Weight Estimation

Abstract

A computer-implemented method is provided for estimating a patient weight when supported on a patient table. Optical image data and depth image data are acquired, and body key points of the patient are extracted from the optical image data. A first frame is selected that includes an image of the patient table by selecting a frame in which no body key points are present in the optical image data. A second frame is selected of the patient on the table by selecting a frame in which there are little or no movement, having key points of the patient's body. A patient volume is acquired based on the difference between the depth image data of the first and second frames (or the depth data at those times), and the patient weight is estimated from the determined patient volume.

Description

[Technical field]

[0001] The present invention relates to estimating the weight of a particular subject, such as a patient on a patient table undergoing a medical imaging procedure. [Background technology]

[0002] Patient weight (ie, total patient weight) is a critical parameter required for most medical imaging modalities.

[0003] In clinical routine, patient metrics such as age, height, and weight are routinely recorded prior to medical imaging examinations such as magnetic resonance imaging, MRI, computed tomography, CT, imaging, and positron emission tomography, PET, imaging. For each modality, patient weight is a critical quantity. In MRI, a safety-related specific absorption rate (SAR) calculation using patient weight as an input parameter is performed prior to each scan to prevent excessive radio frequency, RF, exposure. In CT examinations, weight-based tube currents can be used to minimize radiation dose. Furthermore, in all modalities, the determination of the appropriate contrast volume requires accurate knowledge of the patient's weight. Summary of the Invention [Problem to be solved by the invention]

[0004] One common approach is to ask the patient about their weight. Another is to estimate the patient's weight simply based on the patient's visual appearance. However, these approaches can introduce substantial errors.

[0005] Although using scale to determine weight provides accurate results, it introduces undesirable complexity into the workflow. Furthermore, depending on the patient's medical condition, not all patients can be interviewed or positioned at scale.

[0006] Non-contact weight measurement systems have been proposed. For example, EP 3699929 discloses a system for estimating a patient's weight before a medical examination using a depth camera. A patient model in the form of a mesh is fitted to the depth image and weight is estimated from extracted feature values ​​derived from the patient model using regression analysis. [Means for solving the problem]

[0007] The invention is defined by the claims.

[0008] According to an example according to one aspect of the present invention, 1. A computer-implemented method for estimating a patient's weight when supported on a patient table, comprising: receiving image data comprising optical image data of an optical image of the patient and depth image data of a depth image of the patient, the image data comprising a frame; detecting patient body keypoints from the optical image data; selecting a first frame having an image of the patient table by selecting a frame in which the patient body keypoints are not present in the optical image data; selecting a second frame having an image of a patient on the table by selecting a frame, a number of patient body keypoints present in the optical image data is greater than a first threshold; a movement measurement of a patient body keypoint in the optical image data between the frame and at least one previous frame is below a second threshold; Steps and determining a patient volume comprising an entire patient volume based on a difference between depth image data at a time of the first frame and depth image data at a time of the second frame, the difference defining a patient volume map; estimating a patient weight from the determined patient volume, the patient weight comprising a total patient weight; A method is provided comprising:

[0009] This method provides a non-contact patient weight estimation method that can be seamlessly integrated into clinical workflow. The method uses an optical imaging camera (e.g., RGB image sensor) and a depth camera that provides a video stream of the patient table area during exam preparation.

[0010] The difference between the depth image data at the time of the first frame and the depth image data at the time of the second frame defines a patient volume map, i.e., depth information for the patient's body alone. The estimated patient weight is the complete weight of the patient, i.e., the weight of the entire patient.

[0011] The present invention allows for a fully automatic and robust estimation of patient weight without requiring manual selection of input camera data. Instead, the optimal frame for determining the patient's volume and therefore weight is automatically selected, which can be done during patient preparation.

[0012] The camera system, e.g., a ceiling-mounted RGB camera and a depth camera, allows for non-contact weight determination. The weight estimate can be used as supplemental information to biometric data available via the electronic patient record, e.g., to confirm information in the electronic patient record or to indicate possible discrepancies.

