A DIBH training method and device and a DIBH training apparatus

Abstract

The application relates to a DIBH training method and device and a DIBH training equipment. The method comprises the following steps: determining a pose offset amount of a body surface point cloud relative to a reference point cloud; generating a body position adjustment instruction when the pose offset amount is greater than a preset pose offset amount threshold; repeating the process of generating the body position adjustment instruction until the pose offset amount is less than or equal to the pose offset amount threshold; determining a breathing mode parameter of a user in a breath-holding state; generating a breathing mode adjustment instruction when a difference between the breathing mode parameter and a preset breathing mode reference parameter is greater than a preset breathing mode difference threshold; and repeating the process of generating the breathing mode adjustment instruction until the difference between the breathing mode parameter and the breathing mode reference parameter is less than or equal to the preset breathing mode difference threshold. The method can solve the problem of poor training effect caused by the inability to ensure the consistency of the body position in the breath-holding state between multiple training and between training and clinical treatment.

Description

Technical Field

[0001] This application relates to the fields of signal processing and radiotherapy technology, and in particular to a DIBH training method, apparatus, and DIBH training device. Background Technology

[0002] In radiotherapy for thoracic and abdominal tumors, respiratory motion causes dynamic displacement of the tumor target area on the order of millimeters to centimeters. During radiotherapy, there is an inherent delay in equipment response and beam delivery. If the target displacement caused by respiratory motion cannot be effectively controlled, misalignment between the irradiation field and the tumor target area will occur, leading to insufficient target dose, decreased local tumor control, or excessive radiation dose to organs at risk such as the heart and lungs, causing radiation damage. Deep Inspiration Breath Hold (DIBH) is currently the mainstream respiratory motion control technique in radiotherapy for thoracic and abdominal tumors. It guides patients to hold their breath stably after a deep inspiration, keeping the target area relatively fixed and increasing the spatial distance between the target area and organs at risk such as the heart and lungs, thus reducing the risk of radiation damage. Therefore, DIBH training for patients before radiotherapy is particularly important.

[0003] Existing DIBH training in hospital settings utilizes optical surface monitoring systems to achieve three-dimensional reconstruction of the patient's body surface and six-degree-of-freedom position monitoring through binocular stereo vision or structured light technology. Its positioning accuracy can reach the millimeter level. However, due to the large size and high cost of the equipment, it is deployed in hospital positioning rooms and accelerator rooms, and can only complete training guidance. After completing the training guidance, DIBH training needs to be completed independently at home.

[0004] For home-based training scenarios, existing technologies propose using two-dimensional curves to display changes in respiratory amplitude or flow rate for feedback training. However, this method only trains the duration and volume of breath-holding, failing to quantify respiratory patterns or guide the reproduction of posture during breath-holding. Because the chest expansion pattern during deep inspiration and breath-holding is related to trunk position, the diaphragmatic displacement and chest deformation differ depending on the posture for the same breath-holding volume. This leads to a discrepancy between the actual and labeled spatial locations of the tumor target area during breath-holding. Therefore, existing training methods cannot guarantee consistency in posture during breath-holding across multiple training sessions and between training and clinical treatment, resulting in poor training effectiveness.

[0005] Existing home-based training methods for DIBH (Digital Injury By Bye) cannot guarantee the consistency of postural space during breath-holding between multiple training sessions and between training and clinical treatment, resulting in poor training effects. Currently, no effective solution has been proposed. Summary of the Invention

[0006] Therefore, it is necessary to provide a DIBH training method, apparatus, and DIBH training device to address the aforementioned technical problems.

[0007] Firstly, this application provides a DIBH training method. Applied to a DIBH training device, the method includes:

[0008] Determine the pose offset of the user's body surface point cloud relative to a preset reference point cloud during free breathing; the pose offset includes rotation offset and translation offset.

[0009] When the pose offset is greater than a preset pose offset threshold, a body position adjustment command corresponding to the pose offset is generated; the body position adjustment command is used to guide the user to perform rotation and / or translation.

[0010] The process of repeatedly acquiring the pose offset of the user's body surface point cloud relative to the preset reference point cloud in a free breathing state after the body position adjustment, and generating a body position adjustment command, continues until the pose offset is less than or equal to the preset pose offset threshold.

[0011] Instruct users to take a deep breath and hold it, and determine the user's breathing pattern parameters while holding their breath;

[0012] When the difference between the user's breathing mode parameters in a breath-holding state and the preset breathing mode baseline parameters is greater than the preset breathing mode difference threshold, a breathing mode adjustment command is generated; the breathing mode adjustment command is used to guide the user to perform chest breathing and / or abdominal breathing.

[0013] The process of repeatedly acquiring the user's breathing mode parameters under breath-holding state after the breathing mode adjustment and generating breathing mode adjustment instructions continues until the difference between the user's breathing mode parameters under breath-holding state and the preset breathing mode baseline parameters is less than or equal to the preset breathing mode difference threshold.

[0014] In one embodiment, determining the pose offset of the user's surface point cloud relative to a preset reference point cloud during free breathing includes:

[0015] Acquire the first body surface image of the user in a free breathing state;

[0016] Extract the first body surface point cloud from the target region of the first body surface image;

[0017] Determine the pose offset of the first body surface point cloud relative to the preset reference point cloud.

[0018] In one embodiment, determining the pose offset of the first body surface point cloud relative to the preset reference point cloud includes:

[0019] Calculate the first centroid of the first body surface point cloud and the second centroid of the preset reference point cloud;

[0020] Based on the first centroid and the second centroid, calculate the covariance matrix between the first body surface point cloud and the preset reference point cloud;

[0021] The covariance matrix is ​​solved by SVD to determine the pose offset of the first body surface point cloud relative to the preset reference point cloud.

[0022] In one embodiment, the translational offset includes X-axis offset, Y-axis offset, and Z-axis offset; the rotational offset includes rotational offset around the X-axis, rotational offset around the Y-axis, and rotational offset around the Z-axis; the preset pose offset threshold includes preset X-axis offset threshold, preset Y-axis offset threshold, preset Z-axis offset threshold, preset rotational offset threshold around the X-axis, preset rotational offset threshold around the Y-axis, and preset rotational offset threshold around the Z-axis; when the pose offset is greater than the preset pose offset threshold, generating a body position adjustment command corresponding to the pose offset includes:

[0023] When the absolute value of the X-axis offset is greater than or equal to the preset X-axis offset threshold, an adjustment command to move the body left or right is generated.

[0024] When the absolute value of the Y-axis offset is greater than or equal to the preset Y-axis offset threshold, an adjustment command to move the body up and down is generated.

[0025] When the absolute value of the Z-axis offset is greater than or equal to the preset Z-axis offset threshold, an adjustment command to lift or relax the chest is generated.

[0026] When the absolute value of the rotation offset around the Y-axis is greater than or equal to the preset rotation offset threshold around the Y-axis, a turning adjustment command is generated.

[0027] When the absolute value of the rotation offset around the X-axis is greater than or equal to the preset rotation offset threshold around the X-axis, an upper body pitch adjustment command is generated.

[0028] And / or, when the absolute value of the rotational offset around the Z-axis is greater than or equal to a preset rotational offset threshold around the Z-axis, an adjustment command for shoulder leveling is generated.

[0029] In one embodiment, after repeatedly acquiring the pose offset of the user's surface point cloud relative to a preset reference point cloud in a free breathing state after posture adjustment, and generating a posture adjustment command, until the pose offset is less than or equal to the preset pose offset threshold, the method further includes:

[0030] A second body surface image is acquired after the user has adjusted their posture and is in a free breathing state. Based on a preset reference point cloud and the second body surface image, a target transformation matrix corresponding to the user's adjusted posture is determined. The target transformation matrix is ​​a transformation matrix that adjusts the coordinate system corresponding to the second body surface point cloud in the target area of ​​the second body surface image to the coordinate system corresponding to the preset reference point cloud.

[0031] In one embodiment, the breathing pattern parameters include the proportion of thoracic breathing and the proportion of abdominal breathing; guiding the user to perform deep inhalation and breath-holding, and determining the user's breathing pattern parameters during breath-holding, includes:

[0032] Instruct users to take a deep breath and hold it, and capture a third-party image of the user's body surface when the deep breath is completed;

[0033] Extract the third surface point cloud from the target region of the third surface image;

[0034] The third body surface point cloud is divided into different regions according to its location; the different regions include an upper region, a middle region, and a lower region;

[0035] The quotient of the average value of each point cloud in the upper region along the Z-axis and the average value of each point cloud in the middle region along the Z-axis is determined as the proportion of chest breathing in the user's breath-holding state; the Z-axis direction is the user's abdominal and back direction.

[0036] The quotient of the average value of each point cloud in the lower region along the Z-axis and the average value of each point cloud in the middle region along the Z-axis is determined as the proportion of abdominal breathing of the user in the breath-holding state.

