Optical body imaging for respiratory monitoring

Abstract

The invention relates to a method (100) for measuring respiratory parameters of a subject using a range imaging sensor, wherein the method comprises: receiving at least one raw image of at least a portion of the subject's torso from the range imaging sensor (110), wherein each point of the raw image represents a distance between the range imaging sensor and the subject; generating a surface image of at least a portion of the subject's torso surface by surface interpolation of the raw image (120); estimating a respiratory signal over time, which is computed as a spatial average of the difference between the depth values of the surface image at a given time and the depth values of a reference surface image in a given region of interest (ROI) defined on the subject's torso (130); and estimating a lung volume (140).

Description

Technical Field

[0001] This invention relates to the field of physiological parameter monitoring. In particular, this invention relates to the field of monitoring a patient's respiratory parameters by means of image analysis. Background Technology

[0002] Mechanical ventilation, or assisted ventilation, uses mechanical means to assist or replace spontaneous breathing. The clinical goals of mechanical ventilation are to maintain gas exchange, reduce or replace respiratory effort, and monitor systemic oxygen consumption. Mechanical ventilation is administered via a device called a ventilator. A ventilator is an artificial organ that delivers air to a patient via non-invasive devices (via a mask or nasal sphincter) or invasive devices (via intubation). Continuous monitoring of respiratory parameters in mechanically ventilated patients is crucial for assessing changes in airway dynamics, but this continuous monitoring is currently only provided by invasive and airtight mask ventilation devices, not by non-airtight, non-invasive ventilation devices such as high-flow oxygen therapy systems. Mechanical ventilation is essential for the survival of patients with respiratory failure. Ventilators provide respiratory assistance by pressure or volume. During pressure ventilation, inspiratory pressure is adjusted by monitoring the volumetric changes caused by pressure changes that are not limited by the ventilator. Vital capacity and tidal volume alarms are set to avoid the risk of trauma. Volumetric ventilation requires adjustment of tidal volume or vital capacity. It has been shown that setting mechanical ventilation parameters is particularly crucial for patients with unstable breathing, and it should be noted that abnormal tidal volumes may cause damage such as lung disease. Intubation is associated with a higher risk of complications in unstable patients.

[0003] In this context, continuous monitoring of lung volume allows clinicians to track the health status of these patients and improve their prognosis by taking actions, particularly by fine-tuning ventilation parameters. This can help reduce the risk of intubation, shorten the duration of invasive ventilation, and decrease the risk of complementary lung complications that may result from mechanical ventilation. Therefore, non-invasive, non-contact monitoring to continuously measure the respiration of mechanically ventilated patients is a critical service in clinical intensive care.

[0004] Several clinical devices are available for monitoring patient ventilation. The most commonly used clinical device is a ventilator that allows air to be delivered in place of the respiratory muscles while also allowing monitoring of ventilation parameters to be adapted to the patient's needs.

[0005] Currently, there are no available non-invasive and non-contact monitoring systems in the clinical field capable of continuously monitoring a patient's ventilation. Technologies that provide accurate respiratory measurements, such as plethysmography, chest impedance, impedance respiration, optical plethysmography, and magnetometers, are non-invasive. However, these technologies are cumbersome, expensive, and not all are suitable for clinical settings.

[0006] Traditional techniques such as spirometers and respiratory velocimeters measure airflow rate during respiration. These contact techniques require patient cooperation during implementation. They allow for the measurement of lung volume but do not permit the detection of abnormalities in chest wall movement associated with pathologies such as pulmonary edema and chest region dysfunction. Breathing bands are another well-known contact-based non-invasive technique that allows for the measurement of a patient's respiratory rhythm.

[0007] Conventional respiratory testing is typically performed via contact methods, which are considered disadvantageous to patients and inaccurate. In fact, in the field of respiratory dynamics analysis, it is advantageous to avoid physical contact between the measurement system and the patient.

[0008] In this context, the present invention aims to overcome these disadvantages by providing a method and system for measuring patient respiratory parameters through non-invasive and non-contact monitoring of patient respiration.

[0009] Another application of respiratory monitoring involves pulmonary function testing, which examines the function of gas exchange during respiration. Some subjects, such as children, the elderly, or those with lung conditions like chronic obstructive pulmonary disease (COPD), may experience difficulties using a spirometer for this test due to a feeling of constraint (because the mouthpiece and nasal plug must be attached to the subject's face) or a lack of strength to perform forced exhalation / inhalation. Furthermore, measurements may be inaccurate due to occasional leaks. Finally, the mouthpiece must be replaced for each patient.

[0010] In this context, the present invention provides a solution to these drawbacks by proposing a method and system for measuring a patient's respiratory parameters in a non-invasive and non-contact manner. Summary of the Invention

[0011] This invention relates to a computer-implemented method for estimating respiratory parameters of a subject, wherein the method includes:

[0012] - Receive a set of acquisitions obtained from a distance imaging sensor, the set of acquisitions including at least one raw image of at least a portion of the subject's torso, wherein each point of the raw image represents the distance between the distance imaging sensor and the subject;

[0013] - Generate a surface image of at least a portion of the subject's torso by surface interpolation of the original image;

[0014] - Estimate the respiratory signal as it changes over time, which is calculated as the spatial average of the difference between the depth values ​​of a surface image obtained from the set of acquisitions at a given time and the depth values ​​of a reference surface image in a given area of ​​interest defined on the subject's torso.