[0013] Detecting the patient's body keypoints may, for example, utilize neural networks. Algorithms for the analysis of image data to extract body keypoints are well known.

[0014] The method may include identifying a second set of frames including all frames in which a number of patient body key points present is greater than a first threshold and a patient body key point movement measure between the frame and at least one previous frame is less than a second threshold, and determining a patient volume for each of the frames in the second set.

[0015] In this manner, multiple volume determinations may be made so that the best one may be selected for use in estimating patient weight, for example, estimating patient weight may be based on the lowest patient volume determined.

[0016] Determining the patient volume may include, for example, deriving a volume map from the depth data.

[0017] In the most basic approach, a typical patient density may be assumed to be associated with a particular type (age, sex, etc.) of patient (i.e., patient category) for volume to weight conversion. However, patient weight may instead be derived more accurately by using an anatomical model of typical tissue distribution fitted to the patient volume map. Patient weight may be estimated, for example, by using a neural network with one or more volume maps as input.

[0018] Therefore, estimating patient weight is Typical patient density, or the typical patient density for a particular category of patients, or Typical tissue distribution fitted to patient volume map This may involve converting the determined patient volume to patient weight using

[0019] The method can include receiving image data for a fixed patient table position, such that depth data of any second frame can be combined with depth data of any first frame.

[0020] Alternatively, image data may be received for different patient table positions, and the method includes determining the table position from the image data and calibrating the depth data using the determined table position, such that corrections can be made when different frames contain data for different table positions.

[0021] When image data is received for a different patient table position, the method may instead include selecting first and second frames that correspond to the same table position, e.g., for a new table position, a new first frame may be required.

[0022] Determining the patient volume includes, for example, measuring the volume of a portion of the patient's body using the difference between depth image data and extrapolating from the determined patient volume to an estimated patient weight. The volume estimation may be based only on a central body part, such as, for example, the trunk, since the limbs and head are more likely to give inaccurate results.

[0023] We also provide a computer program comprising computer program code means adapted to carry out the method defined above, when said program is run on a computer.

[0024] The invention also provides a controller programmed with a computer program as defined above.

[0025] The present invention also provides An optical camera; A depth camera and the controller for processing image data, including optical image data from the optical camera and depth image data from the depth camera, to estimate a patient's weight; The present invention provides a weight estimation system including:

[0026] The optical and depth cameras are, for example, ceiling mounted units.

[0027] The present invention also provides A patient support section; A medical imaging unit; The body weight estimation system; The present invention provides a medical imaging system including:

[0028] The medical imaging unit comprises for example a CT scanner, an MRI scanner, or a PET scanner.

[0029] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter.

[0030] For a better understanding of the invention and to show more clearly how it may be carried into effect, reference will now be made, by way of example only, to the accompanying drawings in which: [Brief description of the drawings]

[0031] [Figure 1] A method for estimating patient weight is shown. [Diagram 2] 1 shows an image-based representation of the method. [Diagram 3] Various image patient measurements are shown to illustrate how the method works. [Figure 4] 1 illustrates an example of an imaging system that can be used to provide images for analysis using the methods of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0032] The present invention will now be described with reference to the drawings.

[0033] It should be understood that the detailed description and specific examples, while indicating exemplary embodiments of the devices, systems, and methods, are for purposes of illustration only and are not intended to limit the scope of the invention. These and other features, aspects, and advantages of the devices, systems, and methods of the present invention will become better understood from the following description, appended claims, and accompanying drawings. It should be understood that the figures are merely schematic and are not drawn to scale. It should also be understood that the same reference numerals are used throughout the figures to denote the same or similar parts.

[0034] The present invention provides a computer-implemented method for estimating a patient's weight (i.e., their total body weight) when supported on a patient table. Optical image data and depth image data are acquired and patient body keypoints are extracted from the optical image data. A first frame is selected that includes an image of the patient table by selecting a frame in which no patient body keypoints are present in the optical image data. A second frame is selected of the patient on the table by selecting a frame in which there are no patient body keypoints present in the optical image data. A patient volume is obtained based on the difference between the depth image data of the first and second frames (or more generally the difference between depth images taken at approximate or exact times of these first and second frames) and the patient weight is estimated from the determined patient volume.

