Method and device for detecting closed loop in magnetic resonance imaging and magnetic resonance imaging system

Abstract

The embodiment of the invention discloses a method and device for detecting a closed loop in magnetic resonance imaging and a magnetic resonance imaging system. The method comprises the following steps: acquiring three-dimensional coordinates of each key feature point of an MRI examination object; the three-dimensional coordinates of all the key feature points are matched with a three-dimensional human body standard model, and a three-dimensional virtual human body model of the examination object is obtained; extracting a three-dimensional surface area of each pre-specified body part on the three-dimensional virtual human body model; calculating the distance between the three-dimensional surface areas of the two body parts in each body part pair according to predefined body parts which are prone to skin contact; moreover, if the distance between the three-dimensional surface areas of the two body parts in any body part pair is smaller than a preset first threshold value, it is determined that the examination object has the closed loop risk at present. According to the embodiment of the invention, the closed loop of the examination object in the MRI can be detected.

Description

Technical Field

[0001] The present invention relates to the technical field of MRI (Magnetic Resonance Imaging), and in particular to a method and device for detecting a closed loop in MRI and an MRI system. Background Art

[0002] In clinical scanning applications of MRI systems, due to the presence of gradient fields and high-frequency radio frequencies, if a closed loop exists in a patient's body during scanning, a closed loop current will be generated, thereby generating heat. When the closed loop current is too large or the scanning duration is long, the heat generated is enough to make the patient feel hot or even burn the patient, seriously threatening the patient's safety.

[0003] Burns during MRI scanning are very serious, with a large number of first and second degree burns reported. Statistical studies have shown that MRI burns are mainly caused by contact between body parts such as overlapping skin, crossed legs or overlapping hands.

[0004] The closed loop during MRI scanning is mainly caused by the following three reasons:

[0005] 1. Cross-positioning of the patient’s body parts

[0006] When the patient lies on the examination bed and prepares for coil positioning, the patient's nervousness will unconsciously bring the body parts together. If the scanning technician does not find the crossed parts of the patient and correct them during positioning, there is a risk of a closed loop during scanning. Therefore, burn cases caused by the intersection of body parts are the most common. According to statistical surveys, about 77% of the burn cases received were caused by skin contact between the patient's body parts. Figure 1 An example diagram of a closed circuit resulting from skin contact between patient body parts, such as Figure 1 As shown, skin contact between the two hands 11 and 12, between the two feet 13, between the hands and the torso 14 and 15, between the two inner thighs 16, etc. will all result in a closed circuit.

[0007] 2. Direct contact between the patient's skin and other objects such as scanning equipment or coils

[0008] When the skin of a patient's body comes into direct contact with the outer wall of the scanning device or other objects such as the coil, a closed circuit may occur, which may lead to burns.

[0009] 3. Metal implants in the patient’s body, such as pacemakers, prostheses, stents, braces, etc.

[0010] Because before performing an MRI scan, the scanning technician will ask the patient if he has any metal implants in his body, so the number of burns caused by this reason is relatively small.

[0011] Currently, in the preparation stage before performing an MRI scan, the scanning technician will perform manual checks to prevent the occurrence of closed loops. Since the patient may move his position or posture during the positioning process, it is difficult to completely eliminate the possibility of closed loops. Summary of the invention

[0012] In view of this, the embodiments of the present invention propose, on the one hand, a method and apparatus for detecting a closed loop in MRI to detect a closed loop of an object under examination in MRI; and on the other hand, propose an MRI system to detect a closed loop of an object under examination in MRI.

[0013] A method for detecting a closed loop in a magnetic resonance imaging (MRI), the method comprising:

[0014] Obtain the three-dimensional coordinates of each key feature point of the MRI examination object;

[0015] The three-dimensional coordinates of each key feature point are matched with a three-dimensional human body standard model to obtain a three-dimensional virtual human body model of the inspection object;

[0016] Extracting three-dimensional surface areas of pre-specified body parts on the three-dimensional virtual human body model;

[0017] According to predefined pairs of body parts that are prone to skin contact, the distance between the three-dimensional surface areas of two body parts in each pair of body parts is calculated;

[0018] Furthermore, if the distance between the three-dimensional surface areas of two body parts in any pair of body parts is smaller than a preset first threshold, it is determined that the inspection object currently has a closed loop risk.