[0015] - Estimating respiratory parameters, including at least the time-varying lung volume, by multiplying the respiratory signal by the surface area of ​​the region of interest, and

[0016] - The respiratory parameters are provided as output.

[0017] This method is based on the analysis of 3D surface images (i.e., 3D point clouds) to estimate one-dimensional respiratory signals and further calculate other respiratory parameters.

[0018] Advantageously, this method provides absolute quantitative estimates of the subject's breathing and respiratory parameters in a real-time and continuous manner.

[0019] Advantageously, this method allows for accurate estimation of respiratory parameters using raw data acquired with only a single camera. Therefore, less data must be analyzed to estimate respiratory parameters, which allows for reduced computation time and facilitates real-time implementation of the method.

[0020] The respiratory parameters defined according to this invention are also called respiratory volume. Vital capacity is obtained as the sum of different respiratory volumes and represents the amount of air that can be inhaled or exhaled during one respiratory cycle. In this invention, chest region movement (chest wall movement) can be tracked by analyzing 3D spatial information provided by a distance imaging sensor. The difference between the surface image (chest wall shape) and a reference surface image provides information about the amount of air that can be inhaled or exhaled, thereby allowing for accurate estimation of lung volume (changes in chest wall relative to a reference).

[0021] In one embodiment, the present invention relates to a method for measuring respiratory parameters of a subject using a distance imaging sensor, wherein the method includes:

[0022] - Receive at least one raw image of at least a portion of the subject's torso from a distance imaging sensor, wherein each point in the raw image represents the distance between the distance imaging sensor and the subject;

[0023] - Generate a surface image of at least a portion of the subject's torso by surface interpolation of the original image;

[0024] - Estimate the respiratory signal over time, which is calculated as the spatial average of the difference between the depth value of a surface image at a given time and the depth value of a reference surface image within a given region of interest (ROI) defined on the subject's torso.

[0025] - Estimating lung volume over time by multiplying the respiratory signal by the surface area of ​​the region of interest.

[0026] In this invention, the area of ​​interest defined on the torso includes the body region from the subject's shoulders to the hip bones. Thoracic breathing relies primarily on the contraction of the intercostal muscles, while abdominal breathing relies primarily on the contraction of the diaphragm. Chest wall movements include both thoracic and abdominal breathing. Therefore, in this invention, respiratory movements across the entire torso region are considered, thereby advantageously allowing for more accurate estimation of respiratory parameters.

[0027] According to one embodiment, the method of the present invention is a computer-implemented method.

[0028] This method advantageously allows for real-time, non-invasive, and non-contact monitoring of a subject's respiratory parameters. This monitoring provides clinicians with useful information, allowing for improvements in the efficiency of mechanical ventilation by fine-tuning ventilation parameters, particularly by adjusting the volume or pressure of air to be delivered. Therefore, monitoring respiratory parameters allows for reductions in the duration of mechanical ventilation, thus preventing supplemental lung complications arising from such ventilation. Monitoring respiratory parameters also allows for the detection of respiratory abnormalities such as pneumothorax, lung collapse, and diaphragmatic paralysis. Furthermore, monitoring respiratory parameters, especially for patients undergoing prolonged invasive ventilation, allows for monitoring their health status during or after invasive ventilation cessation.

[0029] According to one embodiment, the method further includes estimating the tidal volume as the respiratory signal multiplied by the difference between the maximum and minimum values ​​of the surface of the region of interest during a respiratory cycle.

[0030] According to one embodiment, the method further includes estimating the respiratory rate calculated based on the detection of inspiratory peaks in the respiratory signal.

[0031] In one embodiment, respiratory parameters, in addition to lung volume, include tidal volume, minute ventilation, respiratory rate, vital capacity, expiratory reserve, inspiratory reserve, inspiratory volume, and / or inspiratory vital capacity. Inspiratory and expiratory reserves can only be estimated in conscious subjects capable of performing forced inspiration and expiration.

[0032] According to one embodiment, the surface image is obtained using a base spline function.

[0033] According to one embodiment, the distance imaging sensor is a time-of-flight (ToF) camera.

[0034] According to one embodiment, the method further includes the step of filtering the original image to remove noise originating from other objects in the scene.

[0035] According to one embodiment, the method further includes a calibration step of applying a rotation matrix to the original image to align the subject's torso in the original image with the xy plane of the distance imaging sensor.

[0036] This calibration advantageously allows the subject's torso in the raw image to be aligned with the xy-plane of the distance imaging sensor, enabling the method to be implemented on the subject regardless of his / her actual position, as long as the torso is included in the camera's field of view. Therefore, this method is suitable for estimating respiratory parameters of subjects who are standing, sitting, or lying down.

[0037] According to one embodiment, the method further includes the step of controlling the ventilator by modifying the value of at least one ventilation parameter, wherein the value of the ventilation parameter is calculated using at least one of the estimated respiratory parameters, particularly tidal volume. This embodiment advantageously allows for a reduction in the occurrence of ventilation-related complications.