[0035] FIG. 1 illustrates the method 10.

[0036] Image data is received, step 20, including optical image data for an optical image of the patient and depth image data for a depth image of the patient. The image data is obtained, for example, from a ceiling mounted camera system.

[0037] The image data includes frames of optical data and frames of depth data. The frames may be identical, such that each frame includes optical and depth data. However, this is not required and the frame structure (e.g., frame rate and timing) may be different for the optical and depth data. Thus, separate cameras may be used.

[0038] The optical image data is, for example, 2D RGB camera image data.

[0039] Depth images can be obtained by various possible camera designs. One example is a time-of-flight camera, where the time of flight to an object and back is measured. Alternatively, a stereo vision camera may be used, based on a pair of regular optical cameras mounted a distance apart. Structured light cameras are also known for 3D imaging, where the scene is illuminated with specially designed light patterns.

[0040] With a depth camera, either of these options are possible.

[0041] Image data is received over a period covering initial patient preparation, ie, as the patient arrives on the patient table of the imaging system.

[0042] At step 22, body keypoints of the patient are detected in the optical image data using known algorithms. The patient's body keypoints are anatomical landmarks such as body joints or other anatomical features.

[0043] In step 24, a first frame containing an image of the patient table is selected by selecting a frame in which there are no patient body keypoints present in the optical image data. This frame may be considered to be an empty frame.

[0044] This first empty frame is selected to show a completely empty patient table, which is usually possible between two image examinations. Fully automatic selection of this frame may be made possible using a dedicated neural network for body keypoint detection in step 22, which is applied to each frame (i.e., RGB data) of the camera optical image data stream. The neural network is "activated" when a body keypoint is detected, and thus the level of neural network activation may represent the number of keypoints detected.

[0045] By way of example, the set of keypoints may include 13 keypoints including the 12 major joints (ankle, knee, hip, shoulder, elbow, wrist) and the head. By way of example, a "heat map" of these keypoints may be generated, e.g., a 2D image with Gaussian-like peaks when keypoints are detected.

[0046] To ensure that neither the operator nor the patient occludes the table, a threshold criterion is applied to all activations of the network, for example to ensure that there are no body keypoints in the image.

[0047] In one embodiment, the remaining steps of the method rely on the assumption of the same patient table height, so that any table movement recorded by the imaging system can be used to trigger an updated selection of this first empty frame.

[0048] Alternatively, detection of patient table keypoints can be used to automatically infer the table position and correct for possible table motion by applying corresponding parameters (longitudinal and vertical translations) to the depth data, e.g. in combination with a camera calibration matrix.

[0049] In step 26, a second frame is selected that includes an image of the patient on the table, preferably showing the patient lying on the table prior to placement of the coil or other accessory.

[0050] A second frame is selected in which the number of patient body keypoints present in the optical image data is greater than the first threshold. To perform this test of patient body keypoints, the total activation of a keypoint detection neural network in the image region of the table may be obtained for each frame of the optical image data stream. A minimum total activation is required, corresponding to the detection of this minimum number of keypoints. This ensures that the patient is truly positioned on the table. It has been found that activations caused by the operator during examination preparation do not result in a false positive trigger of this criterion.

[0051] The selection of the second frame also includes performing a motion measurement of the patient's body keypoints in the optical image data between the frame and at least one previous frame, such as between two successive frames of the optical image data, which motion measurement must be below a second threshold.

[0052] In this way, frames with high patient motion are not selected as the second frame.The motion measure is derived as a body keypoint motion metric, which is calculated by measuring the total frame-to-frame difference in body keypoint activations.

[0053] In step 28, a patient volume is obtained based on the difference between the depth image data of the first frame (or more generally, the depth image data as close as possible to the timing of the first frame) and the depth image data of the second frame (or more generally, the depth image data as close as possible to the timing of the second frame).