[0037] In one embodiment, after extracting the third surface point cloud from the target region of the third surface image, the method further includes:

[0038] Based on the target transformation matrix, the point cloud of the third body surface is corrected to obtain the corrected point cloud of the third body surface.

[0039] In one embodiment, determining the target transformation matrix corresponding to the user's adjusted body position based on a preset reference point cloud and a second body surface image includes:

[0040] Using the rotation offset and the translation offset as the initial values ​​for iteration, the second body surface point cloud in the target area of ​​the second body surface image is projected onto the coordinate system corresponding to the preset reference point cloud to obtain the projected point cloud, and the root mean square error of the distance between each pair of points between the projected point cloud and the preset reference point cloud is calculated.

[0041] The root mean square error is used as the objective function for iterative optimization, and the new rotation offset and the new translation offset are solved by SVD decomposition.

[0042] Using the new rotation offset and the new translation offset as iteration parameters, the process of updating the root mean square error is repeated until the root mean square error is less than the preset error threshold, or the root mean square error is less than the preset convergence threshold and the number of iterations reaches the preset number of iterations threshold; the iteration parameters at this time are determined as the target transformation matrix corresponding to the user's adjusted body position.

[0043] Secondly, this application also provides a DIBH training device for use in DIBH training equipment. The device includes:

[0044] The offset determination module is used to determine the pose offset of the user's body surface point cloud relative to a preset reference point cloud in a free breathing state; the pose offset includes rotation offset and translation offset.

[0045] The body position adjustment instruction generation module is used to generate a body position adjustment instruction corresponding to the pose offset when the pose offset is greater than a preset pose offset threshold; the body position adjustment instruction is used to guide the user to perform rotation and / or translation.

[0046] The body position adjustment module is used to repeatedly acquire the pose offset of the user's body surface point cloud in a free breathing state after body position adjustment relative to a preset reference point cloud, and generate a body position adjustment command until the pose offset is less than or equal to the preset pose offset threshold.

[0047] The breathing mode parameter determination module is used to guide users to perform deep inhalation and breath-holding, and to determine the user's breathing mode parameters during breath-holding.

[0048] The breathing mode adjustment instruction generation module is used to generate a breathing mode adjustment instruction when the difference between the user's breathing mode parameters in the breath-holding state and the preset breathing mode baseline parameters is greater than a preset breathing mode difference threshold; the breathing mode adjustment instruction is used to guide the user to perform chest breathing and / or abdominal breathing.

[0049] The system also includes a breathing mode adjustment module, which repeatedly acquires the user's breathing mode parameters in the breath-holding state after the breathing mode adjustment and generates a breathing mode adjustment command until the difference between the user's breathing mode parameters in the breath-holding state and the preset breathing mode baseline parameters is less than or equal to the preset breathing mode difference threshold.

[0050] Thirdly, this application also provides a DIBH training device. The DIBH training device includes a memory, a processor, and a blood oxygen monitoring unit. The memory stores a computer program, and the processor executes the computer program to implement the DIBH training method described in the first aspect. The blood oxygen monitoring unit is used to monitor the user's blood oxygen saturation.

[0051] The aforementioned DIBH training method, apparatus, and DIBH training device determine the pose offset of the user's body surface point cloud relative to a preset reference point cloud during free breathing. When the pose offset exceeds a preset pose offset threshold, a posture adjustment command corresponding to the pose offset is generated to guide the user to rotate and / or translate. This process of generating posture adjustment commands to guide the user to adjust their posture is repeated until the pose offset is less than or equal to the preset pose offset threshold. By adjusting the rotation and / or translation during free breathing, the device ensures that the user's body rotation angle and position are consistent with the posture corresponding to the preset reference point cloud. Furthermore, it guides the user to perform deep inhalation and breath-holding and determines the user's breathing pattern parameters during breath-holding. Furthermore, when the difference between the user's breathing pattern parameters during breath-holding and the preset breathing pattern baseline parameters exceeds a preset breathing pattern difference threshold, a breathing pattern adjustment command is generated to guide the user to perform thoracic and / or abdominal breathing. This method adjusts the breathing pattern parameters, and by adjusting the proportion of thoracic and abdominal breathing, the diaphragm's downward displacement and chest deformation during breath-holding are further adjusted. This ensures that the user's posture during breath-holding is consistent with the posture corresponding to the baseline point cloud, guaranteeing consistency between the user's posture during breath-holding and during clinical treatment, as well as consistency during multiple training sessions. This solves the problem of poor training effect caused by the inability of existing DIBH training methods in home training scenarios to guarantee the consistency of posture space during breath-holding between multiple training sessions and between training and clinical treatment.

[0052] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0053] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0054] Figure 1 A hardware structure block diagram of a terminal for the DIBH training method provided in an embodiment of this application;

[0055] Figure 2A flowchart of a DIBH training method provided in an embodiment of this application;

[0056] Figure 3 A flowchart of a preferred embodiment of the DIBH training method provided in this application;

[0057] Figure 4 This is a structural block diagram of a DIBH training device provided in an embodiment of this application. Detailed Implementation

[0058] To better understand the purpose, technical solution, and advantages of this application, the application is described and explained below in conjunction with the accompanying drawings and embodiments.

[0059] Unless otherwise defined, the technical or scientific terms used in this application shall have the general meaning understood by one of ordinary skill in the art to which this application pertains. Words such as “a,” “an,” “an,” “the,” “the,” and “these” used in this application do not indicate quantitative limitation and may be singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or modules (units) is not limited to the listed steps or modules (units) but may include steps or modules (units) not listed, or may include other steps or modules (units) inherent to these processes, methods, products, or devices. Words such as “connected,” “linked,” and “coupled” used in this application are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. Normally, the character " / " indicates that the objects before and after it are in an "or" relationship. The terms "first," "second," "third," etc., used in this application are merely to distinguish similar objects and do not represent a specific order of objects.

[0060] The method embodiments provided in this example can be executed on a terminal, computer, or similar computing device. For example, it can run on a terminal. Figure 1 This is a hardware structure block diagram of the terminal for the DIBH training method in this embodiment. For example... Figure 1 As shown, a terminal may include one or more ( Figure 1Only one is shown in the diagram. A processor 102 and a memory 104 for storing data are also included. The processor 102 may be, but is not limited to, a microprocessor (MCU) or a programmable logic device (FPGA). The terminal may also include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that… Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the terminal described above. For example, the terminal may also include components that are larger than... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown are illustrated.

[0061] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the DIBH training method in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0062] The transmission device 106 is used to receive or send data via a network. This network includes a wireless network provided by the terminal's communication provider. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 can be a Radio Frequency (RF) module used for wireless communication with the Internet.

[0063] This embodiment provides a DIBH training method, applied to a DIBH training device. Figure 2 This is a flowchart of the DIBH training method in this embodiment, as follows: Figure 2 As shown, the process includes the following steps:

[0064] Step S210: Determine the pose offset of the user's body surface point cloud relative to the preset reference point cloud during free breathing; the pose offset includes rotation offset and translation offset.

[0065] Before using the DIBH training device at home, training guidance needs to be completed in a hospital setting. During the training guidance process, a baseline calibration and coordinate system establishment need to be performed using an optical body surface monitoring system, that is, to determine the preset baseline point cloud and the coordinate system corresponding to the baseline point cloud. The preset baseline point cloud mentioned above is the preset baseline point cloud of the user in a free breathing state. The process of benchmark calibration and coordinate system establishment is as follows: The user being calibrated lies supine, exposing the chest and abdominal area, maintaining a standard treatment position; after completing the binocular vision camera calibration, the display unit is calibrated to establish a one-to-one correspondence between the binocular vision acquisition image and the anatomical position mapping coordinates of the body surface area within the user's actual field of vision, and left and right mirror matching is completed simultaneously; the patient maintains a stable state of resting and free breathing. After the doctor confirms that the user is in a standard free breathing state and reaches a suitable acquisition position, the system acquires real-time images of the body surface area within the field of vision under the patient's stable free breathing state, and quickly reconstructs the benchmark point cloud of the body surface area within the field of vision; the doctor manually and independently delineates the body surface area within the field of vision corresponding to the tumor target area on the real-time reconstructed point cloud image of the body surface area under free breathing state, i.e., the benchmark point cloud of the target area, i.e., the preset benchmark point cloud. The system automatically generates a binary mask of the preset benchmark point cloud: the target area delineated by the doctor is marked as 1 (valid area), and other areas are marked as 0 (invalid area). The preset benchmark point cloud is extracted from the point cloud image of the body surface area under free breathing state within the field of vision using the binary mask of the preset benchmark point cloud. Given a pre-defined reference point cloud, a coordinate system corresponding to the reference point cloud is constructed according to a pre-defined coordinate system construction method. This pre-defined coordinate system construction method can be specifically set according to specific circumstances and requirements; this embodiment does not impose any specific limitations on it.

