from the input interface 53.
[0067] The input interface 53 is composed of input devices such as a keyboard, a mouse, a joystick, and a trackball. The input interface 53 receives input from the user via the input devices. If at least one input device is operated by the operator, the input interface 53 generates an input signal corresponding to the operation and outputs the input signal to the processing circuit 51.
[0068] The display 55 is a display device such as a liquid crystal display panel, a plasma display panel, an organic EL (ElectroLuminescence) panel, etc. The display 55 displays images such as imaging conditions and positioning images under the control of the processing circuit 51 .
[0069] Figure 2 It is a schematic side view of the X-ray CT apparatus 1 according to the first embodiment for explaining the tilt mechanism. Figure 2 It is a side view of the scanner 11 viewed from the x-axis direction in a state where the rotating body 25 is tilted. Figure 2 The dot-dash line L1 is a straight line parallel to the y-axis and passes through the center of the tilt mechanism 26. The dot-dash line L2 is a straight line on the yz plane and passes through the center of the tilt mechanism 26. The tilt angle θ is defined as the angle between the dot-dash line L1 and the dot-dash line L2.
[0070] exist Figure 2 , the case where the rotating body 25 is tilted in the negative direction along the z-axis (i.e., backward tilt) is described, but depending on the situation, the rotating body 25 may be tilted in the positive direction along the z-axis (i.e., forward tilt). When backward tilt is defined as the positive tilt direction and forward tilt is defined as the negative tilt direction, the rotating body 25 is tilted in the positive tilt direction or the negative tilt direction within a predetermined angle range in such a manner that the reference angular position of the rotating body 25 becomes the tilt angle 0 degree where the dot-dash line L2 coincides with the dot-dash line L1.
[0071] Figure 3 Indicates the details of the processing circuit 51. The processing circuit 51 includes a scan controller function 60, an image reconstruction function 61, a deep convolutional neural network (DCNN) 62, and an inclined surface generation function 63. These functions can be programs stored in a memory and executed by a processor, or can be hard-wired to the processor. In addition, the scan controller function 60 is an example of a generation unit and a control unit. In addition, DCNN62 is an example of a detection unit and an example of a learning completion model (a trained machine learning model). In addition, the inclined surface generation function 63 is an example of a determination unit.
[0072] The DCNN62 described in detail below extracts and outputs points ((x, y, z) coordinates) in the 3D positioning image body corresponding to the desired biomarker. In addition, DCNN62 can output scan lines or scan planes. In one embodiment, the center of the supraorbital ridge of the left and right eyes and the center of the external auditory canal are extracted and represented as P1 to P3, respectively. These three points P1 to P3 define the desired scan plane. The supraorbital ridge points P1 to P2 can be determined using the average value or center of the eye socket. Other points such as the inner plate of the posterior edge of the foramen magnum or the posterior cranial point (Opisthion of the Occipital Bone: OO) of the occipital skull can be extracted to define the scan plane. In another embodiment, when the subject has an implant or has undergone head surgery, the location or area of the implant or surgery can be used as a biomarker when it is necessary to avoid irradiation of these locations during the diagnostic scan.
[0073] Figure 4A and Figure 4B As an example, a biomarker, an inclined surface, and an inclined line are shown. Figure 4A This is a front view of the skull with the supraorbital ridge points P1 to P2 confirmed. The line L3 between the points is within the scanning plane. Figure 4B In FIG. 1 , OMBL is represented as an inclined line L4 defined by one of the supraorbital ridge points P1 to P2 and the point P3 of the external auditory canal. The inclined surface is defined by three points P1 to P3 (or two lines L3 to L4). Line L5 represents the inclined surface defined by the points P1 to P2 and the point corresponding to the posterior cranial point (OO) of the occipital skull. Whichever surface is used, the angle of the inclined surface can be determined.
[0074] Figure 5A : is a flowchart representing the method involved in the first embodiment. In the X-ray CT device 1, the subject is configured and positioned in the scanner 11 (step 70), and a 3D positioning image is obtained (step 71). The processing circuit 51 inputs the 3D positioning image to the DCNN 62, and detects biomarkers from the 3D positioning image (step 72). In the case of head CT, points defined as the posterior cranial point of the occipital skull, the external auditory canal, and the center of the supraorbital ridge of the left and right eyes can become biomarkers. For example, please also refer to the American Association of Physicists in Medicine, Adult routine headCT protocolsversion 2.0 (2016). In order to avoid or reduce the radiation received by the lens of the eye, the scanning angle can be set to be parallel to the line formed by the inner plate of the lower orbit and the posterior edge of the foramen magnum. In one embodiment, the center of the supraorbital ridge and the external auditory canal is used to determine the auditory canthal line (Orbitomeatal Base Line: OMBL).