[0054] In the most basic approach, only one primary empty frame is selected and only one secondary patient frame is selected and then the volume is acquired.

[0055] In a preferred approach, for every second frame that satisfies the two previous criteria, a patient volume map is calculated by subtracting the corresponding depth image from the depth image of the first, empty frame.

[0056] The volume estimate is obtained, for example, by the sum of all pixel values ​​in the patient volume map. Then, one most suitable second frame, i.e. one most suitable volume calculation, can be selected. For example, the frame corresponding to the smallest volume estimate may be selected as the second patient frame. The rationale for this is based on the observation that depth camera-based volume estimation is compromised in case of elevated body parts, which frequently occurs during the early stages of patient preparation, or in case of shielding by accessories such as MR coils or the operator.

[0057] If the frames of optical image data and depth image data are the same (i.e., there is a single camera system with a single frame timing), there is a direct correspondence between the frames identified by the optical analysis and the depth data used. However, if there is no direct correspondence between the frames of optical image data and the frames of depth image data, time values ​​can be identified for the frames selected as the first and second frames based on the analysis of key points, and then the frame of depth image data that most closely matches the first and second frames selected in time can be selected. Thus, the patient volume is based on the difference between the depth image data at (or closest to) the timing of the first frame and the depth image data at (or closest to) the timing of the second frame.

[0058] In step 30, patient weight is estimated from the determined patient volume. Patient weight may be estimated using a simple regression model. Calibration of the model may be performed by applying the algorithm to a range of subjects with known weights.

[0059] Patient weight can be estimated using anatomical models of typical tissue distribution that are fitted to the patient volume map. These anatomical models can be based on the same body keypoints used in the method. The analytical models take into account the relative densities of different tissues.

[0060] Patient weight can be estimated using a neural network using one or several patient volume maps as input. If reliable patient weight data is available at the clinical site (e.g., using a scale), such a system can be trained and even improved after deployment.

[0061] Figure 2 shows an image-based representation of this method.

[0062] Image 40 represents an RGB camera image of the patient table for the first selected frame, and image 42 represents a depth image of the patient table for the first selected frame.

[0063] Image 44 represents an RGB camera image of the patient table and the patient for the selected second frame, and image 46 represents a depth image of the patient table and the patient for the selected second frame.

[0064] Image 48 represents a depth map based on subtraction of image 46 from image 42 (or vice versa).

[0065] FIG. 3 shows various patient measurements to illustrate how the method works.

[0066] FIG. 3A shows volume estimate (y-axis) versus frame number, where a volume estimate is made from each frame by comparing the depth image data with the depth image data of the first frame.

[0067] The first frame is acquired first before there is any interaction with the patient table, for example the first frame selected as a free frame may be frame number 1.

[0068] The first non-zero volume estimate is made at frame 35, as shown by line 50. However, at this point the patient has just begun to climb onto the patient table, so it is not a suitable frame to use as the second patient frame.

[0069] After frame 35, the patient has moved a great deal while getting into position on the patient table.

[0070] FIG. 3B shows the activation level of the keypoint detection algorithm, which represents the number of keypoints detected, as an arbitrary scale indicating the number of keypoints detected.

[0071] The minimum neural network activation level is denoted as the first threshold 60 (which thus corresponds to the number of patient body keypoints present in the optical image data). If only this criterion is used, frame 40, for example, will be erroneously selected as a suitable second patient frame, which is still in the stage where the patient is on the table.

[0072] 3C shows a motion measure based on the number of patient body keypoints in the optical image data that have moved (by an amount greater than the motion threshold) between frames. A second threshold (of the number of moved keypoints) is shown as 62.

[0073] By using additional patient motion criteria, the frame indicated by box 64 is eliminated, leading to the selection of an appropriate frame, such as frame 95 as indicated by line 52 in FIG. 3A.

[0074] This frame is the one with the smallest volume calculation in FIG. 3A among the frames that meet the two criteria for selection as the second frame.