[0066] In addition, during training guidance, baseline calibration includes calibration under deep inspiration and breath-holding conditions, specifically calibrating the DIBH baseline and the target region of DIBH, thereby determining preset respiratory mode baseline parameters. These preset respiratory mode baseline parameters may include the baseline proportion of abdominal breathing and the baseline proportion of thoracic breathing. The calibration of the DIBH baseline and the target region of DIBH allows doctors to guide users in performing DIBH actions that conform to clinical radiotherapy guidelines. After visual observation and experience-based judgment that the current breath-holding state meets treatment requirements, the system acquires real-time images of the body surface region within the field of view under this state, quickly reconstructing the DIBH baseline point cloud within the field of view. Doctors then manually and independently delineate the target region (ROI) of the body surface region corresponding to the tumor target area on the real-time reconstructed DIBH point cloud image. The system automatically generates a binary mask: marking the DIBH region delineated by the doctor as 1 (valid region) and other regions as 0 (invalid region). This mask is used for training guidance and target assessment under the DIBH state. The target reference point cloud of DIBH-ROI is extracted from the DIBH reference point cloud of the body surface region within the visual field using the ROI mask of DIBH, and used as the target reference for subsequent DIBH training. Furthermore, the process of determining the preset respiratory mode reference parameters is as follows: the system automatically calculates and stores personalized respiratory mode reference parameters: based on the current DIBH state that meets the treatment requirements, the reference abdominal breathing ratio and the reference thoracic breathing ratio are calculated. The reference thoracic breathing ratio reflects the degree of thoracic breathing participation during patient treatment. The calculation method for the reference thoracic breathing ratio is: removing the average Z-axis (patient anterior-posterior direction) position of the upper part of the DIBH-ROI (from below the clavicle to above the nipple) by the quotient of the average Z-axis displacement of the middle part of the DIBH-ROI (nipple level region). The larger the value of the reference thoracic breathing ratio, the more obvious the thoracic breathing, and the more the shoulders and upper chest are involved. The reference abdominal breathing ratio reflects the degree of abdominal breathing participation during patient treatment. The above-mentioned baseline abdominal breathing percentage is calculated as follows: the average Z-axis displacement of the lower DIBH-ROI (from below the nipple to the xiphoid process) divided by the average Z-axis displacement of the middle DIBH-ROI (the nipple-level region). A higher baseline abdominal breathing percentage indicates more pronounced abdominal breathing and greater abdominal involvement.

[0067] The aforementioned DIBH training device includes an image acquisition unit, which includes a binocular camera that simultaneously acquires images of the user's body surface through two cameras with different perspectives. Determining the pose offset of the user's body surface point cloud relative to a preset reference point cloud during free breathing involves: acquiring a first body surface image of the user during free breathing; extracting the first body surface point cloud from the target region of the first body surface image; and determining the pose offset of the first body surface point cloud relative to the preset reference point cloud. The free breathing state refers to a state where the user maintains a calm, natural breathing rhythm without actively inhaling or holding their breath. The first body surface point cloud is the user's body surface point cloud during free breathing. In three-dimensional space, the state of a rigid body relative to a reference coordinate system can be described by six degrees of freedom: three translational degrees of freedom and three rotational degrees of freedom. Generally, position, used to describe the location of an object in space, can be represented by a translation vector containing X, Y, and Z directional components. In this embodiment, the current rigid body is the human body. We can define the left-right direction (from right to left as positive in a supine position) as the X-direction, the head-to-toe direction (from feet to head as positive) as the Y-direction, and the abdomen-back direction (from back to abdomen as positive, reflecting chest out / inward posture and body surface undulations) as the Z-direction. Posture describes the object's orientation / angle and can be represented by a rotation matrix or Euler angles, containing rotational components about the X, Y, and Z axes, referred to as the rotation around the X-axis, rotation around the Y-axis, and rotation around the Z-axis. Therefore, the aforementioned pose offset can represent the position and orientation deviation, i.e., the deviation of the current body surface point cloud from the reference point cloud in six degrees of freedom. The aforementioned rotational offset can include rotational offsets around the X-axis, rotational offsets around the Y-axis, and rotational offsets around the Z-axis. The aforementioned translational offset can include X-axis offsets, Y-axis offsets, and Z-axis offsets.

[0068] Step S220: When the pose offset is greater than the preset pose offset threshold, a body position adjustment command corresponding to the pose offset is generated; the body position adjustment command is used to guide the user to perform rotation and / or translation.

[0069] The aforementioned preset pose offset thresholds may include one or more of the following: preset X-axis offset threshold, preset Y-axis offset threshold, preset Z-axis offset threshold, preset rotational offset threshold around the X-axis, preset rotational offset threshold around the Y-axis, and preset rotational offset threshold around the Z-axis. When generating the body position adjustment command corresponding to the pose offset, if deviations in multiple directions exceed the limit simultaneously, the system generates commands according to a preset priority strategy to avoid confusing the user by outputting multiple commands simultaneously. Specifically, body position adjustment commands are generated according to the principle of translation before rotation. Translational deviations directly affect the absolute spatial position of the target area and have the highest priority, followed by rotational deviations. Within the same category, the direction with the larger deviation is given priority in guidance (generation of body position adjustment commands). The aforementioned body position adjustment commands can be voice commands or visual commands. Voice commands can be voice guidance broadcast through a speaker. Visual commands can be guidance commands displayed as text on the display unit. The aforementioned body position adjustment commands may include the direction and magnitude of adjustment related to the pose offset.

[0070] Step S230: Repeat the process of acquiring the pose offset of the user's body surface point cloud in free breathing state after posture adjustment relative to the preset reference point cloud and generating posture adjustment instructions until the pose offset is less than or equal to the preset pose offset threshold.

[0071] In this step, the pose offset of the user's body surface point cloud relative to the preset reference point cloud under the free breathing state is repeatedly obtained, and a body position adjustment command is generated. That is, the process of repeating steps S210 to S220 is repeated, that is, the process of repeatedly updating the body position and obtaining the body position offset to generate the body position adjustment command.

[0072] The determination that the pose offset is less than or equal to a preset pose offset threshold can be made as follows: the absolute value of the X-axis offset is less than the preset X-axis offset threshold; the absolute value of the Y-axis offset is less than the preset Y-axis offset threshold; the absolute value of the Z-axis offset is less than the preset Z-axis offset threshold; the absolute value of the rotation offset around the Y-axis is less than the preset rotation offset threshold around the Y-axis; the absolute value of the rotation offset around the X-axis is less than the preset rotation offset threshold around the X-axis; and the absolute value of the rotation offset around the Z-axis is greater than or equal to the preset rotation offset threshold around the Z-axis.

[0073] The determination that the pose offset is less than or equal to a preset pose offset threshold can also be made if the absolute value of the X-axis offset is less than a preset X-axis offset threshold, the absolute value of the Y-axis offset is less than a preset Y-axis offset threshold, the absolute value of the Z-axis offset is less than a preset Z-axis offset threshold, the absolute value of the rotational offset around the Y-axis is less than a preset rotational offset threshold around the Y-axis, the absolute value of the rotational offset around the X-axis is less than a preset rotational offset threshold around the X-axis, the absolute value of the rotational offset around the Z-axis is greater than or equal to a preset rotational offset threshold around the Z-axis, and the overall centroid offset is less than a preset centroid offset threshold. The centroid offset threshold can be specifically set according to specific needs and actual application scenarios; this embodiment does not impose specific limitations on it. The centroid offset threshold is greater than a preset X-axis offset threshold, greater than a preset Y-axis offset threshold, and greater than a preset Z-axis offset threshold.

[0074] The above-mentioned overall centroid offset The calculation process is as follows:

[0075] ;

[0076] Where X1 represents the X-axis offset, Y1 represents the Y-axis offset, and Z1 represents the Z-axis offset.

[0077] Step S240: Instruct the user to take a deep breath and hold it, and determine the user's breathing mode parameters while holding their breath.

[0078] The aforementioned guidance for users to perform deep inhalation and breath-holding can be provided via voice prompts or by displaying text on a screen. The aforementioned determination of the user's breathing pattern parameters during breath-holding can be based on the specific breathing pattern parameters the user determines when they complete a deep inhalation and begin holding their breath. By determining these breathing pattern parameters, it ensures that the user can complete the preset target duration of breath-holding training after adjusting the breathing pattern parameters.

[0079] Step S250: When the difference between the user's breathing mode parameters in the breath-holding state and the preset breathing mode baseline parameters is greater than the preset breathing mode difference threshold, a breathing mode adjustment command is generated; the breathing mode adjustment command is used to guide the user to perform chest breathing and / or abdominal breathing.