[0038] The present invention also relates to a program including instructions that, when executed by a computer, cause the computer to perform the steps of the method according to any embodiment as described above.

[0039] The present invention also relates to a non-transitory computer-readable medium comprising instructions that, when executed by a computer, cause the computer to perform the steps of the method according to any embodiment as described above.

[0040] In the following text, modules will be understood as functional entities rather than materially or physically distinct components. Therefore, they can be implemented as combined within the same tangible and concrete component, or distributed across several such components. Furthermore, each of these modules may be shared between at least two physical components. Moreover, these modules can be implemented in hardware, software, firmware, or any hybrid form thereof.

[0041] The present invention also relates to a system for measuring respiratory parameters of a subject, comprising:

[0042] - Acquisition module configured to control distance imaging sensor to acquire at least one raw image including at least a portion of the subject's torso, wherein each point of the raw image represents the distance between the distance imaging sensor and the subject;

[0043] - A surface generation module configured to generate a surface image of at least a portion of the subject's torso surface through surface interpolation of the original image;

[0044] - A calculation module configured to calculate the time-varying respiratory signal as the spatial average of the difference between the depth value of a surface image at a given time and the depth value of a reference surface image in a given region of interest defined on the subject's torso, and configured to calculate the time-varying lung volume by multiplying the respiratory signal by the surface of the region of interest.

[0045] The present invention also relates to a system for estimating respiratory parameters of a subject, comprising:

[0046] - An input module configured to receive a set of acquisitions from a distance imaging sensor, the set of acquisitions including at least one raw image containing at least a portion of the subject's torso, wherein each point of the raw image represents the distance between the distance imaging sensor and the subject;

[0047] - A surface generation module configured to generate a surface image of at least a portion of the subject's torso surface through surface interpolation of the original image;

[0048] - A calculation module configured to calculate a time-varying respiratory signal as a spatial average of the difference between a depth value of a surface image obtained from the set of acquisitions at a given time in a given region of interest defined on the subject's torso and a depth value of a reference surface image, and configured to calculate respiratory parameters including at least a time-varying lung volume by multiplying the respiratory signal by the surface of the region of interest.

[0049] - Output module, configured to output the respiratory parameters.

[0050] In one embodiment, the calculation module is configured to also calculate the respiratory rate based on the detection of inspiratory peaks in the respiratory signal.

[0051] In one embodiment, the calculation module is further configured to calculate the tidal volume as the respiratory signal multiplied by the difference between the maximum and minimum values ​​of the surface of the region of interest during a respiratory cycle.

[0052] According to one embodiment, the system also includes a distance imaging sensor and an acquisition module configured to control the distance imaging sensor for acquiring the set of raw images.

[0053] According to one embodiment, the system also includes a time-of-flight camera with infrared illumination.

[0054] According to one embodiment, the distance imaging sensor is placed in front of the subject's torso.

[0055] According to one embodiment, the distance imaging sensor is a time-of-flight camera, and the system includes a calibration module configured to apply a rotation matrix to the raw image in order to align the subject's torso in the raw image with the xy plane of the time-of-flight camera.

[0056] According to one embodiment, the surface generation module is also configured to filter the original image to remove noise originating from other objects in the scene.

[0057] definition

[0058] In this invention, the following terms have the following meanings:

[0059] - "Lung volume" refers to the amount of air in the lungs at different stages of the respiratory cycle.

[0060] The term "processor" as used herein is not limited to hardware capable of executing software, but also refers generally to processing devices, which may include, for example, computers, microprocessors, integrated circuits, or programmable logic devices (PLDs). A processor may also include one or more graphics processing units (GPUs), whether for computer graphics and image processing or for other functions. Furthermore, instructions and / or data capable of performing associated and / or generated functions may be stored on any processor-readable medium, such as integrated circuits, hard disks, CDs (optical discs), optical discs (such as DVDs (Digital Versatile Optical Discs), RAM (Random Access Memory), or ROM (Read-Only Memory)). Instructions may be specifically stored in hardware, software, firmware, or any combination thereof.

[0061] "Real-time": refers to the system's ability to control the environment by receiving data, processing it, and returning results quickly enough to influence the environment at that time. Real-time response (i.e., output) is usually understood in milliseconds, sometimes in microseconds.

[0062] - "Respiratory parameters" refer to lung volume, tidal volume, and minute ventilation calculated over a region of interest (ROI) encompassing the patient's entire trunk or over multiple ROIs corresponding to the abdominal, chest, left, or right lung regions. Additionally, respiratory parameters also include vital capacity (the volume of air exhaled after a maximum inspiration), expiratory reserve (the maximum volume of air that can be exhaled from the end of expiration), inspiratory reserve (the maximum volume of air that can be inhaled from the end of inspiration), inspiratory volume (the sum of inspiratory reserve and tidal volume), and inspiratory vital capacity (the maximum volume of air inhaled from the point of maximal expiration). Inspiratory and expiratory reserves can only be estimated in conscious subjects capable of performing forced inspiration and expiration.

[0063] - "Subject" refers to a mammal, preferably a human. In the sense of this invention, a subject may be an individual suffering from any mental or physical illness requiring regular or frequent medication, or may be a patient, i.e., a person receiving medical care, undergoing or having undergone drug treatment, or whose disease progression is being monitored.