[0019] The step of obtaining the three-dimensional coordinates of each key feature point of the MRI examination object includes:

[0020] Acquire an RGB image and a depth image of the inspection object;

[0021] Detecting each key feature point in the RGB image of the inspection object and recording the two-dimensional coordinates of each key feature point;

[0022] The three-dimensional coordinates of each key feature point are acquired according to the two-dimensional coordinates of each key feature point and the depth image of the inspection object.

[0023] The detecting each key feature point in the RGB image of the inspection object comprises:

[0024] The RGB image of the inspection object is input into a pre-trained key feature point detection deep learning network model to obtain the two-dimensional coordinates of each key feature point of the inspection object.

[0025] After acquiring the three-dimensional coordinates of each key feature point of the MRI inspection object and before adapting the three-dimensional coordinates of each key feature point to the three-dimensional human body standard model, the method further includes:

[0026] The three-dimensional coordinates of each key feature point are converted into a predefined coordinate system.

[0027] After obtaining the three-dimensional coordinates of each key feature point of the MRI examination object, the method further includes:

[0028] According to the three-dimensional coordinates of each key feature point, it is detected whether the inspection object currently has a cross-position. If so, it is determined that the inspection object currently has a closed loop risk.

[0029] The detecting, according to the three-dimensional coordinates of each key feature point, whether the inspection object currently has a cross-position includes:

[0030] According to predefined key feature point pairs that are prone to cross-positioning and the three-dimensional coordinates of the key feature points, the distance between two key feature points in each key feature point pair is calculated, and if the distance between two key feature points in any key feature point pair is less than a preset second threshold, it is determined that the inspection object currently has a cross-positioning; or,

[0031] The three-dimensional coordinates of each key feature point of the inspection object are input into a pre-trained cross-position detection deep learning network model to detect whether the inspection object is currently in a cross-position.

[0032] After the distance between the three-dimensional surface areas of two body parts in any pair of body parts is less than a preset first threshold and before determining that the inspection object currently has a closed loop risk, the method further includes:

[0033] It is determined whether it is detected that the inspection object currently has a cross-position. If so, it is determined that the inspection object currently has a closed loop risk.

[0034] After determining that the inspection object currently has a closed loop risk, the method further includes:

[0035] Send a closed loop alarm to the inspected object; or / and,

[0036] A closed-loop alarm is issued to the inspected object, and correct positioning guidance is given.

[0037] A device for detecting a closed loop in a magnetic resonance imaging (MRI), the device comprising:

[0038] A key feature point acquisition module is used to obtain the three-dimensional coordinates of each key feature point of the MRI examination object;

[0039] A three-dimensional surface area extraction module is used to adapt the three-dimensional coordinates of each key feature point to a three-dimensional human body standard model to obtain a three-dimensional virtual human body model of the inspection object; and extract the three-dimensional surface area of ​​each pre-specified body part on the three-dimensional virtual human body model;

[0040] The skin contact detection module is used to calculate the distance between the three-dimensional surface areas of two body parts in each pair of body parts based on pre-defined pairs of body parts that are prone to skin contact; and if the distance between the three-dimensional surface areas of two body parts in any pair of body parts is less than a preset first threshold, it is determined that the inspection object currently has a closed loop risk.

[0041] The key feature point acquisition module acquires the three-dimensional coordinates of each key feature point of the MRI examination object, including:

[0042] Acquire an RGB image and a depth image of the inspection object;

[0043] Detecting each key feature point in the RGB image of the inspection object and recording the two-dimensional coordinates of each key feature point;

[0044] The three-dimensional coordinates of each key feature point are acquired according to the two-dimensional coordinates of each key feature point and the depth image of the inspection object.

[0045] The key feature point acquisition module detects each key feature point in the RGB image of the inspection object, including:

[0046] The RGB image of the inspection object is input into a pre-trained key feature point detection deep learning network model to obtain the two-dimensional coordinates of each key feature point of the inspection object.

[0047] After the key feature point acquisition module acquires the three-dimensional coordinates of each key feature point of the MRI examination object, the method further includes:

[0048] The three-dimensional coordinates of each key feature point are converted into a predefined coordinate system.

[0049] The device further includes: a cross-position detection module, which is used to detect whether the inspection object is currently cross-positioned according to the three-dimensional coordinates of the key feature points, and if so, determine that the inspection object currently has a closed loop risk.