[0080] The aforementioned preset breathing pattern difference thresholds may include the difference thresholds for the proportion of chest breathing and the difference thresholds for the proportion of abdominal breathing. The aforementioned breathing pattern adjustment commands may include commands such as "Please relax / tighten your shoulders slightly," "Inhale more through your nose / mouth," "Return to the breathing pattern at the starting position," and "Please slightly increase / decrease abdominal exertion." The specific output breathing pattern adjustment command can be determined based on the difference between the user's breathing pattern parameters during breath-holding and the preset breathing pattern baseline parameters, and is used to reduce the absolute value of the difference between the user's breathing pattern parameters during breath-holding and the preset breathing pattern baseline parameters.

[0081] The aforementioned preset breathing pattern reference parameters may include preset thoracic breathing pattern reference values ​​and preset abdominal breathing pattern reference values. When the difference between the user's breathing pattern parameters during breath-holding and the preset breathing pattern reference parameters is greater than a preset breathing pattern difference threshold, this can be either the absolute value of the difference between the user's thoracic breathing percentage during breath-holding and the preset thoracic breathing pattern reference value being greater than a threshold for the difference in thoracic breathing percentage, or the absolute value of the difference between the user's abdominal breathing percentage during breath-holding and the preset abdominal breathing pattern reference value being greater than a threshold for the difference in abdominal breathing percentage.

[0082] Step S260: Repeat the process of acquiring the user's breathing mode parameters in the breath-holding state after the breathing mode adjustment and generating a breathing mode adjustment command until the difference between the user's breathing mode parameters in the breath-holding state and the preset breathing mode baseline parameters is less than or equal to the preset breathing mode difference threshold.

[0083] When the difference between the user's breathing mode parameters during breath-holding and the preset breathing mode baseline parameters is less than or equal to the preset breathing mode difference threshold, it can be that the absolute value of the difference between the user's chest breathing ratio during breath-holding and the preset chest breathing mode baseline value is less than or equal to the chest breathing ratio difference threshold, and the absolute value of the difference between the user's abdominal breathing ratio during breath-holding and the preset abdominal breathing mode baseline value is less than or equal to the abdominal breathing ratio difference threshold.

[0084] Steps S210 to S260 above determine the pose offset of the user's body surface point cloud relative to a preset reference point cloud during free breathing. When the pose offset is greater than a preset pose offset threshold, a posture adjustment command corresponding to the pose offset is generated to guide the user to rotate and / or translate. The process of generating posture adjustment commands to guide the user to adjust their posture is repeated until the pose offset is less than or equal to the preset pose offset threshold. By adjusting the rotation and / or translation during free breathing, the rotation angle and position of the user's body are ensured to be consistent with the posture corresponding to the preset reference point cloud. In turn, the user is guided to take a deep breath and hold their breath, and the breathing mode parameters of the user during breath-holding are determined. Furthermore, when the difference between the user's breathing pattern parameters during breath-holding and the preset breathing pattern baseline parameters exceeds a preset breathing pattern difference threshold, a breathing pattern adjustment command is generated to guide the user to perform thoracic and / or abdominal breathing. This method adjusts the breathing pattern parameters, and by adjusting the proportion of thoracic and abdominal breathing, the diaphragm's downward displacement and chest deformation during breath-holding are further adjusted. This ensures that the user's posture during breath-holding is consistent with the posture corresponding to the baseline point cloud, guaranteeing consistency between the user's posture during breath-holding and during clinical treatment, as well as consistency during multiple training sessions. This solves the problem of poor training effect caused by the inability of existing DIBH training methods in home training scenarios to guarantee the consistency of posture space during breath-holding between multiple training sessions and between training and clinical treatment.

[0085] In one embodiment, step S210, determining the pose offset of the user's body surface point cloud relative to a preset reference point cloud during free breathing, includes:

[0086] Step S212: Obtain the first body surface image of the user in a free breathing state.

[0087] In this step, acquiring the first body surface image of the user in a free breathing state can be achieved by using the image acquisition unit of the DIBH training device to acquire the first body surface image of the user in a free breathing state. Because the image acquisition unit includes a binocular camera, the first body surface image is obtained by simultaneously capturing two two-dimensional body surface images of the user's body surface from different perspectives using two cameras at the same time.

[0088] In a preferred embodiment, the image acquisition unit may further include multiple calibration subunits, each including a reflective marker for attaching to different marking points on the body surface. The positions of the marking points corresponding to each calibration subunit are the positions marked during training guidance in a hospital setting. With reflective markers attached to each marking point, multiple marking points can be quickly located on the captured first body surface image. It should be noted that the number of calibration subunits can be specifically set according to specific circumstances and needs; this embodiment does not impose a specific limitation. To ensure the accuracy of body position adjustment, the number of calibration subunits should not be less than nine, and at least nine marking points should be marked. These include three distributed in the upper region (from below the clavicle to above the nipple), three distributed in the middle region (at the level of the nipple), and three distributed in the lower region (from below the nipple to the xiphoid process).

[0089] Step S214: Extract the first body surface point cloud from the target region of the first body surface image.

[0090] The aforementioned target region is the region of interest (FB-ROI) under free breathing conditions. This target region may include the user's chest and abdomen. Specifically, the scope of the target region may include at least the projection area of ​​the tumor target area onto the body surface, and at least the relevant areas reflecting breathing pattern parameters. The first body surface point cloud extracted from the target region of the first body surface image can be a high-density three-dimensional point cloud of the target region obtained after three-dimensional reconstruction and FB-ROI masking of the first body surface image. The masking can employ binary masking or other masking methods to mask the data after three-dimensional reconstruction of the first body surface image; the specific masking method is not specifically limited in this embodiment.

[0091] Step S216: Determine the pose offset of the first body surface point cloud relative to the preset reference point cloud.

[0092] The aforementioned determination of the pose offset of the first body surface point cloud relative to the preset reference point cloud can be achieved by calculating the first centroid of the first body surface point cloud and the second centroid of the preset reference point cloud, calculating the covariance matrix between the first body surface point cloud and the preset reference point cloud based on the first centroid and the second centroid, and then performing SVD (Singular Value Decomposition) on the covariance matrix to determine the pose offset of the first body surface point cloud relative to the preset reference point cloud.

[0093] Steps S212 to S216 above involve acquiring a first body surface image of the user in a free breathing state, then extracting a first body surface point cloud from the target area of ​​the first body surface image, and finally determining the pose offset of the first body surface point cloud relative to a preset reference point cloud. By determining the pose offset, it is easier to generate corresponding body position adjustment instructions based on the pose offset to guide the user to rotate and / or translate, so that the user can adjust their body position according to the prompts.

[0094] In a preferred embodiment, the DIBH training device further includes a visualization display unit, which is used to simultaneously display two sets of FB-ROI point cloud contours that correspond one-to-one with the anatomical locations of the body surface regions within the patient's visual field: an initial baseline contour (blue semi-transparent area) and a current real-time contour (red solid area). The initial baseline contour is the FB-ROI baseline point cloud read during the initial clinical calibration. , This represents the 3D coordinates of the i-th marker point in the statically acquired FB-ROI reference point cloud (a preset reference point cloud), projected onto the screen via a calibration matrix, and used as a fixed alignment reference. The above current real-time contour (the solid red area) is the extracted current real-time FB-ROI point cloud. The image is projected onto the screen via a calibration matrix. (The above...) This represents the three-dimensional coordinates of the i-th marker point within the statically acquired FB-ROI point cloud (first body surface point cloud).

[0095] Specifically, in one embodiment, step S216, determining the pose offset of the first body surface point cloud relative to a preset reference point cloud, includes:

[0096] Step S2162: Calculate the first centroid of the first body surface point cloud and the second centroid of the preset reference point cloud.

[0097] The above calculation of the first centroid of the first surface point cloud The calculation process is as follows:

[0098] ;

[0099] The above calculation pre-sets the second centroid of the reference point cloud. The calculation process is as follows:

[0100] ;

[0101] Where N is the number of effective points in the high-density point cloud.

[0102] Step S2164: Calculate the covariance matrix between the first body surface point cloud and the preset reference point cloud based on the first centroid and the second centroid.

[0103] The calculation process for the covariance matrix H between the first surface point cloud and the preset reference point cloud, based on the first centroid and the second centroid, is as follows:

[0104] ;

[0105] Step S2166: Perform SVD on the covariance matrix to determine the pose offset of the first body surface point cloud relative to the preset reference point cloud.

[0106] The above-mentioned SVD solution of the covariance matrix determines the pose offset of the first body surface point cloud relative to the preset reference point cloud, i.e., the covariance matrix is ​​solved. Perform SVD to determine the rotational offset. Translation offset The process is as follows:

[0107] ;

[0108] ;

[0109] ;

[0110] in, This is the transpose of the right singular vector matrix. It is a singular value diagonal matrix. V is a left singular vector matrix. rt It is a right singular vector matrix.