[0064] - "Tidal volume" refers to the normal amount of air expelled between a normal inspiration and an expiration without additional effort. This respiratory parameter is expressed in milliliters.

[0065] - The "trunk" or "body" refers to the central part or core of the body of many animals (including humans), from which the neck and limbs extend. The trunk includes: the thoracic section of the trunk (i.e., the chest or pleural cavity), the abdominal section of the trunk (i.e., the abdomen, which is the part of the body between the chest and the pelvis), and the perineum. Attached Figure Description

[0066] The following detailed description will be better understood when read in conjunction with the accompanying drawings. For illustrative purposes, block diagrams of the system and the described method are shown in preferred embodiments. However, it should be understood that this application is not limited to the precise arrangements, structures, features, implementations, and aspects shown. The drawings are not drawn to scale and are not intended to limit the scope of the claims to the depicted embodiments. Therefore, it should be understood that where reference numerals follow features mentioned in the appended claims, these included reference numerals are only for enhancing the comprehensibility of the claims and do not in any way limit the scope of the claims.

[0067] The features and advantages of the present invention will become apparent from the following description of system embodiments, which are given by way of example only and with reference to the accompanying drawings, wherein:

[0068] Figure 1 This is a block diagram illustrating the steps of a method according to an embodiment of the present invention.

[0069] Figure 2 This is a schematic representation of a system according to one embodiment of the present invention.

[0070] Figure 3 This is a graph showing the change in lung volume over time in the subject's left lung (black marker) and right lung (gray marker).

[0071] Figure 4 This is a graph showing the change in lung volume over time in the subject's chest region (black marker) and abdominal region (gray marker).

[0072] Figure 5 This is a box plot showing the median, one standard deviation above and below the data mean, minimum, and maximum values ​​of the respiratory rate parameter calculated for 6 patients.

[0073] Figure 6 It is a box plot showing the median, one standard deviation above and below the data mean, minimum and maximum values ​​of the tidal volume parameters calculated for 6 patients.

[0074] Figure 7 (a) and (b) show the correlation and Bland-Altman plots between the reference respiratory rate and the estimated respiratory rate. The reference RR value was provided by a human model ventilator. The estimated RR was implemented using a Kinect-based monitoring system. In 21 recordings, the estimated RR was highly correlated with the reference method (r = 0.99; p < 0.001). Comparison of the two methods using the Bland-Altman plot showed low deviation (0 bps) and bias (< ± 1.86 bps).

[0075] Figure 8(a) and (b) show the correlation and Bland-Altman plots between the reference tidal volume and the estimated tidal volume. The reference Vt value was provided by a human model ventilator. The estimated Vt was implemented using a Kinect-based monitoring system. In 21 recordings, the estimated RR was highly correlated with the reference method (r = 0.99; p < 0.001). Comparison of the two methods using the Bland-Altman plot showed a low deviation (7.90 ml) and bias (< ± 22.03 ml).

[0076] Figure 9 This is a graph representing the estimated regional ventilation (ROI) for the left and right parts of the chest in a human model. The volume-time curves were obtained by analyzing the left and right lung ROIs separately using the asynchronous ventilation mode of the human model. The black curve corresponds to the volume-time curve for the left chest, and the gray curve corresponds to the volume-time curve for the right chest.

[0077] Figure 10 (a) and (b) show the correlation and Bland-Altman plots between the reference respiratory rate (RR) and the estimated RR. The reference RR was provided by a ventilator for patients receiving ventilator support. The estimated RR was performed using a Kinect-based monitoring system. In the recordings of 16 ICU patients, the estimated RR was highly correlated with the reference method (r = 0.95; p < 0.001). Comparison of the two methods using the Bland-Altman plot showed a low deviation (0.40 bps) and bias (< ± 1.67 bps).

[0078] Figure 11 (a) and (b) show the correlation and Bland-Altman plots between the reference tidal volume and the estimated tidal volume. The reference Vt value was provided by a ventilator for patients receiving ventilatory support. Vt estimation was performed using a Kinect-based monitoring system. In the records of 16 ICU patients, the estimated Vt was highly correlated with the reference method (r = 0.90; p < 0.001). Comparison of the two methods using the Bland-Altman plot showed a low deviation (-5.36 ml) and an acceptable deviation (< ± 23.70 ml).

[0079] Figure 12 This is a graph that shows whether the patient's average error is more or less than the standard deviation of the estimated RR.

[0080] Figure 13 This is a graph that shows whether the patient's average error is more or less than the standard deviation of the estimated Vt.

[0081] Figure 14(a) represents equal ventilation in the left and right chests of two ICU patients, and (b) represents unequal ventilation in the left and right chests of two ICU patients. Volume-time curves were obtained by analyzing the left and right lung regions of interest (ROIs) separately. The black curve corresponds to the volume-time curve of the left chest, and the gray curve corresponds to the volume-time curve of the right chest.

[0082] Figure 15 (a) represents synchronized ventilation between the chest and abdomen of two ICU patients, and (b) represents asynchronous ventilation between the chest and abdomen of two ICU patients. Volume-time curves were obtained by analyzing the chest and abdominal ROIs separately. The black curve corresponds to the abdominal volume-time curve, and the gray curve corresponds to the chest volume-time curve.