[0075] The processing circuit 51 executes the inclined plane generation function 63 to generate an inclined plane based on the coordinates of the biomarker derived from the positioning image. The orientation of the inclined plane relative to the mechanical axis is determined by the inclined plane generation function 63. The scan controller function 60 generates an instruction to the scan controller 45 that controls the tilt controller 44 and moves the gantry to an appropriate angle. In addition, as needed, it generates an instruction to the table drive device 31 for configuring the subject in an appropriate position, and performs a CT scan with the determined gantry tilt angle and table position to minimize the radiation received by the lens and obtain the desired head image. As described above, it is preferable to determine the gantry tilt angle based on the plane defined by the biomarker.
[0076] return Figure 5A In step 73, the processing circuit generates an inclined plane using the detected biomarkers. In step 74, the processing circuit instructs the tilt controller 44 to tilt the gantry. In step 75, the processing circuit also sends an instruction to the stage drive device 31 and arranges the subject in an appropriate position in the z direction, thereby enabling the subject to be moved, as needed. Thereafter, a diagnostic CT scan of the subject is performed in step 75. In order to optimize the workflow and ensure the implementation of safety protocols for protecting the lens, the scout CT scan and the diagnostic CT scan can be performed as part of an overall scanning operation.
[0077] In addition, the angles of lines L3 and L4 may obviously change depending on the positioning of the subject and the movement after positioning. With respect to line L3, if the head of the subject deviates to the left or right relative to the desired position, the determined scanning plane may expose one side of the lens to radiation during the diagnostic scan. This is because the gantry is configured to tilt only around the axis in the x direction (see Figure 2 ). Figure 2 The stand is not configured to be tilted about an axis in the yz plane. Figure 5B In the second embodiment shown, the X-ray CT apparatus 1 includes a function of correcting the positioning of the subject.
[0078] Figure 4C The displacement of the head includes the radiated EL based on the position of the subject. In the figure, the subject is radiated above the EL. The processing circuit 51 can determine the EL based on the position of the subject and the mechanical shape. Figure 4C At the offset shown, the subject's right lens may be exposed to radiation.
[0079] Steps 70 to 74 and Figure 5AThe method is the same as that of step 74. After step 74, the head tilt is confirmed in step 77A. In one embodiment, the angle of line L3 in the xz plane determined by DCNN62 relative to the x-axis is compared with the expected range by processing circuit 51. In the case where the angle of line L3 is greater than the determined degree relative to the y-axis, in step 77B, a warning or alarm is provided to the operator on the display 55. In another embodiment, processing circuit 51 is capable of determining the distance between P1 or P2 and EL that is orthogonal to EL. In the case where the distance is above a threshold value such as several mm, a warning or alarm is provided again. Figure 4D The figure shows the displacement of the head of the subject when it moves relative to the desired chin-tightening position, including the radiated line EL based on the position of the subject. In the figure, above the EL, the subject is radiated. The processing circuit 51 can determine the EL based on the position of the subject and the shape of the machine. Figure 4D By determining the angle between L4 and L5, or by determining the distance between P2 and EL that is perpendicular to EL, it is possible to use the Figure 4C Give a warning or alarm in the same way as above.
[0080] In step 77C, the operator can communicate with the subject using the communication device 46 (speaker / microphone) to reposition the subject. If the operator wishes, in step 77D, another 3D positioning image can be photographed to confirm that the subject is in the appropriate position. This process can be repeated as needed. The subject can be moved to the appropriate z-direction position in step 75 as needed, and a diagnostic scan is performed in step 76 in the same manner as in the first embodiment.
[0081] Figure 5C and Figure 5D 3 and 4 embodiments of the method according to the present disclosure are shown. In these embodiments, the gantry is fixed at the L1 position without being tilted. Figure 5C In the case of the method, in step 70, the subject is positioned by typically tightening the jaw in order to arrange the head at a desired angle so that the lens is not irradiated. The above steps 71 and 72 are performed. The processing circuit 51 uses the biomarker to determine the appropriate z-direction position in step 75 as needed, and scans the subject in step 76. Figure 5D The method shown is similar to the third embodiment except that steps 73 and 74 are omitted. Figure 5B The method is the same.
[0082] Figure 6An example of DCNN 62 is shown. Other neural network structures may also be used. A multilayer network is a network in which a plurality of layers are arranged in such a way that only adjacent layers are connected, and information is transmitted in a single direction from the input layer side to the output layer side. Figure 6 In the example, DCNN 62 has an input layer (I=1), inner layers I=2 to I=L-1, and an output layer I=L. Other configurations may also be used. The positioning image data is input to the x 1 ~x N The data is processed in the inner layers and the output is generated in the output layer I=L. Each layer is given appropriate weights and biases for detecting biomarkers.