[0111] For translation offset The data is decomposed in three directions to obtain the X-axis offset, Y-axis offset, and Z-axis offset. The rotation offset is then... Converting to Euler angles and decomposing them, we obtain Roll(γ), which is the rotational offset around the Y-axis (head-to-feet major axis), representing the user's "left-right turning / turning deviation"; Pitch(β), which is the rotational offset around the X-axis (left-right axis), representing the user's "head up / down, uneven upper body posture / collapse, and partial deviation of the waist and back from the bed surface"; and Yaw(α), which is the rotational offset around the Z-axis (abdomen-back axis), representing the user's "uneven shoulder height and upper body tilt deviation".

[0112] Steps S2162 to S2166 above involve calculating the first centroid of the first body surface point cloud and the second centroid of a preset reference point cloud. Then, based on the first and second centroids, the covariance matrix between the first body surface point cloud and the preset reference point cloud is calculated. Finally, the covariance matrix is ​​solved by SVD to determine the pose offset of the first body surface point cloud relative to the preset reference point cloud. By determining the pose offset of the first body surface point cloud relative to the preset reference point cloud, the deviation between the user and the reference body position is determined, so as to facilitate subsequent adjustments to the body position based on the deviation between the user and the reference body position.

[0113] In another embodiment, the translation offset includes X-axis offset, Y-axis offset, and Z-axis offset; the rotation offset includes rotation offset around the X-axis, rotation offset around the Y-axis, and rotation offset around the Z-axis; the preset pose offset threshold includes preset X-axis offset threshold, preset Y-axis offset threshold, preset Z-axis offset threshold, preset rotation offset threshold around the X-axis, preset rotation offset threshold around the Y-axis, and preset rotation offset threshold around the Z-axis; step S220, when the pose offset is greater than the preset pose offset threshold, generates a body position adjustment command corresponding to the pose offset, including:

[0114] Step S221: When the absolute value of the X-axis offset is greater than or equal to the preset X-axis offset threshold, an adjustment command for moving the body left and right is generated.

[0115] In this step, the preset X-axis offset threshold can be set according to specific needs and actual application scenarios, and this embodiment does not impose specific limitations. For example, the preset X-axis offset threshold is 0.3cm. Whether a leftward or rightward adjustment command is generated depends on the sign of the X-axis offset. A positive X-axis offset generates a leftward adjustment command, and a negative X-axis offset generates a rightward adjustment command. The specific value of the adjustment command can be set to the value of the X-axis offset.

[0116] Step S222: When the absolute value of the Y-axis offset is greater than or equal to the preset Y-axis offset threshold, an adjustment command for moving the body up and down is generated.

[0117] The preset Y-axis offset threshold can be set according to specific needs and actual application scenarios, and this embodiment does not impose specific limitations. For example, the preset Y-axis offset threshold is 0.3cm. Whether the adjustment command generated is for upward or downward movement of the body is determined by the sign of the Y-axis offset. When the Y-axis offset is positive, an adjustment command for upward (headward) movement of the body is generated; when the Y-axis offset is negative, an adjustment command for downward (footward) movement of the body is generated. The specific value of the adjustment command can be set to the value of the Y-axis offset.

[0118] Step S223: When the absolute value of the Z-axis offset is greater than or equal to the preset Z-axis offset threshold, an adjustment command to lift or relax the chest is generated.

[0119] The preset Z-axis offset threshold can be set according to specific needs and actual application scenarios, and this embodiment does not impose specific limitations. For example, the preset Z-axis offset threshold is 0.3cm. Whether a chest-lifting adjustment command or a chest-relaxing adjustment command is generated is determined by the sign of the Z-axis offset. A positive Z-axis offset generates a chest-lifting adjustment command, and a negative Z-axis offset generates a chest-relaxing adjustment command. The specific adjustment command value can be set to the Z-axis offset value.

[0120] Step S224: When the absolute value of the rotation offset around the Y-axis is greater than or equal to the preset rotation offset threshold around the Y-axis, a turning adjustment command is generated.

[0121] Furthermore, the aforementioned preset rotational offset threshold around the Y-axis can be specifically set according to specific needs and actual application scenarios, and this embodiment does not impose specific limitations here. For example, the aforementioned preset rotational offset threshold around the Y-axis is 2°. The aforementioned turning adjustment commands include left turn adjustment commands and right turn adjustment commands. Whether a left turn adjustment command or a right turn adjustment command is generated is determined by the sign of the rotational offset around the Y-axis. When the rotational offset around the Y-axis is positive, a right turn adjustment command is generated; when the rotational offset around the Y-axis is negative, a left turn adjustment command is generated. The specific value of the adjustment command can be set to the value of the rotational offset around the Y-axis.

[0122] Step S225: When the absolute value of the rotation offset around the X-axis is greater than or equal to the preset rotation offset threshold around the X-axis, an adjustment command for upper body pitch is generated.

[0123] The aforementioned preset rotational offset threshold around the X-axis can be specifically set according to specific needs and actual application scenarios, and this embodiment does not impose specific limitations on it. For example, the aforementioned preset rotational offset threshold around the X-axis is 2°. The aforementioned upper body tilt adjustment commands include upper body relaxation and upper body straightening adjustments. Whether an upper body relaxation or upper body straightening adjustment command is generated depends on the sign of the rotational offset around the X-axis. When the rotational offset around the X-axis is positive, an upper body straightening adjustment command is generated; when the rotational offset around the X-axis is negative, an upper body relaxation adjustment command is generated. The specific value of the adjustment command can be determined based on the value of the rotational offset around the X-axis.

[0124] Step S226: When the absolute value of the rotational offset around the Z-axis is greater than or equal to the preset rotational offset threshold around the Z-axis, an adjustment command for shoulder leveling is generated.

[0125] The preset rotational offset threshold around the Z-axis can be set according to specific needs and actual application scenarios, and this embodiment does not impose specific limitations on it. For example, the preset rotational offset threshold around the Z-axis is 2°. The above-mentioned shoulder leveling adjustment commands include commands to relax the upper body and commands to straighten the upper body. Whether a command to lower the left shoulder or the right shoulder is generated is determined by the sign of the rotational offset around the Z-axis. When the rotational offset around the Z-axis is positive, a command to lower the left shoulder is generated; when the rotational offset around the Z-axis is negative, a command to lower the right shoulder is generated. The specific value of the adjustment command can be determined based on the value of the rotational offset around the Z-axis.

[0126] Steps S221 to S226 above generate a posture adjustment command corresponding to the posture offset when the posture offset is greater than a preset posture offset threshold. By generating each posture adjustment command, the system guides the user to make posture adjustments, enabling the user to make posture adjustments under guidance. This effectively solves the problem that users cannot judge the direction and degree of posture deviation in home scenarios. It allows users to independently complete accurate posture adjustments based on system guidance even without the presence of professional medical personnel, ensuring the consistency between posture in free breathing and clinically calibrated posture, and providing a spatial benchmark guarantee for the accuracy and effectiveness of subsequent breath-holding training.

[0127] In one embodiment, after step S230, the above-described DIBH training method further includes:

[0128] Step S232: Obtain the second body surface image of the user in a free breathing state after the user's body position adjustment is completed, and determine the target transformation matrix corresponding to the user's adjusted body position based on the preset reference point cloud and the second body surface image; the target transformation matrix is ​​the transformation matrix that adjusts the coordinate system corresponding to the second body surface point cloud in the target area of ​​the second body surface image to the coordinate system corresponding to the preset reference point cloud.

[0129] During steps S210 to S230, the user is guided to actively adjust their body position to resolve large positional deviations at the centimeter level. After completing step S230, if there are still small deviations at the sub-centimeter level, this embodiment determines the target transformation matrix corresponding to the user's adjusted body position. By determining the target transformation matrix, the image captured in the adjusted body position is corrected according to the target transformation body position to achieve millimeter-level coordinate correction and ensure rigid alignment between multi-session training data and the clinical baseline coordinate system.

[0130] In another embodiment, the breathing pattern parameters include the proportion of thoracic breathing and the proportion of abdominal breathing; step S240 involves guiding the user to perform deep inhalation and breath-holding, and determining the user's breathing pattern parameters during breath-holding, including:

[0131] Step S241: Guide the user to take a deep breath and hold it, and capture a third body surface image of the user when the deep breath is completed.

[0132] The determination of the completion of the deep inhalation action described above can be based on the detection that the chest expansion amplitude has reached a stable state and the movement amplitude tends to be stable, thus confirming that the user has completed the deep inhalation action and has begun to enter a breath-holding state. The acquisition of the user's third body surface image at the completion of the deep inhalation action described above can be achieved by using the image acquisition unit of the DIBH training device to acquire the user's third body surface image after determining that the deep inhalation action has been completed.

[0133] Step S242: Extract the third body surface point cloud from the target region of the third body surface image.

[0134] The point cloud of the third body surface in the target region of the above-mentioned extraction of the third body surface image can be a high-density three-dimensional point cloud of the target region obtained after three-dimensional reconstruction and FB-ROI masking of the third body surface image.