[0050] The cross-position detection module detects whether the inspection object is currently cross-positioned according to the three-dimensional coordinates of each key feature point, including:

[0051] According to predefined key feature point pairs that are prone to cross-positioning and the three-dimensional coordinates of the key feature points, the distance between two key feature points in each key feature point pair is calculated, and if the distance between two key feature points in any key feature point pair is less than a preset second threshold, it is determined that the inspection object currently has a cross-positioning; or,

[0052] The three-dimensional coordinates of each key feature point of the inspection object are input into a pre-trained cross-position detection deep learning network model to detect whether the inspection object is currently in a cross-position.

[0053] After determining that the distance between the three-dimensional surface areas of two body parts in any pair of body parts is less than a preset first threshold and before determining that the inspection object currently has a closed loop risk, the skin contact detection module further includes:

[0054] It is determined whether the cross-position detection module detects that the inspection object is currently in a cross-position. If so, it is determined that the inspection object currently has a closed loop risk.

[0055] After the skin contact detection module determines that the inspection object currently has a closed loop risk, the method further includes:

[0056] Send a closed loop alarm to the inspected object; or / and,

[0057] A closed-loop alarm is issued to the inspected object, and correct positioning guidance is given.

[0058] A magnetic resonance imaging (MRI) system comprises any one of the above-described devices for detecting a closed loop in a magnetic resonance imaging (MRI).

[0059] In an embodiment of the present invention, the three-dimensional coordinates of each key feature point of the inspection object of MRI are obtained, the three-dimensional coordinates of each key feature point are adapted to the three-dimensional human body standard model to obtain a three-dimensional virtual human body model of the inspection object, and then the three-dimensional surface areas of each pre-specified body part are extracted on the three-dimensional virtual human body model, and the distance between the three-dimensional surface areas of two body parts in each body part pair is calculated according to pre-defined pairs of body parts that are prone to skin contact. If the distance between the three-dimensional surface areas of two body parts in any body part pair is less than a preset first threshold, it is determined that the inspection object currently has a closed loop risk, thereby being able to detect the closed loop of the inspection object in MRI. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings so that the above and other features and advantages of the present invention will be more apparent to those skilled in the art. In the accompanying drawings:

[0061] Figure 1 An example diagram of a closed circuit resulting from skin contact between patient body parts;

[0062] Figure 2 A flow chart of a method for detecting a closed loop in MRI provided by an embodiment of the present invention;

[0063] Figure 3 This is a schematic diagram of using a 3D camera to capture an RGB image and a depth image of an inspection object in an application example of the present invention;

[0064] Figure 4 An example diagram of detecting key feature points in an RGB image of an inspection object provided in an application example of the present invention;

[0065] Figure 5 An example diagram of an application example of the present invention is provided in which the three-dimensional coordinates of each key feature point of the inspection object are adapted to the three-dimensional human body standard model to obtain a three-dimensional virtual human body model of the inspection object;

[0066] Figure 6 This is an example diagram of extracting a three-dimensional surface area of ​​a right foot on a three-dimensional virtual human body model of an inspection object in an application example of the present invention;

[0067] Figure 7 A flow chart of a method for detecting a closed loop in MRI provided by another embodiment of the present invention;

[0068] Figure 8 A schematic diagram of the structure of a device for detecting a closed loop in MRI provided by an embodiment of the present invention.

[0069] The reference numerals are as follows:

[0070]

[0071] DETAILED DESCRIPTION

[0072] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described in detail with reference to the following embodiments.

[0073] Figure 2 A flowchart of a method for detecting a closed loop in MRI provided by an embodiment of the present invention, wherein the specific steps are as follows:

[0074] Step 201: Acquire the three-dimensional coordinates of each key feature point of the MRI examination object.

[0075] In an optional embodiment, this step 201 specifically includes: obtaining an RGB image and a depth image of the inspection object; detecting each key feature point in the RGB image of the inspection object and recording the two-dimensional coordinates of each key feature point; and obtaining the three-dimensional coordinates of each key feature point based on the two-dimensional coordinates of each key feature point and the depth image of the inspection object.

[0076] Figure 3 The schematic diagram of using a 3D camera to collect RGB images and depth images of an inspection object in an application example of the present invention is shown in FIG. 31 is a 3D camera installed directly above an inspection bed 32 of an MRI system, with a field of view covering the inspection bed, 33 is an example of an inspection object, and 34 is an example of a depth image of an inspection object.

[0077] In practical applications, a 3D camera or a depth camera + a color camera may be used to collect the RGB image and the depth image of the inspection object, which is not limited in this embodiment.