[0075] To improve accuracy, especially if the optical axis of the camera system is not orthogonal to the patient table plane, the depth data can first be transformed to 3D scanner coordinates using a camera calibration matrix, and the subtraction between the patient depth data and the patient table depth data can be performed in this Cartesian coordinate system.

[0076] Instead of relying purely on RGB data for automatic camera frame selection, other sensors such as infrared may be used.

[0077] Volume estimation can only be done for central body parts such as tree trunks, since the limbs and head tend to be elevated during preparation, leading to inaccurate results. A regression model can then be calculated between this subvolume and the total patient weight.

[0078] Hygiene checks can be performed to identify cases where the algorithm may have produced inaccurate results, for example when the estimated patient weight is highly abnormal for the detected patient dimensions (very high / low body mass index). In particular, a quality check can be performed using biometric data available in electronic patient records or previous examinations, and significant mismatches can be communicated to the operator.

[0079] Additional standard image filters may be used to smooth or correct the depth image or patient volume map.

[0080] FIG. 4 shows an example of an imaging system that can be used to provide images for analysis using the methods of the present invention.

[0081] The imaging device 100 in this figure is an X-ray computed tomography (CT) scanner.

[0082] The imaging apparatus 100 generally includes a stationary gantry 102 and a rotating gantry 104. The rotating gantry 104 is rotatably supported by the stationary gantry 102 and rotates about an examination region about a longitudinal, axial, or z-axis.

[0083] A patient table 120, such as a treatment couch, supports an object or body, such as a human patient, within an examination region. The support 120 is configured to move the object or body in order to load, scan, and / or remove the object or body. The support 120 is movable along an axial direction, i.e., along the direction of the z-axis or longitudinal axis. Moving the support changes the axial position of the rotating gantry relative to the support (and thus relative to the body supported by the support).

[0084] A radiation source 108, such as an x-ray tube, is rotatably supported by the rotating gantry 104. The radiation source 108 rotates with the rotating gantry 104 and emits radiation that traverses an examination region 106.

[0085] A radiation sensitive detector array 110 subtends an angular arc opposite the radiation source 108 across the examination region 106. The detector array 110 extends along the z-axis direction and includes one or more rows of detectors that detect radiation traversing the examination region 106 and generate projection data indicative thereof.

[0086] Rotation of the gantry 104 changes the angular or rotational position of the scanner relative to the subject, and movement of the support along the z-axis changes the axial position of the scanner relative to the subject.

[0087] A typical scan is pre-configured using a scan protocol. The scan protocol includes multiple scan parameters. The scan parameters define, among other things, the spatial extent of the scan relative to the axial and rotational axes of the scanner. For example, the scan parameters may include boundaries (i.e., start and end points) of the scan extent along one or more axes of the imaging device, e.g., the rotational and / or axis. The scan extent defines the field of view (FOV) from which imaging data is acquired during the scan. The scan parameters may also typically include several other parameters, including, for example, tube current, tube voltage, scan spatial resolution, scan temporal resolution, and / or fan angle. The resolution parameters may be defined by the rotational speed of the gantry 104 and the axial movement speed of the support 120 through the gantry.

[0088] A general-purpose computing system or computer serves as an operator console 112 and includes input devices 114, such as a mouse, keyboard, etc., and output devices 116, such as a display monitor. The console, the input devices, and the output devices form a user interface 130. The console 112 allows an operator to control the operation of the system 100.

[0089] A reconstruction unit 118 processes the projection data to reconstruct volumetric image data, which may be displayed through one or more display monitors of output unit(s) 116.

[0090] The reconstruction unit 118 may use filtered backprojection (FBP) reconstruction, (image domain and / or projection domain) reduced noise reconstruction algorithms (e.g., iterative reconstruction), and / or other algorithms. It should be understood that the reconstruction unit 118 may be implemented via a microprocessor executing computer-readable instructions encoded or embedded on a computer-readable storage medium, such as a physical memory and other non-transitory medium. Additionally or alternatively, the microprocessor may execute computer-readable instructions carried by a carrier wave, signal, and other transitory (or non-transitory) medium.