[0135] Step S243: Divide the third body surface point cloud into different regions according to its location; the different regions include the upper region, the middle region and the lower region.

[0136] Step S244: The quotient of the average value of each point cloud in the upper region along the Z-axis and the average value of each point cloud in the middle region along the Z-axis is determined as the proportion of chest breathing of the user in the breath-holding state; the Z-axis direction is the direction of the user's abdomen and back.

[0137] Step S245: The quotient of the average value of each point cloud in the lower region along the Z-axis and the average value of each point cloud in the middle region along the Z-axis is determined as the proportion of abdominal breathing of the user in the breath-holding state.

[0138] Steps S241 to S245 above determine the user's breathing mode parameters in the breath-holding state. By determining the breathing mode parameters, it is easier to determine the current breathing mode parameters after the user's body position is adjusted, and compare them with the preset baseline breathing mode parameters. This allows it to determine whether the breathing force mode in the current breath-holding state is consistent with the clinical calibration state, and triggers the generation of a breathing mode adjustment command when the deviation exceeds the threshold.

[0139] Furthermore, in one embodiment, after step S242, the method further includes:

[0140] Step S243: Based on the target transformation matrix, the third body surface point cloud is corrected to obtain the corrected third body surface point cloud.

[0141] The process of correcting the third body surface point cloud based on the target transformation matrix to obtain the corrected third body surface point cloud is as follows:

[0142] ;

[0143] in, The corrected third body surface point cloud, For the third body surface point cloud, R opt Let t be the rotation vector corresponding to the target transformation matrix. opt This is the translation vector corresponding to the target transformation matrix.

[0144] Preferably, based on the corrected coordinates, the heatmap mapping matrix of the display unit can be updated synchronously to ensure that in subsequent training, the display area of ​​the heatmap corresponds one-to-one with the anatomical location of the body surface region in the user's current position, without visual misalignment.

[0145] In one embodiment, step S232, determining the target transformation matrix corresponding to the user's adjusted body position based on a preset reference point cloud and a second body surface image, includes:

[0146] Step S2322: Using the rotation offset and translation offset as the initial values ​​for iteration, project the second surface point cloud in the target area of ​​the second surface image onto the coordinate system corresponding to the preset reference point cloud to obtain the projected point cloud, and calculate the root mean square error of the distance between each pair of points between the projected point cloud and the preset reference point cloud.

[0147] In this embodiment, by using rotational offset Translation offset As the initial values ​​for iteration, the second surface point cloud in the target region of the second surface image is projected onto the coordinate system corresponding to the preset reference point cloud to obtain the projected point cloud. The process is as follows:

[0148] ;

[0149] in, The number of iterations, at the initial iteration. , , Let i be the projected coordinates of the i-th marker point in the k-th iteration. Let R(k) be the rotation vector corresponding to the k-th iteration, and t(k) be the translation vector corresponding to the k-th iteration.

[0150] Each point pair between the aforementioned projected point cloud and the preset reference point cloud can be a point pair where the projected point cloud and the preset reference point cloud are located at the same marker point.

[0151] The above calculation calculates the root mean square error of the distance between each pair of points in the projected point cloud and the preset reference point cloud. The process is as follows:

[0152] ;

[0153] in, denoted as the root mean square error of the k-th iteration, in mm, used to measure the degree of matching between the two sets of point clouds. For projected point clouds The nearest neighbor, In the preset reference point cloud The normal vector at that location.

[0154] Step S2324: Using the root mean square error as the objective function for iterative optimization, solve for the new rotation offset and the new translation offset through SVD decomposition.

[0155] The process described above, which uses the root mean square error as the objective function for iterative optimization and solves for the new rotation offset and the new translation offset through SVD decomposition, is as follows:

[0156] First, calculate the third centroid of the second body surface point cloud. The calculation process is as follows:

[0157] ;

[0158] Then, based on the third and second centroids, the covariance matrix H(k) between the second surface point cloud and the preset reference point cloud is calculated. The specific process is as follows:

[0159] ;

[0160] Furthermore, SVD decomposition is performed on the covariance matrix to solve for the optimal rotation matrix increment (solving for the new rotation offset). The process of (with the new translation offset) is as follows:

[0161] ;

[0162] ;

[0163] Where U(k) is the left singular vector matrix corresponding to the k-th iteration. V is the singular value diagonal matrix corresponding to the k-th iteration, and V(k) is the right singular vector matrix corresponding to the k-th iteration.

[0164] Then, update the rotation matrix and translation vector:

[0165] ;

[0166] ;

[0167] in, Let be the rotation matrix corresponding to the (k+1)th iteration. Let be the translation vector corresponding to the (k+1)th iteration. Let be the rotation matrix corresponding to the k-th iteration.

[0168] Step S2326: Using the new rotation offset and the new translation offset as iteration parameters, repeat the process of updating the root mean square error until the root mean square error is less than the preset error threshold, or the root mean square error is less than the preset convergence threshold and the number of iterations reaches the preset number of iterations threshold; the iteration parameters at this time are determined as the target transformation matrix corresponding to the user's adjusted body position.

[0169] The above process of repeatedly updating the root mean square error using new rotation and translation offsets as iterative parameters: if RMSE(k) - RMSE(k+1) > the preset iteration convergence threshold, it indicates that the error is still decreasing significantly, and iteration continues; if RMSE(k) - RMSE(k+1) ≤ the preset iteration convergence threshold, it indicates that the error has converged, and iteration can be terminated early. The root mean square error (RMSE) is the result of the (k+1)th iteration. The preset error threshold can be set according to specific needs and application scenarios; this embodiment does not impose a specific limitation. The preset iteration convergence threshold can be set according to specific needs and application scenarios; this embodiment does not impose a specific limitation. The number of iterations can be set according to specific needs and application scenarios; this embodiment does not impose a specific limitation. For example, the number of iterations can be 30. The preset convergence threshold can be set according to specific needs and application scenarios; this embodiment does not impose a specific limitation. The process of repeatedly updating the RMSE continues until the RMSE is less than the preset error threshold, or the RMSE is less than the preset convergence threshold and the number of iterations reaches the preset iteration number threshold. That is, the RMSE update stops when the RMSE meets any of the preset stopping conditions. The preset stopping conditions may include the RMSE being less than the preset error threshold, or the RMSE being less than the preset convergence threshold and the number of iterations reaching the preset iteration number threshold. When the root mean square error meets any of the preset stopping conditions, the fine registration is successful, and the final optimal transformation matrix and optimal rotation offset are output. With optimal translation offset That is, the rotation vector and translation vector corresponding to the target transformation matrix.

[0170] It should be noted that if the RMSE is still greater than the preset convergence threshold after the number of iterations reaches the preset maximum number of iterations, it indicates that the user's body position has shifted beyond the coarse registration range or the device calibration parameters have shifted. The system will automatically terminate the fine registration and first prompt the user to check whether the device position has moved. If there is no abnormality in the device, the system will return to the coarse registration steps S210 to S230 and guide the user to adjust their body position again.

[0171] Steps S2322 to S2326 above determine the target transformation matrix corresponding to the user's adjusted body position based on the preset reference point cloud and the second body surface image. The determination of the target transformation matrix facilitates the subsequent correction of the third body surface point cloud based on the target transformation matrix, so as to obtain the corrected third body surface point cloud and use the corrected third body surface point cloud to determine the user's breathing mode parameters in the breath-holding state.

[0172] The present embodiment will now be described and illustrated through preferred embodiments.

[0173] Figure 3 This is a flowchart of a preferred embodiment of the DIBH training method provided in this application. It is applied to a DIBH training device, such as... Figure 3 As shown, the DIBH training method includes the following steps:

[0174] Step S301: Obtain the first body surface image of the user in a free breathing state, and extract the first body surface point cloud in the target area of ​​the first body surface image;

[0175] Step S302: Determine the pose offset of the first body surface point cloud relative to the preset reference point cloud; the pose offset includes rotation offset and translation offset.

[0176] Step S303: When the pose offset is greater than the preset pose offset threshold, a body position adjustment command corresponding to the pose offset is generated; the body position adjustment command is used to guide the user to perform rotation and / or translation.

[0177] Step S304: Repeat the process of acquiring the first body surface image after body position adjustment, determining the pose offset and generating a body position adjustment command until the pose offset is less than or equal to the preset pose offset threshold.

[0178] Step S305: Obtain the second body surface image of the user in a free breathing state after the user's body position adjustment is completed, and determine the target transformation matrix corresponding to the user's adjusted body position based on the preset reference point cloud and the second body surface image; the target transformation matrix is ​​the transformation matrix that adjusts the coordinate system corresponding to the second body surface point cloud in the target area of ​​the second body surface image to the coordinate system corresponding to the preset reference point cloud.

[0179] Step S306: Guide the user to take a deep breath and hold it, and capture a third-party body surface image of the user when the deep breath is completed.