[0078] In an optional embodiment, each key feature point is detected in the RGB image of the inspection object, including: inputting the RGB image of the inspection object into a pre-trained key feature point detection deep learning network model to obtain the two-dimensional coordinates of each key feature point of the inspection object.

[0079] The key feature points may be predefined, including at least the key joints of the human body. The key feature points may represent the feature points of the following parts: nose, left eye, right eye, left ear, right ear, left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip, right hip, left knee, right knee, left ankle, right ankle, etc.

[0080] Figure 4 This is an example diagram of detecting key feature points in an RGB image of an inspection object provided as an application example of the present invention, wherein each small dot represents a key feature point.

[0081] In practical applications, after the three-dimensional coordinates of each key feature point of the MRI inspection object are acquired in step 201, the three-dimensional coordinates of each key feature point of the inspection object may be further converted into a predefined coordinate system.

[0082] The examination subject refers to the human body undergoing MRI.

[0083] Step 202: The three-dimensional coordinates of each key feature point of the inspection object are matched with the three-dimensional human body standard model to obtain a three-dimensional virtual human body model of the inspection object.

[0084] Specifically, for example, the three-dimensional coordinates of each key feature point of the inspection object are compared with the three-dimensional coordinates of each key feature point in the three-dimensional human standard model to obtain the position and posture of the inspection object, and then determine the three-dimensional virtual human model of the inspection object.

[0085] Figure 5 An example diagram of an application example of the present invention is provided in which the three-dimensional coordinates of each key feature point of the inspection object are adapted to the three-dimensional human body standard model to obtain an example of a three-dimensional virtual human body model of the inspection object. Among them, 51 is an example of the key feature point of the inspection object, and 52 is an example of the adapted three-dimensional virtual human body model of the inspection object.

[0086] Step 203: extracting three-dimensional surface areas of pre-specified body parts on the three-dimensional virtual human body model of the inspection object.

[0087] The pre-designated body parts refer to the relevant body parts that are prone to skin contact, such as: head, palms, arms, chest, abdomen, thighs, calves, feet, etc. For example: contact between the left and right palms, contact between any palm and the head, contact between any palm and the chest, contact between any palm and the abdomen, contact between the inner thighs, contact between the calves, contact between the feet, etc. will all result in a closed loop.

[0088] After obtaining the three-dimensional virtual human body model of the inspection object, the position and range of the surface area of ​​each body part on the three-dimensional virtual human body model can be known, that is, the three-dimensional coordinates of each point on the surface area of ​​each body part can be known.

[0089] Figure 6 This is an example diagram of extracting the three-dimensional surface area of ​​the right foot on the three-dimensional virtual human body model of the inspection object in an application example of the present invention, wherein the black area 61 is the three-dimensional surface area of ​​the right foot.

[0090] Step 204: Calculate the distance between the three-dimensional surface areas of two body parts in each pair of body parts based on predefined pairs of body parts that are prone to skin contact; and if the distance between the three-dimensional surface areas of two body parts in any pair of body parts is less than a preset first threshold, determine that the inspection object currently has a closed loop risk.

[0091] In practical applications, the distance between the three-dimensional surface areas of the two body parts in each body part pair can be represented by the distance between the two closest points on the three-dimensional surface areas of the two body parts. For example, for the two body parts A and B in a body part pair, the distance between each point in the three-dimensional surface area of ​​body part A and each point in the three-dimensional surface area of ​​body part B is calculated respectively, where the minimum distance is the distance between the three-dimensional surface areas of body parts A and B.

[0092] When it is determined that the subject has a closed loop risk, the body part with the closed loop risk can be combined with whether the subject enters the hole head first or feet first to determine the name of the body part with the closed loop risk, and the subject or / and the scanning technician can be prompted with the name of the body part with the closed loop risk. For example, for parts such as hands and feet, it is necessary to distinguish between left and right, so at this time, it is necessary to combine whether the subject enters the hole head first or feet first to determine left or right.

[0093] In the above embodiment, the three-dimensional coordinates of each key feature point of the MRI examination object are obtained, the three-dimensional coordinates of each key feature point are adapted to the three-dimensional human body standard model to obtain a three-dimensional virtual human body model of the examination object, and then the three-dimensional surface areas of each pre-specified body part are extracted on the three-dimensional virtual human body model, and the distance between the three-dimensional surface areas of the two body parts in each body part pair is calculated according to the pre-defined pairs of body parts that are prone to skin contact. If the distance between the three-dimensional surface areas of the two body parts in any body part pair is less than a preset first threshold, it is determined that the examination object currently has a closed loop risk, thereby being able to detect the closed loop of the examination object in the MRI.