[0091] The reconstruction device 118 may incorporate a processor programmed with a computer program for carrying out the above-mentioned methods for analyzing the generated 3D CT scan images and thereby performing an analysis of collateral vessels within the region of interest.

[0092] A ceiling mounted camera system is shown comprising an optical camera 140 and a depth camera 141 .

[0093] It provides image data to a controller 142 which, under the control of a suitable computer program, performs the methods described above in the formation of optical image data and depth image data, and thus the system utilizes a controller to perform image data processing. The controller can be implemented in a number of ways, using software and / or hardware, to perform the various functions required. The controller typically uses one or more microprocessors which can be programmed using software (e.g., microcode) to perform the required functions. The controller can be implemented as a combination of dedicated hardware to perform some functions and one or more programmed microprocessors and associated circuitry to perform other functions.

[0094] Examples of circuitry that may be used in various embodiments of the present disclosure include, but are not limited to, conventional microprocessors, application specific integrated circuits (ASICs), and field programmable gate arrays (FPGAs).

[0095] In various implementations, the processor may be associated with one or more storage media, such as volatile and non-volatile computer memories, such as RAM, PROM, EPROM, and EEPROM. The storage media may be encoded with one or more programs that, when executed on the one or more processors and / or controllers, perform the necessary functions. The various storage media may be fixed within the processor or controller, or may be portable, such that the one or more programs stored thereon may be loaded into the processor.

[0096] The present invention can be used for non-contact patient weight estimation during examination preparation in medical imaging systems, including magnetic resonance (MR), computed tomography (CT), positron emission tomography (PET)-CT, PET-MR, and MR linear accelerator (Linac) systems. The estimated weight is used, for example, to derive specific absorption rate (SAR) values ​​for MR images or to derive recommended contrast agent volumes.

[0097] Variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality.

[0098] The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0099] The computer program may be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state medium, provided together with or as part of other hardware, but may also be distributed in other forms, such as over the Internet or other wired or wireless telecommunications systems.

[0100] It should be noted that when the term "adapted for" is used in the claims or description, it is intended to be equivalent to the term "configured for."

[0101] Any reference signs in the claims should not be construed as limiting the scope. Hereinafter, embodiments of the present invention will be described. (Appendix 1) 1. A computer-implemented method for estimating a patient's weight when supported on a patient table, comprising: receiving image data comprising optical image data of an optical image of the patient and depth image data of a depth image of the patient, the image data comprising a frame; detecting patient body keypoints from the optical image data; selecting a first frame having an image of the patient table by selecting a frame in which the patient body keypoints are not present in the optical image data; selecting a second frame having an image of a patient on the table by selecting a frame, a number of patient body keypoints present in the optical image data is greater than a first threshold; a movement measurement of a patient body keypoint in the optical image data between the frame and at least one previous frame is below a second threshold; Steps and determining a patient volume comprising an entire patient volume based on a difference between depth image data at a time of the first frame and depth image data at a time of the second frame, the difference defining a patient volume map; Typical patient density, the typical patient density of a particular patient category that includes said patient; or A typical tissue distribution is fitted to the patient volume map. estimating a patient weight, comprising a total patient weight, from the determined patient volume by converting the determined patient volume to the patient weight using The method comprising: (Appendix 2) The number of patient body keypoints present is greater than the first threshold, and the patient body keypoints between the frame and the at least one previous frame are identifying a second set of frames comprising all frames for which the body keypoint movement measure is below the second threshold; determining a patient volume for each of the second set of frames; The step of estimating the patient weight includes converting a minimum of the determined patient volumes to the patient weight. 2. The method of claim 1, comprising: (Appendix 3) 3. The method of claim 1 or 2, comprising estimating the patient weight by using an anatomical model of typical tissue distribution and associated densities of different tissues fitted to the patient volume map. (Appendix 4) 4. The method of claim 3, comprising estimating the patient weight by using a neural network having one or more patient volume maps as input. (Appendix 5) 5. The method of any one of claims 1 to 4, comprising receiving image data relating to a fixed patient table position. (Appendix 6) 5. A method according to any one of claims 1 to 4, comprising receiving image data for different patient table positions, the method comprising determining a table position from the image data, and calibrating the depth data using the determined table position by applying corresponding longitudinal and vertical translation parameters to the depth data. (Appendix 7) 5. The method of any one of claims 1 to 4, comprising receiving image data for different patient table positions, the method comprising selecting first and second frames corresponding to the same table position. (Appendix 8) A computer program comprising computer program code means adapted to perform a method according to any one of claims 1 to 7, when the computer program is run on a computer. (Appendix 9) 9. A controller programmed with the computer program of claim 8. (Appendix 10) An optical camera; A depth camera and 10. A controller as described in claim 9 for processing image data comprising optical image data from the optical camera and depth image data from the depth camera to estimate patient weight. A weight estimation system comprising: (Appendix 11) A patient support section; A medical imaging unit; The weight estimation system according to claim 10. A medical imaging system comprising: (Appendix 12) 11. The medical imaging system of claim 10, wherein the medical imaging unit comprises a CT scanner, an MRI scanner, or a PET scanner.