[0180] Step S307: Extract the third surface point cloud from the target region of the third surface image;

[0181] Step S308: Based on the target transformation matrix, the point cloud of the third body surface is corrected to obtain the corrected point cloud of the third body surface.

[0182] Step S309: Determine the user's breathing pattern parameters in the breath-holding state based on the corrected third body surface point cloud.

[0183] Step S310: When the difference between the user's breathing mode parameters in the breath-holding state and the preset breathing mode baseline parameters is greater than the preset breathing mode difference threshold, a breathing mode adjustment command is generated.

[0184] Step S311: Repeat the process of acquiring the third body surface image after the breathing mode adjustment, determining the breathing mode parameters and generating the breathing mode adjustment command until the difference between the user's breathing mode parameters in the breath-holding state and the preset breathing mode baseline parameters is less than or equal to the preset breathing mode difference threshold; the breathing mode adjustment command is used to guide the user to perform chest breathing and / or abdominal breathing.

[0185] Steps S301 to S311 above determine the pose offset of the user's body surface point cloud relative to a preset reference point cloud during free breathing. When the pose offset is greater than a preset pose offset threshold, a posture adjustment command corresponding to the pose offset is generated to guide the user to rotate and / or translate. The process of generating posture adjustment commands to guide the user to adjust their posture is repeated until the pose offset is less than or equal to the preset pose offset threshold. By adjusting the rotation and / or translation during free breathing, the rotation angle and position of the user's body are ensured to be consistent with the posture corresponding to the preset reference point cloud. In turn, the user is guided to take a deep breath and hold their breath, and the breathing mode parameters of the user during breath-holding are determined. Furthermore, when the difference between the user's breathing pattern parameters during breath-holding and the preset breathing pattern baseline parameters exceeds a preset breathing pattern difference threshold, a breathing pattern adjustment command is generated to guide the user to perform thoracic and / or abdominal breathing. This method adjusts the breathing pattern parameters, and by adjusting the proportion of thoracic and abdominal breathing, the diaphragm's downward displacement and chest deformation during breath-holding are further adjusted. This ensures that the user's posture during breath-holding is consistent with the posture corresponding to the baseline point cloud, guaranteeing consistency between the user's posture during breath-holding and during clinical treatment, as well as consistency during multiple training sessions. This solves the problem of poor training effect caused by the inability of existing DIBH training methods in home training scenarios to guarantee the consistency of posture space during breath-holding between multiple training sessions and between training and clinical treatment.

[0186] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0187] Based on the same inventive concept, this embodiment also provides a DIBH training device for implementing the above embodiments and preferred embodiments, which will not be repeated as already described. The terms "module," "unit," "subunit," etc., used below can refer to combinations of software and / or hardware that implement a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0188] In one embodiment, Figure 4 This is a structural block diagram of a DIBH training device provided in one embodiment of this application, as shown below. Figure 4 As shown, the DIBH training device, applied to DIBH training equipment, includes:

[0189] The offset determination module 41 is used to determine the pose offset of the user's body surface point cloud relative to the preset reference point cloud in the free breathing state; the pose offset includes rotation offset and translation offset.

[0190] The body position adjustment instruction generation module 42 is used to generate a body position adjustment instruction corresponding to the pose offset when the pose offset is greater than a preset pose offset threshold; the body position adjustment instruction is used to guide the user to perform rotation and / or translation.

[0191] The body position adjustment module 43 is used to repeatedly acquire the pose offset of the user's body surface point cloud in the free breathing state after body position adjustment relative to the preset reference point cloud, and generate a body position adjustment command until the pose offset is less than or equal to the preset pose offset threshold.

[0192] The breathing mode parameter determination module 44 is used to guide the user to perform deep inhalation and breath-holding, and to determine the user's breathing mode parameters in the breath-holding state.

[0193] The breathing mode adjustment instruction generation module 45 is used to generate a breathing mode adjustment instruction when the difference between the user's breathing mode parameters in the breath-holding state and the preset breathing mode baseline parameters is greater than the preset breathing mode difference threshold; the breathing mode adjustment instruction is used to guide the user to perform chest breathing and / or abdominal breathing.

[0194] The breathing mode adjustment module 46 is used to repeatedly acquire the user's breathing mode parameters in the breath-holding state after the breathing mode adjustment, and generate a breathing mode adjustment command, until the difference between the user's breathing mode parameters in the breath-holding state and the preset breathing mode reference parameters is less than or equal to the preset breathing mode difference threshold.

[0195] The aforementioned DIBH training device determines the pose offset of the user's body surface point cloud relative to a preset reference point cloud during free breathing. When the pose offset exceeds a preset pose offset threshold, it generates a posture adjustment command corresponding to the pose offset to guide the user to rotate and / or translate. This process of generating posture adjustment commands to guide the user to adjust their posture continues until the pose offset is less than or equal to the preset pose offset threshold. By adjusting the rotation and / or translation during free breathing, it ensures that the user's body rotation angle and position are consistent with the posture corresponding to the preset reference point cloud. Furthermore, it guides the user to take a deep breath and hold their breath, and determines the user's breathing mode parameters during breath-holding. Furthermore, when the difference between the user's breathing pattern parameters during breath-holding and the preset breathing pattern baseline parameters exceeds a preset breathing pattern difference threshold, a breathing pattern adjustment command is generated to guide the user to perform thoracic and / or abdominal breathing. This method adjusts the breathing pattern parameters, and by adjusting the proportion of thoracic and abdominal breathing, the diaphragm's downward displacement and chest deformation during breath-holding are further adjusted. This ensures that the user's posture during breath-holding is consistent with the posture corresponding to the baseline point cloud, guaranteeing consistency between the user's posture during breath-holding and during clinical treatment, as well as consistency during multiple training sessions. This solves the problem of poor training effect caused by the inability of existing DIBH training methods in home training scenarios to guarantee the consistency of posture space during breath-holding between multiple training sessions and between training and clinical treatment.

[0196] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can reside in the same processor; or the above modules can be located in different processors in any combination.

[0197] In one embodiment, a DIBH training device is provided, including a memory, a processor, and a blood oxygen monitoring unit. The memory stores a computer program, and the processor executes the computer program to implement any of the DIBH training methods described in the above embodiments. The blood oxygen monitoring unit is used to monitor the user's blood oxygen saturation.

[0198] The aforementioned blood oxygen monitoring unit monitors the user's blood oxygen saturation throughout the training process, preventing hypoxia due to overtraining and ensuring training safety. Simultaneously, the monitoring data is displayed in real-time on the sidebar of the display unit. The blood oxygen monitoring unit collects patient blood oxygen data at a preset blood oxygen saturation sampling rate, aligning it with the visual-motor data timestamps. Abnormal states immediately trigger corresponding intervention rules; these include tiered safety intervention rules: Safe state: SpO2 ≥ preset Level 1 warning blood oxygen threshold, normal training, no additional intervention; Level 1 warning: SpO2 is between the preset Level 2 warning blood oxygen threshold and the Level 1 warning blood oxygen threshold, and continues to reach the doctor's preset abnormal blood oxygen duration. A yellow prompt box pops up on the display unit's interface, and simultaneous voice guidance leads the user to slow down breathing, smoothly end the breath-holding phase, and restore FB (Free Breathing). Breath (free breathing); Level 2 warning: If SpO2 falls between the doctor's preset mandatory termination blood oxygen threshold and the Level 2 warning blood oxygen threshold, and continues to reach the doctor's preset abnormal blood oxygen duration, the training timer will be immediately paused, the interface will display a full-screen red highlight, and the patient will be forced to resume free deep breathing via voice guidance. Training can only be restarted after blood oxygen continuously reaches the doctor's preset blood oxygen restart training stabilization time is greater than or equal to the doctor's preset Level 1 warning blood oxygen threshold. Breath-holding completed under these conditions is considered invalid. Mandatory termination: If SpO2 falls below the doctor's preset mandatory termination blood oxygen threshold and continues to reach the doctor's preset abnormal blood oxygen duration, the entire training session will be immediately terminated, an emergency safety prompt will pop up, and the user will be instructed to maintain a sitting or semi-reclining position for deep breathing. If blood oxygen continues to reach the doctor's preset blood oxygen recovery observation time and does not recover to a level greater than or equal to the doctor's preset blood oxygen recovery target threshold, the user will be prompted to contact medical personnel. The SpO2 mentioned above refers to the degree to which oxygen in arterial blood is bound to hemoglobin, that is, the percentage of oxygen-carrying hemoglobin in the total hemoglobin in the blood.

[0199] In another embodiment, the training is terminated when any of the following conditions are met: the number of training sets and duration preset by the doctor are completed; the doctor or patient manually triggers the termination command; SpO2 is less than the doctor's preset mandatory termination blood oxygen threshold and continues to reach the doctor's preset duration of abnormal blood oxygen, thus forcibly terminating; or the user experiences severe discomfort and triggers the emergency stop button.