[0094] In the above embodiment, whether the inspection object currently has a closed loop risk is detected by detecting the distance between the three-dimensional surface areas of the body parts that are prone to skin contact. In actual applications, whether the inspection object currently has a closed loop risk can also be detected by detecting the distance between the key feature points that are prone to cross-positioning.

[0095] In an optional embodiment, in step 201, after obtaining the three-dimensional coordinates of each key feature point of the MRI examination object, it further includes: based on the three-dimensional coordinates of each key feature point of the examination object, detecting whether the examination object currently has a cross-position, and if so, determining that the examination object currently has a closed loop risk.

[0096] In an optional embodiment, based on the three-dimensional coordinates of each key feature point of the inspection object, it is detected whether the inspection object currently has a cross-position, including: based on pre-defined key feature point pairs that are prone to cross-position and the three-dimensional coordinates of each key feature point, calculating the distance between two key feature points in each key feature point pair; if the distance between the two key feature points in any key feature point pair is less than a preset second threshold, it is determined that the inspection object currently has a cross-position; or, inputting the three-dimensional coordinates of each key feature point of the inspection object into a pre-trained cross-position detection deep learning network model to detect whether the inspection object currently has a cross-position.

[0097] For example, when the distance between the feature points of the left and right wrists is less than a preset second threshold, it is considered that the left and right palms are crossed and the inspection object currently has a closed loop risk.

[0098] In actual applications, when the following schemes are adopted simultaneously: scheme 1, detecting whether the inspection object currently has a closed loop risk by detecting the distance between the three-dimensional surface areas of body parts prone to skin contact, and scheme 2, detecting whether the inspection object currently has a closed loop risk by detecting the distance between key feature points prone to cross-positioning, an alarm may be issued to the inspection object when any scheme detects that the inspection object currently has a closed loop risk, or a closed loop alarm may be issued to the inspection object only when both schemes detect that the inspection object currently has a closed loop risk.

[0099] In practical applications, by combining the two closed-loop detection schemes, the closed-loop risks of the inspection object can be detected more comprehensively and effectively.

[0100] When the two closed loop detection schemes are combined, that is, in step 204, if the distance between the three-dimensional surface areas of two body parts in any pair of body parts is less than the preset first threshold and before determining that the inspection object currently has a closed loop risk, it further includes: determining whether it is detected that the inspection object currently has a cross-position, and if so, determining that the inspection object currently has a closed loop risk.

[0101] In an optional embodiment, after determining that the inspection object currently has a closed loop risk, it further includes: issuing a closed loop alarm to the inspection object; or / and, issuing a closed loop alarm to the inspection object and providing correct positioning guidance.

[0102] There is no limitation on the form and specific content of issuing a closed-loop alarm to the inspection object and providing correct positioning guidance, which may be in the form of text and / or voice and / or image.

[0103] It should be noted that the embodiments of the present invention can be applied not only in the scanning preparation stage of MRI to guide the correct positioning of the inspection object, but also in the formal scanning stage of MRI to timely detect and remind when the inspection object has a closed loop risk, thereby realizing the detection of whether the inspection object has a closed loop risk throughout the entire MRI, greatly reducing labor costs, speeding up MRI scanning, and improving MRI scanning safety.

[0104] In addition, in the embodiment of the present invention, when the inspection object is positioned, a positioning image of the inspection object at this time can be collected so that if human body damage caused by the closed loop occurs later, the cause can be found according to the inspection object positioning image.

[0105] Figure 7 A flowchart of a method for detecting a closed loop in MRI provided by another embodiment of the present invention, wherein the specific steps are as follows:

[0106] Step 701: Acquire an RGB image and a depth image of an MRI examination object.

[0107] Step 702: Input the RGB image of the inspection object into a pre-trained key feature point detection deep learning network model to obtain the two-dimensional coordinates of each key feature point of the inspection object.

[0108] Step 703: Acquire the three-dimensional coordinates of each key feature point of the inspection object according to the two-dimensional coordinates of each key feature point of the inspection object and the depth image of the inspection object.

[0109] Step 704: convert the three-dimensional coordinates of each key feature point of the inspection object into a predefined coordinate system.

[0110] Step 705: Input the converted three-dimensional coordinates of each key feature point of the inspection object into a pre-trained cross-position detection deep learning network model, and the model outputs an indication of whether the inspection object currently has a cross-position.