[0083] Although various embodiments have been described and illustrated, the detailed description should not be construed as limiting thereto. Various modifications can be made to the embodiments by those skilled in the art without departing from the true spirit and scope of this disclosure as defined by the claims. Detailed Implementation

[0084] Figure 1 A block diagram illustrating some of the main steps of a method 100 for measuring respiratory parameters of a subject using a distance imaging sensor is shown. In a preferred embodiment, the subject is a mechanically ventilated patient.

[0085] Because clinical environments such as intensive care units, rehabilitation rooms, and emergency medical services are equipped with a variety of health monitoring devices (i.e., monitors, ventilation equipment, cardiac monitoring equipment, etc.), and all of these devices must be mobile to facilitate and quickly access patients for care management or in emergency situations, distance imaging sensors can be advantageously placed on top of the bed.

[0086] In one embodiment, the first step 110 of the method comprises receiving at least one raw image of at least a portion of the patient's torso acquired from a distance imaging sensor. Each point in the raw image acquired from the distance imaging sensor represents the distance between the distance imaging sensor and the patient.

[0087] Currently, there are various distance imaging sensors or cameras available, which are based on different types of distance imaging technologies, such as stereo triangulation, optical triangulation, structured light, interferometry, coded aperture, and any other technology known to those skilled in the art.

[0088] According to a preferred embodiment, the distance imaging sensor is a time-of-flight (ToF) camera. A ToF camera is a distance imaging camera system that uses time-of-flight technology to resolve the distance between the camera and the subject for each point in the image by measuring the round-trip time of artificial light signals provided by lasers, laser diodes, or LEDs of different wavelengths (particularly infrared or near-infrared). Instead of measuring the intensity of ambient light as in a standard camera, a ToF camera measures reflected light from its own light source emitter.

[0089] Different measurement principles of time-of-flight cameras can be used to achieve the objectives of this invention, including: (i) a pulsed light camera that directly measures the time it takes for a light pulse to travel from a device to an object and back; and (ii) a continuous wave modulated light camera that measures the phase difference between the transmitted and received signals, thereby indirectly obtaining the propagation time.

[0090] ToF cameras can combine single or multiple laser beams (possibly mounted on a rotating mechanism) with a 2D photodetector array and a time-to-digital converter to produce an array of 1-D or 2-D depth values.

[0091] In this method, the ToF camera receiving the raw images can include two types of sensors: pulsed light sensors or continuous wave modulation (CW) sensors. Pulsed light sensors directly measure the round-trip time of a light pulse. The pulse width is a few nanoseconds. Continuous wave (CW) modulation sensors measure the phase difference between the emitted continuous sinusoidal light signal and the backscattered signal received by each photodetector. The phase difference between the emitted and received signals is estimated via cross-correlation (demodulation). Given a known modulation frequency, the phase is directly related to distance. These sensors typically operate indoors and can only perform short-range measurements (from a few centimeters to a few meters), making them well-suited for use in clinical settings.

[0092] Time-of-Flight (ToF) cameras allow for rapid acquisition and real-time processing of scene information. A key feature of ToF cameras is their high acquisition frequency, which enables real-time 3D volumetric analysis. The surface information provided by this type of camera is advantageously suited for the management and monitoring of dynamic motion.

[0093] In contrast to radar systems with transmitting and receiving antennas, the use of range imaging sensors is particularly advantageous. In fact, the electromagnetic waves used by transmitting antennas can have side effects on the patient and disrupt the function of medical devices inside the patient (i.e., peacemaker or insulin pumps) or other medical devices near the patient. Conversely, this method and system are configured to use infrared radiation that does not cause side effects on the patient and does not interfere with the function of medical devices in the clinical setting. Furthermore, images obtained from range imaging sensors allow for granular information about the patient's entire trunk displacement, while the use of radar systems only provides timing information, resulting in poorer estimations of lung volume.

[0094] The method can also be configured to provide instructions to a distance imaging sensor to control image acquisition and transmission to a data processing device configured to perform the steps of this method. Image acquisition can be scheduled based on a predefined timetable or triggered upon receiving a trigger signal.

[0095] When using a ToF camera that implements continuous wave modulation technology, the cross-correlation between the optical power s(t) of the emitted signal formed by the reflection of light on the imaging object and the optical power r(t) of the received signal is expressed by the following equation:

[0096]

[0097] If we consider the correlation function values ​​of four equally spaced samples within a modulation period:

[0098]

[0099] These four sample values ​​are sufficient to clearly calculate the obtained phase. As shown below:

[0100]

[0101] The depth value d for each pixel is calculated using the following formula:

[0102]

[0103] Where c is the speed of light and f is the modulation frequency.

[0104] According to one embodiment, a ToF camera includes an emitter, a light emitter (typically an LED or light-emitting diode) configured to send light onto an object, such that the time required for light to travel from an illumination source to the object and return to the sensor is measured. In the case of continuous wave (CW), the emitted signal is a sinusoidally modulated light signal. Due to the round-trip journey of the light signal, the received signal undergoes a phase shift. Furthermore, the received signal is affected by the object's reflectivity, attenuation along the light path, and background illumination. Each pixel independently performs demodulation of the received signal, and thus it is possible to measure its phase delay as well as amplitude and offset (background illumination).