[0083] pass Figure 7 The training process shown above determines the weights and biases. The DCNN training process is shown in the figure above and is designed to map the training input X to the desired target Y. The training process can be described as solving the following optimization problem.
[0084]
[0085] Here, Θ is the set of parameters of the neural network to be optimized, N is the total number of training instances during training, f is the neural network to be optimized, and x i is the i-th element of the training input, y i is the i-th element of the training target. By solving this optimization problem, the optimal network parameter Θ can be obtained * , so that the difference Δ between the network output and the target biomarker Y (for example, it can also be MAE (mean absolute error), MSE (mean square error), GDL (Generalized Dice Loss), Dice loss, cross entropy, etc.) is minimized.
[0086] exist Figure 7In the present invention, a 3D positioning image 80 is input to a DCNN 81, and a biomarker 82 is generated as an output of the DCNN 81. A target biomarker 83 (supraorbital ridge point and external auditory canal center, etc.) is prepared and input to a loss function 84 to find the difference between the generated target biomarker and the target biomarker of the true value. The optimizer 85 adjusts the weights and biases of the DCNN 81 in order to adjust θ so that the difference between the output biomarker and the target biomarker is minimized. The present inventors collected a training set consisting of 48 brain CT scans with manual labels in order to find the coordinates of three landmarks defined as the posterior cranial point (OO) of the occipital skull and the approximate center of the supraorbital ridge of the left and right eyes (LE and RE). A test set consisting of 12 brain CT and 8 real 3D positioning image data sets also had the coordinates manually labeled as the true value. The images were pre-processed by downscaling (x4) and bone thresholding (HU300-1900 was set to 1 and other feature quantities were set to 0). Fig. 9 The results are shown. The left, middle and right views represent the biomarker points determined by DCNN for LE, OO and RE, respectively. The horizontal line at the bottom of each view is the inclination angle (θ) between the mechanical axis and the surface generated as a result, shown for reference. The dotted line represents the inferred inclined surface, and the solid line represents the labeled inclined surface. The following Table 1 shows the results of the pixel data corresponding to 12 brain CT data sets with dFOV = 240mm or 320mm and a spacing of 0.5mm. The inferred biomarker points and inclined surfaces are highly consistent with the points and surfaces with manually labeled surfaces with high precision.
[0087] (Table 1)
[0088] Measurement benchmark Difference (mean ± standard deviation) RE / LE 1.31±0.73 pixels OO 2.04±0.99 pixels OO (Sagittal plane) 0.90±0.62 pixels Tilt Angle 1.27±1.06 degrees
[0089] In addition, the inventors also conducted tests using 3D positioning image data. Eight data sets were used with dFOV = 500 mm and 1 mm pitch. The image data was downsampled (2x) to a cropped image of 256×256 pixels. Fig.10 The results are shown for LE, OO, and RE. The dotted line represents the estimated inclined surface, and the solid line represents the labeled inclined surface. Here, the predicted inclined surface is also highly consistent with the surface of the manually labeled points. The following Table 2 shows the results of the pixel data.
[0090] (Table 2)
[0091] Measurement benchmark Difference (mean + standard deviation) RE / LE 2.12+0.41 pixels OO 3.54+0.91 pixels OO (Sagittal plane) 1.64+1.15 pixels Tilt Angle 1.47+1.26 degrees
[0092] According to Table 1 and Table 2 and Fig. 9 and Fig.10The results show that the present device can accurately infer the inclined plane (and the inclined angle) for safely performing a diagnostic head CT scan in order to avoid additional irradiation to the lens.
[0093] Figure 9-10 The images shown were created using Fig.11 The images obtained for the phantom shown.
[0094] (Variation Example)
[0095] In a variation of the embodiment, two modes of 3D scout scanning are used. In the first mode, the scout scanning is performed at a dose of a normal 3D scout image without detecting the OMBL (or other lines). The mode is switched to the second mode at a higher dose of the scout image, and the OMBL is detected. As a second variation, in a specific view acquisition in the 3D scout image operation, the dose is increased at a specific rotational position in the 3D scan in order to detect the OMBL. Figure 8 Indicates this situation.
[0096] The method and system disclosed herein have the advantages of an automatic gantry tilt workflow, which speeds up the overall throughput of subjects compared to time-consuming manual tilting. It is believed that the method and system ensure that a gantry tilt protocol that protects the lens can be applied and reduce the risk of human error in inferring the tilt angle. Compared to previous gantry tilt methods, the method and system proposed in the present disclosure does not change the current workflow because it seamlessly performs 3D positioning scans and diagnostic scans, and there is no need for additional CT scans for positioning and calculating the gantry tilt angle.
[0097] According to at least one of the embodiments described above, it is possible to reduce radiation exposure to organs with high radiation sensitivity.