[0111] Step 706: adapting the converted three-dimensional coordinates of each key feature point of the inspection object to the three-dimensional human body standard model to obtain a three-dimensional virtual human body model of the inspection object.

[0112] Step 707: extracting the three-dimensional surface area of ​​each pre-specified body part on the three-dimensional virtual human body model of the inspection object.

[0113] Step 708: Calculate the distance between the three-dimensional surface areas of the two body parts in each pair of body parts based on predefined pairs of body parts that are prone to skin contact; and if the distance between the three-dimensional surface areas of the two body parts in any pair of body parts is less than a preset first threshold, determine that the two body parts are in skin contact.

[0114] Step 709: Determine whether the following conditions are met: in step 705, the model outputs an indication that the inspection object is currently in a cross-position, and in step 708, at least one pair of body parts is in skin contact. If so, execute step 710; otherwise, execute step 711.

[0115] Step 710: Determine that the inspection object currently has a closed loop risk, issue a closed loop alarm to the inspection object, and provide correct positioning guidance, and this process ends.

[0116] Step 711: Determine that the inspection object currently does not have a closed loop risk, and this process ends.

[0117] In practical applications, step 705 and steps 706 to 708 may be performed in parallel.

[0118] In addition, steps 701-711 are performed in real time during the MRI process (including the scanning preparation stage and the formal scanning stage) to achieve the purpose of real-time detection of whether there is a closed loop in the inspection object.

[0119] The beneficial technical effects of the embodiments of the present invention are as follows:

[0120] 1. It can detect the closed loop risk of the object in MRI (including the preparation and formal scanning stages), reduce the labor cost of MRI, shorten the time of MRI scanning preparation stage, improve the MRI scanning speed, and detect the closed loop risk before the formal scanning, thus improving the safety of MRI scanning;

[0121] Second, the 3D coordinates of the key feature points of the inspection object are obtained by combining the RGB image and the depth image, so that even if the inspection object is covered by a coil or blanket, the closed loop detection of the inspection object can be accurately performed;

[0122] Third, by combining scheme 1, detecting whether the inspection object currently has a closed loop risk by detecting the distance between the three-dimensional surface areas of body parts prone to skin contact, and scheme 2, detecting whether the inspection object currently has a closed loop risk by detecting the distance between key feature points prone to cross-positioning, the closed loop risk of the inspection object can be detected more comprehensively and effectively.

[0123] Figure 8 A schematic diagram of the structure of a device 80 for detecting a closed loop in an MRI according to an embodiment of the present invention. The device 80 mainly includes:

[0124] The key feature point acquisition module 81 is used to acquire the three-dimensional coordinates of each key feature point of the MRI examination object.

[0125] The three-dimensional surface area extraction module 82 is used to adapt the three-dimensional coordinates of each key feature point of the inspection object to the three-dimensional human body standard model to obtain a three-dimensional virtual human body model of the inspection object; and extract the three-dimensional surface areas of each pre-specified body part on the three-dimensional virtual human body model of the inspection object.

[0126] The skin contact detection module 83 is used to calculate the distance between the three-dimensional surface areas of two body parts in each pair of body parts based on pre-defined pairs of body parts that are prone to skin contact; and if the distance between the three-dimensional surface areas of two body parts in any pair of body parts is less than a preset first threshold, it is determined that the inspection object currently has a closed loop risk.

[0127] In an optional embodiment, the key feature point acquisition module 81 acquires the three-dimensional coordinates of each key feature point of the MRI examination object, including: acquiring an RGB image and a depth image of the examination object; detecting each key feature point in the RGB image of the examination object and recording the two-dimensional coordinates of each key feature point; acquiring the three-dimensional coordinates of each key feature point based on the two-dimensional coordinates of each key feature point and the depth image of the examination object.

[0128] In an optional embodiment, the key feature point acquisition module 81 detects each key feature point in the RGB image of the inspection object, including: inputting the RGB image of the inspection object into a pre-trained key feature point detection deep learning network model to obtain the two-dimensional coordinates of each key feature point of the inspection object.

[0129] In an optional embodiment, after the key feature point acquisition module 81 acquires the three-dimensional coordinates of each key feature point of the MRI examination object, it further includes: converting the three-dimensional coordinates of each key feature point into a predefined coordinate system.

[0130] In an optional embodiment, the device 80 further includes: a cross-position detection module 84, which is used to detect whether the inspection object currently has a cross-position based on the three-dimensional coordinates of each key feature point. If so, it is determined that the inspection object currently has a closed loop risk.