[0105] The original image is a depth map, or a depth image that can be presented as a two-dimensional array representing a grayscale or RGB image, where the size of the array depends on the ToF camera, and in particular the photoelectric sensor used for image acquisition. The original image can be encoded, for example, in 16 bits, where the depth measurement information in each pixel (u,v) directly corresponds to the distance between the ToF camera and the object (i.e., the patient).

[0106] According to one embodiment, the method further includes a calibration step, which involves applying a rotation matrix to the original image to align the patient's torso in the original image with the xy-plane of the time-of-flight camera. In one example, the following three rotations are performed:

[0107]

[0108]

[0109]

[0110] Apply the following equation to the original image:

[0111]

[0112] Where (X) t ,Y t Z t ) is from the rotation matrix R = R x ×R y ×R z The coordinates of the transformed point, α, β, and γ, are radian angles along the three axes of the ToF camera, (X... i ,Y i Z i ) are the coordinates of the original image points acquired by the ToF camera.

[0113] According to one embodiment, the method further includes step 111 of filtering the original image to remove noise originating from other objects in the scene.

[0114] In one embodiment, the method includes step 120 of generating a surface image of at least a portion of the patient's torso surface through surface interpolation of the original image. In one embodiment, the surface image is obtained using a base spline (also known as a B-spline) function on the original image. A B-spline function is a combination of a number of flexible bands of points called control points that create a smooth curve. These functions can create and manage complex shapes and surfaces using a finite number of points. B-spline functions and Bessel functions are widely used in shape optimization methods. This embodiment advantageously allows respiratory motion dynamics to be derived on a point-by-point basis rather than a region-by-region basis. Furthermore, B-spline modeling significantly improves depth estimation accuracy and reduces the error of respiratory signal measurements to 0.22 ± 0.14 mm, while improving measurement repeatability by a factor of 3.

[0115] In one embodiment, the method includes step 130 of obtaining a time-varying respiratory signal by analyzing data within a given region of interest (ROI) defined on the patient's torso. The time-varying respiratory signal is then estimated as the spatial average of the difference between a depth value of a surface image at a given time and a depth value of a reference surface image within the given ROI. The reference surface image may be the first image of the set of acquisitions. Advantageously, the use of this reference surface image allows for the definition of baseline drift. In effect, the respiratory signal is relative to a reference value (referred to as the baseline). The baseline of the respiratory measurement is defined by the reference image (ideally the zero-flow line).

[0116] Therefore, the respiratory signals that change over time are obtained as follows:

[0117]

[0118] Where R represents the reference surface image, L represents the torso surface image corresponding to the k-th depth image acquired, and N is the number of pixels in the region of interest defined on the patient's torso.

[0119] Advantageously, the position of the reference plane (i.e., the reference surface image) is not important in the estimation of moisture volume and does not affect the accuracy of the volume difference between surfaces.

[0120] According to one embodiment, the region of interest is predefined manually by the user. According to an alternative embodiment, the location and size of the region of interest are automatically defined on the patient's torso using a deep learning algorithm. The deep learning algorithm is configured to receive an RGB image of the patient (particularly including at least the patient's torso) as input and provide a model of the subject's skeleton (i.e., the spatial distribution of the patient's bones and joints) as output. Real-time tracking of the skeletal model can be implemented based on the deep learning algorithm. The subject's skeletal model is then used to select at least one region of interest. In one embodiment, a region of interest is defined as the area corresponding to the entire torso from the hip bone to the scapula. In another embodiment, two regions of interest are defined on the torso (the thoracic region including the ribs, sternum, and scapula, and the abdominal region including the area between the diaphragm and hip bone). Alternatively, a first region can be defined on the left side of the ribcage to cover the left lung, and a second region can be defined on the right side of the ribcage to cover the right lung.

[0121] In one example, a deep learning algorithm for real-time tracking of a skeletal model performs human pose estimation based on a deep neural network model. It has been shown that this algorithm can accurately predict the locations of various human "keypoints" (joints and landmarks, such as elbows, knees, neck, shoulders, hips, chest, etc.). In one embodiment, the deep neural network model is configured to take a color image of the patient as input and generate 2D locations of the "keypoints" (i.e., the spatial distribution of the patient's bones and joints) as output. Each coordinate in the skeleton is called a joint. A pre-trained model can be used on a dataset to detect human joints that generate 18 points, including facial and body "keypoints." This set of coordinates can then be concatenated to identify Regions of Interest (ROIs). The patient's torso ROI can then be defined by the surfaces connecting the identified shoulder and hip joints. This ROI detection algorithm can also be used to track the patient's movements.

[0122] The lung volume, which varies over time, is obtained by calculating the volume of the Region of Interest (ROI) for each frame. This is done by calculating the difference between the current depth surface and the reference depth surface for all pixels within the ROI. The size of the ROI is crucial for obtaining the volume. The size of the ROI can be measured in mm. 2 The number of pixels in the ROI is expressed in units of width and length. Preferably, the size of the ROI is the area measured as the product of its width and length. The width and length are defined as the distance between the minimum and maximum coordinates of pixels within the ROI along the x and y axes, respectively. Since inherent camera calibration is performed before image acquisition begins to find the relationship between the camera's natural units (pixel positions in the image) and real-world units (in millimeters), the surface size can be multiplied by the depth difference (between the actual and reference surfaces).