[0098] Several embodiments are described above, but these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other ways, and various omissions, substitutions, and changes can be made without departing from the scope of the invention. These embodiments and their variations are included in the scope or spirit of the invention, and are included in the invention described in the claims and their equivalents.
Claims
1. A control method, wherein: The steps include: Controlling a CT gantry that supports an X-ray source and an X-ray detector so as to be rotatable, performing a positioning CT scan of a three-dimensional region of a head of a subject, thereby generating image data; inputting cross-sectional image data of the three-dimensional region generated from the image data into a learned model to detect anatomical landmarks for determining an inclined surface for avoiding irradiation of a lens of the subject in a diagnostic CT scan; determining a tilt angle of the CT gantry based on the anatomical landmark and the tilted surface; as well as The CT gantry is controlled to be tilted at the determined tilt angle, and the diagnostic CT scan of the subject is performed.
2. The control method according to claim 1, wherein: The positioning CT scan and the diagnostic CT scan are continuously performed in one scanning operation.
3. The control method according to claim 1, wherein: The scout CT scan is performed by increasing the X-ray dose of a specific view having a specific rotational position in order to determine the inclined plane.
4. The control method according to claim 1, wherein: The step of determining the inclined surface includes the step of determining the inclined surface based on the auditory canthus line of the subject.
5. The control method according to claim 1, wherein: The step of determining the inclined plane includes the step of determining the inclined plane based on the supraorbital crest and the posterior cranial point of the subject.
6. The control method according to claim 1, wherein: The following steps are also included: The position of the head of the subject is determined based on the anatomical landmarks.
7. The control method according to claim 1, wherein: The following steps are also included: determining a position of the subject on a movable table associated with the CT gantry relative to a scanning plane of the tilted CT gantry; and The subject is moved so that a desired scanning plane passing through the head of the subject coincides with the scanning plane of the tilted CT gantry.
8. The control method according to claim 1, wherein: The following steps are also included: Determine the line through the inclined plane of the supraorbital ridge of the left and right eyes; determining whether the angle of the line is within a set angle relative to one of the mechanical axes of the CT gantry; and A signal is generated indicating that the angle of the line is not within the set angle.
9. A control method, wherein: The steps include: Controlling a CT gantry that supports an X-ray source and an X-ray detector so as to be rotatable, performing a positioning CT scan of a three-dimensional region of a subject, and generating image data; inputting the image data into a learning completion model to determine anatomical landmarks within the image data; determining a tilt angle of the CT gantry based on the anatomical landmarks; as well as The CT gantry is controlled to be tilted at the tilt angle, and a diagnostic CT scan of the subject is performed.
10. The control method according to claim 9, wherein: The positioning CT scan and the diagnostic CT scan are continuously performed in one scanning operation.
11. The control method according to claim 9, wherein: The steps include: generating an anatomical line based on the anatomical landmark; Determining the tilt angle based on the anatomical line; as well as The CT gantry is tilted based on the tilt angle.
12. The control method according to claim 11, wherein: The anatomical line is the canthal line.
13. The control method according to claim 9, wherein: The positioning CT scan includes: a first scan configured to perform a scan plan accompanied by detection of the auditory canthus line, or a second scan configured to perform a scan plan not accompanied by detection of the auditory canthus line, The first scan is performed with a higher X-ray dose than the second scan.
14. An X-ray CT device, wherein: have: A CT gantry supports an X-ray source and an X-ray detector so as to be rotatable; a generating unit that performs a positioning CT scan of a three-dimensional region of the subject to generate image data; a detection unit, inputting the image data into the learned model to detect anatomical landmarks; a determination unit, which determines a tilt angle of the CT gantry based on the anatomical landmark; as well as The control unit controls the CT gantry to be tilted at the tilt angle to perform a diagnostic CT scan of the subject.
15. The X-ray CT apparatus according to claim 14, wherein: The generator executes, as the positioning CT scan, a first scan configured with a scan plan accompanied by detection of the auditory canthus line, or a second scan configured with a scan plan not accompanied by detection of the auditory canthus line. The first scan is performed with a higher X-ray dose than the second scan.
16. The X-ray CT apparatus according to claim 14, wherein: The positioning CT scan and the diagnostic CT scan are continuously performed in one scanning operation.
17. The X-ray CT apparatus according to claim 14, wherein: The determination unit generates an anatomical line based on the anatomical landmark, and determines the inclination angle based on the generated anatomical line. The control unit tilts the CT gantry based on the tilt angle.
18. The X-ray CT apparatus according to claim 14, wherein: The determination unit determines the tilt angle based on the canthal line of the subject.
19. The X-ray CT apparatus according to claim 14, wherein: The determination unit determines the tilt angle based on the supraorbital crest and the posterior cranial point of the subject.