[0131] In an optional embodiment, the cross-position detection module 84 detects whether the inspection object currently has a cross-position based on the three-dimensional coordinates of each key feature point, including: calculating the distance between two key feature points in each key feature point pair based on pre-defined key feature point pairs that are prone to cross-position and the three-dimensional coordinates of each key feature point of the body part; if the distance between the two key feature points in any key feature point pair is less than a preset second threshold, it is determined that the inspection object currently has a cross-position; or, inputting the three-dimensional coordinates of each key feature point of the inspection object into a pre-trained cross-position detection deep learning network model to detect whether the inspection object currently has a cross-position.

[0132] In an optional embodiment, after the skin contact detection module 83 determines that the distance between the three-dimensional surface areas of two body parts in any pair of body parts is less than a preset first threshold and before determining that the inspection object currently has a closed loop risk, the skin contact detection module 83 further includes: determining whether the cross-position detection module (84) detects that the inspection object currently has a cross-position, and if so, determining that the inspection object currently has a closed loop risk.

[0133] In an optional embodiment, after the skin contact detection module 83 determines that the inspection object currently has a closed loop risk, it further includes: issuing a closed loop alarm to the inspection object; or / and, issuing a closed loop alarm to the inspection object and providing correct positioning guidance.

[0134] An embodiment of the present invention further provides an MRI system, which includes the device 80 for detecting a closed loop in MRI as described in any of the above embodiments.

[0135] Those skilled in the art will appreciate that the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations or combinations are not explicitly described in the present application. In particular, without departing from the spirit and teachings of the present application, the features described in the various embodiments and / or claims of the present application may be combined and / or combined in a variety of ways, and all of these combinations and / or combinations fall within the scope disclosed in the present application.

[0136] Specific embodiments are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core ideas of the present application, and is not used to limit the present application. For those skilled in the art, changes can be made in the specific implementation methods and application scopes according to the ideas, spirits and principles of the present application, and any modifications, equivalent substitutions, improvements, etc. made by them should be included in the scope of protection of the present application.

Claims

1. A method for detecting a closed loop in magnetic resonance imaging (MRI), characterized in that: The method includes: Obtain the three-dimensional coordinates of each key feature point of the MRI examination object; The three-dimensional coordinates of each key feature point are matched with a three-dimensional human body standard model to obtain a three-dimensional virtual human body model of the inspection object; Extracting three-dimensional surface areas of pre-specified body parts on the three-dimensional virtual human body model; According to predefined pairs of body parts that are prone to skin contact, the distance between the three-dimensional surface areas of two body parts in each pair of body parts is calculated; If the distance between the three-dimensional surface areas of two body parts in any pair of body parts is less than a preset first threshold, it is determined that the inspection object currently has a closed loop risk.

2. The method according to claim 1, characterized in that The step of obtaining the three-dimensional coordinates of each key feature point of the MRI examination object includes: Acquire an RGB image and a depth image of the inspection object; Detecting each key feature point in the RGB image of the inspection object and recording the two-dimensional coordinates of each key feature point; The three-dimensional coordinates of each key feature point are acquired according to the two-dimensional coordinates of each key feature point and the depth image of the inspection object.

3. The method according to claim 2, characterized in that The detecting each key feature point in the RGB image of the inspection object comprises: The RGB image of the inspection object is input into a pre-trained key feature point detection deep learning network model to obtain the two-dimensional coordinates of each key feature point of the inspection object.

4. The method according to claim 1, characterized in that: After acquiring the three-dimensional coordinates of each key feature point of the MRI inspection object and before adapting the three-dimensional coordinates of each key feature point to the three-dimensional human body standard model, the method further includes: The three-dimensional coordinates of each key feature point are converted into a predefined coordinate system.

5. The method according to any one of claims 1 to 4, characterized in that: After obtaining the three-dimensional coordinates of each key feature point of the MRI examination object, the method further includes: According to the three-dimensional coordinates of each key feature point, it is detected whether the inspection object currently has a cross-position. If so, it is determined that the inspection object currently has a closed loop risk.

6. The method according to claim 5, characterized in that The detecting, according to the three-dimensional coordinates of each key feature point, whether the inspection object currently has a cross-position includes: According to predefined key feature point pairs that are prone to cross-positioning and the three-dimensional coordinates of the key feature points, the distance between two key feature points in each key feature point pair is calculated, and if the distance between two key feature points in any key feature point pair is less than a preset second threshold, it is determined that the inspection object currently has a cross-positioning; or, The three-dimensional coordinates of each key feature point of the inspection object are input into a pre-trained cross-position detection deep learning network model to detect whether the inspection object is currently in a cross-position.