[0123] Advantageously, using surfaces (3D point clouds) to estimate breathing parameters (instead of 2D images) allows for the avoidance of additional calibration steps by scaling factors that depend on the camera's intrinsic parameters (especially the camera's focal length).

[0124] According to one embodiment, the method of the present invention further includes the step 140 of estimating the time-varying lung volume by multiplying the respiratory signal by the surface of the region of interest. The equation representing the value of the time-varying lung volume V(k) can be obtained from the following equation:

[0125] V(k)=D(k)×S (8)

[0126] Where D(k) is a measure of the average depth change estimated for the k-th image of the acquired samples, and S is the depth measure expressed in mm. 2 The parameters of the surface of the region of interest are quantified. The application of this equation allows for the acquisition of a curve of lung volume V(k) over time from depth images, showing the peak value of each inhalation and the trough value of each exhalation of the subject.

[0127] According to the implementation example of automatically defined ROI, parameter S is calculated in mm. 2 The surface of the calculated region of interest.

[0128] According to one embodiment, the lung volume changing over time can be calculated as follows:

[0129] V(k)=D(k)×S×10 3

[0130] Among them, the multiplication factor is 10 3 Allowed from mm 3 The conversion from volume in units to volume in ml.

[0131] According to one embodiment, the method of the present invention further includes the step of estimating the tidal volume as the respiratory signal multiplied by the amplitude difference between the maximum and minimum values ​​of the surface of the region of interest during a respiratory cycle (i.e., the amplitude difference between the maximum and minimum values ​​of the lung volume V(k) during a respiratory cycle). This estimation of tidal volume is a key parameter for monitoring ventilated patients.

[0132] In one embodiment, peaks in the curve of lung volume V(k) versus time are detected within a given time window comprising at least one respiratory cycle, for example, in minutes. This step can be implemented using peak detection algorithms known to those skilled in the art. The amplitude between the maximum and minimum values ​​of each peak corresponds to the volume inhaled by the subject during the corresponding respiratory cycle. In this embodiment, tidal volume is calculated as the average of the amplitudes calculated over multiple respiratory cycles within a given time window. Tidal volume can be represented as an average obtained from each respiratory cycle, advantageously allowing for the avoidance of errors caused by random noise and the irregularity and variability of the obtained measurements. This value is crucial for assessing respiration, providing opportunities for early intervention, adjustment of ventilation parameters, or diagnosis of abnormalities.

[0133] According to one embodiment, the method further includes estimating the volume per minute, which is represented as the volume per minute calculated as the product of tidal volume and respiratory rate obtained for the ROI including the entire torso.

[0134] According to one embodiment, the method further includes the step of controlling the ventilator by modifying the value of at least one ventilation parameter, wherein the value of the ventilation parameter is calculated using at least one of the estimated respiratory parameters, particularly tidal volume. For example, the ventilation parameter to be controlled could be the volume or pressure delivered to the patient by the ventilator under mechanical ventilation. This embodiment advantageously allows for the adaptation of respiratory therapy delivered by the ventilator to each patient, particularly to his / her anatomy and current health status.

[0135] According to one embodiment, the method further includes the step of detecting the presence of respiratory abnormalities such as pneumothorax, lung collapse, paralysis of the diaphragm, etc., by monitoring multiple areas of interest on the patient's torso, as described in the embodiments above.

[0136] As described above, the region of interest (ROI) can be defined manually or automatically. In one embodiment, the ROI is automatically defined on half of the chest to cover only the surface of one lung. In this embodiment, the first ROI is defined on the patient's right lung and the second ROI is defined on the patient's left lung. According to the above embodiment, respiratory rate and tidal volume are calculated for the first and second ROIs. Monitoring of respiratory rate and tidal volume on the first and second ROIs corresponds to monitoring of these parameters in the left and right lungs, respectively. Volume per minute can also be estimated, which is represented as the volume per minute calculated as the product of the respiratory rate and tidal volume of the first and second ROIs. Figure 3An example is shown where the calculated left lung volume (marked by the black line) overlaps with the calculated right lung volume (marked by the gray line) for a single patient. In this example, the lung volumes of both the left and right lungs change uniformly over time. Comparison of respiratory rate and tidal volume on the first and second ROIs advantageously allows for the assessment of lung volume distribution between the two lungs and the detection of lung heterogeneity. This information advantageously allows for the detection of non-uniform air volume distribution and the mapping of regional anomalies to diagnose respiratory problems such as lung inflammation or pneumonia, and leads to optimal ventilation adjusted according to the patient's needs.

[0137] In one embodiment, the ROI detection and tracking step is configured to take into account all changes to the chest surface that occur during acquisition.