Claims

1. 1. A computer-implemented method for estimating a patient's weight when supported on a patient table, comprising: receiving image data comprising optical image data of an optical image of the patient and depth image data of a depth image of the patient, the image data comprising a frame; detecting patient body keypoints from the optical image data; selecting a first frame having an image of the patient table by selecting a frame in which the patient body keypoints are not present in the optical image data; selecting a second frame having an image of a patient on the patient table by selecting a frame, a number of patient body keypoints present in the optical image data is greater than a first threshold; a movement measurement of a patient body keypoint in the optical image data between the frame and at least one previous frame is below a second threshold; Steps and determining a patient volume comprising an entire patient volume based on a difference between depth image data at a time of the first frame and depth image data at a time of the second frame, the difference defining a patient volume map; Typical patient density, the typical patient density of a particular patient category that includes said patient; or A typical tissue distribution is fitted to the patient volume map. estimating a patient weight, comprising a total patient weight, from the determined patient volume by converting the determined patient volume to a patient weight using The method comprising:

2. The number of patient body keypoints present is greater than the first threshold, and the patient body keypoints between the frame and the at least one previous frame are identifying a second set of frames comprising all frames for which the body keypoint movement measure is below the second threshold; determining a patient volume for each of the second set of frames; The step of estimating the patient weight includes converting a minimum of the determined patient volumes to the patient weight.

2. The method of claim 1, comprising:

3. 2. The method of claim 1, comprising estimating the patient weight by using an anatomical model of typical tissue distribution and associated densities of different tissues fitted to the patient volume map.

4. 4. The method of claim 3, comprising estimating the patient weight by using a neural network having one or more patient volume maps as input.

5. The method of claim 1 , comprising receiving image data for a fixed patient table position.

6. 2. The method of claim 1, comprising receiving image data for different patient table positions, the method comprising determining a table position from the image data, and calibrating the depth image data using the determined table position by applying corresponding longitudinal and vertical translation parameters to the depth image data.

7. 2. The method of claim 1, further comprising receiving image data for different patient table positions, the method comprising selecting first and second frames that correspond to the same table position.

8. A computer program comprising computer program code means adapted to perform the method according to any one of claims 1 to 7 when the computer program is run on a computer.

9. A controller programmed with the computer program of claim 8.

10. An optical camera; A depth camera and 10. A controller according to claim 9 for processing image data comprising optical image data from the optical camera and depth image data from the depth camera to estimate patient weight. A weight estimation system comprising:

11. A patient support section; A medical imaging unit; The body weight estimation system according to claim 10. A medical imaging system comprising:

12. The medical imaging system of claim 11 , wherein the medical imaging unit comprises a CT scanner, an MRI scanner, or a PET scanner.

Images