7. The method according to claim 5, characterized in that After the distance between the three-dimensional surface areas of two body parts in any pair of body parts is less than a preset first threshold and before determining that the inspection object currently has a closed loop risk, the method further includes: It is determined whether it is detected that the inspection object currently has a cross-position. If so, it is determined that the inspection object currently has a closed loop risk.

8. The method according to claim 1, characterized in that After determining that the inspection object currently has a closed loop risk, the method further includes: Send a closed loop alarm to the inspected object; or / and, A closed-loop alarm is issued to the inspected object, and correct positioning guidance is given.

9. A device (80) for detecting a closed loop in magnetic resonance imaging (MRI), characterized in that: The device (80) comprises: A key feature point acquisition module (81) is used to acquire the three-dimensional coordinates of each key feature point of the MRI examination object; A three-dimensional surface area extraction module (82) is used to adapt the three-dimensional coordinates of each key feature point to a three-dimensional human body standard model to obtain a three-dimensional virtual human body model of the inspection object; and extract the three-dimensional surface area of ​​each pre-specified body part on the three-dimensional virtual human body model; A skin contact detection module (83) is used to calculate the distance between the three-dimensional surface areas of two body parts in each body part pair according to pre-defined pairs of body parts that are prone to skin contact; if the distance between the three-dimensional surface areas of two body parts in any body part pair is less than a preset first threshold, it is determined that the inspection object currently has a closed loop risk.

10. The device (80) according to claim 9, characterized in that The key feature point acquisition module (81) acquires the three-dimensional coordinates of each key feature point of the MRI examination object, including: Acquire an RGB image and a depth image of the inspection object; Detecting each key feature point in the RGB image of the inspection object and recording the two-dimensional coordinates of each key feature point; The three-dimensional coordinates of each key feature point are acquired according to the two-dimensional coordinates of each key feature point and the depth image of the inspection object.

11. The device (80) according to claim 10, characterized in that The key feature point acquisition module (81) detects each key feature point in the RGB image of the inspection object, including: The RGB image of the inspection object is input into a pre-trained key feature point detection deep learning network model to obtain the two-dimensional coordinates of each key feature point of the inspection object.

12. The device (80) according to claim 9, characterized in that After the key feature point acquisition module (81) acquires the three-dimensional coordinates of each key feature point of the MRI examination object, the method further comprises: The three-dimensional coordinates of each key feature point are converted into a predefined coordinate system.

13. The device (80) according to any one of claims 9 to 12, characterized in that: The device (80) further comprises: a cross-position detection module (84) for detecting whether the inspection object currently has a cross-position based on the three-dimensional coordinates of each key feature point, and if so, determining that the inspection object currently has a closed loop risk.

14. The device (80) according to claim 13, characterized in that The cross-position detection module (84) detects whether the inspection object currently has a cross-position according to the three-dimensional coordinates of each key feature point, including: According to predefined key feature point pairs that are prone to cross-positioning and the three-dimensional coordinates of the key feature points, the distance between two key feature points in each key feature point pair is calculated, and if the distance between two key feature points in any key feature point pair is less than a preset second threshold, it is determined that the inspection object currently has a cross-positioning; or, The three-dimensional coordinates of each key feature point of the inspection object are input into a pre-trained cross-position detection deep learning network model to detect whether the inspection object is currently in a cross-position.

15. The device (80) according to claim 13, characterized in that After determining that the distance between the three-dimensional surface areas of two body parts in any pair of body parts is less than a preset first threshold and before determining that the inspection object currently has a closed loop risk, the skin contact detection module (83) further comprises: It is determined whether the cross-position detection module (84) detects that the inspection object currently has a cross-position, and if so, it is determined that the inspection object currently has a closed loop risk.

16. The device (80) according to claim 9, characterized in that After the skin contact detection module (83) determines that the inspection object currently has a closed loop risk, it further comprises: Send a closed loop alarm to the inspected object; or / and, A closed-loop alarm is issued to the inspected object, and correct positioning guidance is given.

17. A magnetic resonance imaging (MRI) system, characterized in that: The MRI system comprises the device (80) for detecting a closed loop in magnetic resonance imaging (MRI) according to any one of claims 9 to 16.

Images