[0200] In one embodiment, data archiving and report output are performed. The specific output content may include one or more of the following: basic information of this training (date, time, number of training groups), all benchmark parameters preset by the doctor (FB-ROI range, DIBH-ROI range, personalized breathing mode benchmark parameters, deviation threshold, upper and lower limits of breath-holding duration), body position correction results (coarse registration deviation, fine registration RMSE), effective duration of breath-holding for each group, DIBH-ROI compliance rate, average DIBH-ROI deviation, respiratory mode parameter statistics (deviation of adjustable parameters from benchmark values, compliance status), statistics on the number of deviation corrections and correction duration, blood oxygen saturation change curve, training compliance status, etc.

[0201] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0202] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0203] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A DIBH training method, applied to a DIBH training device, characterized in that, The method includes: Determine the pose offset of the user's body surface point cloud relative to a preset reference point cloud during free breathing; the pose offset includes rotation offset and translation offset. When the pose offset is greater than a preset pose offset threshold, a body position adjustment command corresponding to the pose offset is generated; the body position adjustment command is used to guide the user to perform rotation and / or translation. The process of repeatedly acquiring the pose offset of the user's body surface point cloud relative to the preset reference point cloud in a free breathing state after the body position adjustment, and generating a body position adjustment command, continues until the pose offset is less than or equal to the preset pose offset threshold. Instruct users to take a deep breath and hold it, and determine the user's breathing pattern parameters while holding their breath; When the difference between the user's breathing mode parameters in a breath-holding state and the preset breathing mode baseline parameters is greater than the preset breathing mode difference threshold, a breathing mode adjustment command is generated; the breathing mode adjustment command is used to guide the user to perform chest breathing and / or abdominal breathing. The process of repeatedly acquiring the user's breathing mode parameters under breath-holding state after the breathing mode adjustment and generating breathing mode adjustment instructions continues until the difference between the user's breathing mode parameters under breath-holding state and the preset breathing mode baseline parameters is less than or equal to the preset breathing mode difference threshold.

2. The DIBH training method according to claim 1, characterized in that, Determining the pose offset of the user's surface point cloud relative to a preset reference point cloud during free breathing includes: Acquire the first body surface image of the user in a free breathing state; Extract the first body surface point cloud from the target region of the first body surface image; Determine the pose offset of the first body surface point cloud relative to the preset reference point cloud.

3. The DIBH training method according to claim 2, characterized in that, Determining the pose offset of the first body surface point cloud relative to the preset reference point cloud includes: Calculate the first centroid of the first body surface point cloud and the second centroid of the preset reference point cloud; Based on the first centroid and the second centroid, calculate the covariance matrix between the first body surface point cloud and the preset reference point cloud; The covariance matrix is ​​solved by SVD to determine the pose offset of the first body surface point cloud relative to the preset reference point cloud.

4. The DIBH training method according to claim 1, characterized in that, The translational offset includes X-axis offset, Y-axis offset, and Z-axis offset; the rotational offset includes rotational offset around the X-axis, rotational offset around the Y-axis, and rotational offset around the Z-axis. The preset pose offset thresholds include preset X-axis offset thresholds, preset Y-axis offset thresholds, preset Z-axis offset thresholds, preset rotation offset thresholds around the X-axis, preset rotation offset thresholds around the Y-axis, and preset rotation offset thresholds around the Z-axis; when the pose offset is greater than the preset pose offset threshold, a body position adjustment command corresponding to the pose offset is generated, including: When the absolute value of the X-axis offset is greater than or equal to the preset X-axis offset threshold, an adjustment command to move the body left or right is generated. When the absolute value of the Y-axis offset is greater than or equal to the preset Y-axis offset threshold, an adjustment command to move the body up and down is generated. When the absolute value of the Z-axis offset is greater than or equal to the preset Z-axis offset threshold, an adjustment command to lift or relax the chest is generated. When the absolute value of the rotation offset around the Y-axis is greater than or equal to the preset rotation offset threshold around the Y-axis, a turning adjustment command is generated. When the absolute value of the rotation offset around the X-axis is greater than or equal to the preset rotation offset threshold around the X-axis, an upper body pitch adjustment command is generated. And / or, when the absolute value of the rotational offset around the Z-axis is greater than or equal to a preset rotational offset threshold around the Z-axis, an adjustment command for shoulder leveling is generated.

5. The DIBH training method according to claim 1, characterized in that, After repeatedly acquiring the pose offset of the user's surface point cloud relative to a preset reference point cloud in a free breathing state after posture adjustment, and generating a posture adjustment command, until the pose offset is less than or equal to the preset pose offset threshold, the method further includes: A second body surface image is acquired after the user has adjusted their posture and is in a free breathing state. Based on a preset reference point cloud and the second body surface image, a target transformation matrix corresponding to the user's adjusted posture is determined. The target transformation matrix is ​​a transformation matrix that adjusts the coordinate system corresponding to the second body surface point cloud in the target area of ​​the second body surface image to the coordinate system corresponding to the preset reference point cloud.

6. The DIBH training method according to claim 5, characterized in that, The breathing pattern parameters include the proportion of thoracic breathing and the proportion of abdominal breathing; The process of guiding the user to perform deep inhalation and breath-holding, and determining the user's breathing pattern parameters during breath-holding, includes: Instruct users to take a deep breath and hold it, and capture a third-party image of the user's body surface when the deep breath is completed; Extract the third surface point cloud from the target region of the third surface image; The third body surface point cloud is divided into different regions according to its location; the different regions include an upper region, a middle region, and a lower region; The quotient of the average value of each point cloud in the upper region along the Z-axis and the average value of each point cloud in the middle region along the Z-axis is determined as the proportion of chest breathing in the user's breath-holding state; the Z-axis direction is the user's abdominal and back direction. The quotient of the average value of each point cloud in the lower region along the Z-axis and the average value of each point cloud in the middle region along the Z-axis is determined as the proportion of abdominal breathing of the user in the breath-holding state.

7. The DIBH training method according to claim 6, characterized in that, After extracting the third surface point cloud from the target region of the third surface image, the process further includes: Based on the target transformation matrix, the point cloud of the third body surface is corrected to obtain the corrected point cloud of the third body surface.

8. The DIBH training method according to claim 5, characterized in that, The step of determining the target transformation matrix corresponding to the user's adjusted body position based on the preset reference point cloud and the second body surface image includes: Using the rotation offset and the translation offset as the initial values ​​for iteration, the second body surface point cloud in the target area of ​​the second body surface image is projected onto the coordinate system corresponding to the preset reference point cloud to obtain the projected point cloud, and the root mean square error of the distance between each pair of points between the projected point cloud and the preset reference point cloud is calculated. The root mean square error is used as the objective function for iterative optimization, and the new rotation offset and the new translation offset are solved by SVD decomposition. Using the new rotation offset and the new translation offset as iteration parameters, the process of updating the root mean square error is repeated until the root mean square error is less than the preset error threshold, or the root mean square error is less than the preset convergence threshold and the number of iterations reaches the preset number of iterations threshold; the iteration parameters at this time are determined as the target transformation matrix corresponding to the user's adjusted body position.

9. A DIBH training device, applied to DIBH training equipment, characterized in that, The device includes: The offset determination module is used to determine the pose offset of the user's body surface point cloud relative to a preset reference point cloud in a free breathing state; the pose offset includes rotation offset and translation offset. The body position adjustment instruction generation module is used to generate a body position adjustment instruction corresponding to the pose offset when the pose offset is greater than a preset pose offset threshold; the body position adjustment instruction is used to guide the user to perform rotation and / or translation. The body position adjustment module is used to repeatedly acquire the pose offset of the user's body surface point cloud in a free breathing state after body position adjustment relative to a preset reference point cloud, and generate a body position adjustment command until the pose offset is less than or equal to the preset pose offset threshold. The breathing mode parameter determination module is used to guide users to perform deep inhalation and breath-holding, and to determine the user's breathing mode parameters during breath-holding. The breathing mode adjustment instruction generation module is used to generate a breathing mode adjustment instruction when the difference between the user's breathing mode parameters in a breath-holding state and the preset breathing mode baseline parameters is greater than a preset breathing mode difference threshold; the breathing mode adjustment instruction is used to guide the user to perform chest breathing and / or abdominal breathing. The system also includes a breathing mode adjustment module, which repeatedly acquires the user's breathing mode parameters in the breath-holding state after the breathing mode adjustment and generates a breathing mode adjustment command until the difference between the user's breathing mode parameters in the breath-holding state and the preset breathing mode baseline parameters is less than or equal to the preset breathing mode difference threshold.

10. A DIBH training device, comprising a memory, a processor, and a blood oxygen monitoring unit, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the DIBH training method according to any one of claims 1 to 8; the blood oxygen monitoring unit is used to monitor the user's blood oxygen saturation.

Images