[0138] In an alternative embodiment, the chest and abdominal regions are defined as regions of interest. According to the above embodiment, respiratory rate and tidal volume are calculated for the chest and abdomen to monitor these parameters separately. Monitoring respiratory parameters in the abdominal and chest regions advantageously allows for the detection of paradoxical breathing (also known as thoracoabdominal paradox) where the lungs and diaphragm move in the opposite direction to their normal direction of movement. This information, in particular, allows for the diagnosis of physiological disorders such as respiratory distress, trauma, neurological problems, etc. Figure 4 This example shows an overlap between lung volume calculated for a patient's chest region (black line and markers) and lung volume calculated for their abdominal region (gray line and markers). In this example, the changes in lung volume over time are correlated for both the chest and abdominal regions.

[0139] According to one embodiment, the method further includes step 150 of analyzing the patient's respiratory signals to calculate the respiratory rate based on morphological changes in a surface image of the torso.

[0140] According to one embodiment, the respiratory rate is calculated based on the detection of inspiratory peaks in the respiratory signal. Peak detection can be performed using an amplitude threshold extremum detection technique. This technique involves detecting the maximum and minimum values ​​in the respiratory signal by searching for changes in the sign of the overall sample. The sample can correspond to acquisition intervals from 10 seconds to 5 minutes. In one example considering sample acquisitions within a 1-minute acquisition interval, the relative extrema (maximum and minimum values) can be considered superior in absolute value to the average value of the respiratory signal within the acquisition interval. Once a peak is detected, the respiratory rate is determined based on the number of peaks in the acquisition interval.

[0141] In one example, the clinical data were obtained from patients in the medical intensive care unit of CHU Cavale de Brest. Monitoring lasted for 10 to 20 minutes. Therefore, for each patient, ventilation parameters were calculated every minute of data collection (10 to 20 measurements were performed per patient).

[0142] Method 100 for measuring respiratory parameters has been validated in 35 patients (30 intubated, 5 spontaneously ventilated). For the 30 intubated patients, the results were compared with clinical system measurements (ventilator), and the mean difference in tidal volume calculations was 30 ± 19 mL. As for respiratory rate, the mean difference was 1.8 ± 1.4 cpm. These values ​​are well below clinically acceptable limits.

[0143] Regional volume monitoring (right and left lungs) was tested on a phantom attached to a respirator. The phantom had an asymmetrical operating mode that allowed air to be injected only into the right lung (one valve open, the other closed). The calculated volume was compared to this setting, and 90% of the theoretical volume was detected being blown into the right lung.

[0144] Figure 5 and Figure 6 Box plots of respiratory rate and tidal volume for six patients are shown. The mean error obtained in the measurements for these patients was 20 ± 9 mL (for tidal volume calculation) and 1.6 ± 1.1 cpm (for respiratory rate calculation).

[0145] The present invention also relates to a program including instructions that, when executed by a computer, cause the computer to perform the steps of the method according to any embodiment as described above.

[0146] A computer program product for performing the methods described above can be written as a computer program, code segment, instruction, or any combination thereof, for individually or collectively instructing or configuring a processor or computer to operate as a machine or special-purpose computer to perform operations performed by hardware components. In one example, the computer program product includes machine code that is directly executed by a processor or computer, such as machine code generated by a compiler. In another example, the computer program product includes higher-level code that is executed by a processor or a computer using an interpreter. Those skilled in the art can readily write instructions or software based on the block diagrams and flowcharts shown in the accompanying drawings and the corresponding descriptions in the specification, which discloses algorithms for performing the methods described above.

[0147] The present invention also relates to a non-transitory computer-readable medium comprising instructions that, when executed by a computer, cause the computer to perform the steps of the method according to any embodiment as described above.

[0148] Computer programs implementing the methods of this embodiment can typically be distributed to a user on distributed computer-readable storage media such as, but not limited to, SD cards, external storage devices, microchips, flash memory devices, portable hard drives, and software sites. The computer program can be copied from the distributed medium to a hard drive or similar intermediate storage medium. The computer program can be run by loading computer instructions from their distributed medium or from their intermediate storage medium into a computer execution memory to configure the computer to function according to the methods of the invention. All of these operations are well known to those skilled in the art of computer systems.

[0149] Instructions or software for controlling a processor or computer to implement hardware components and perform the methods described above, along with any associated data, data files, and data structures, are recorded, stored, or fixed in or on one or more non-transitory computer-readable storage media. Examples of non-transitory computer-readable storage media include read-only memory (ROM), random access memory (RAM), flash memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-RLTH, BD-RE, magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid-state drive, and any device known to those skilled in the art capable of storing instructions or software and any associated data, data files, and data structures in a non-transitory manner and providing instructions or software and any associated data, data files, and data structures to a processor or computer so that the processor or computer can execute the instructions. In one example, instructions or software, along with any associated data, data files, and data structures, are distributed across a network-coupled computer system so that processors or computers can store, access, and execute the instructions and software, along with any associated data, data files, and data structures, in a distributed manner.

[0150] Another aspect of the invention relates to a system S for measuring respiratory parameters of a subject, particularly a patient.

[0151] like Figure 2 As shown, the system of the present invention includes multiple modules configured to cooperate with each other and perform the steps of the method of the present invention.

[0152] According to one embodiment, the system includes an input module aM configured to receive a set of acquisitions from a distance imaging sensor Sri, the set of acquisitions including at least one raw image containing at least a portion of the subject's torso, wherein each point of the raw image represents the distance between the distance imaging sensor Sri and the subject.

[0153] According to one embodiment, the system includes an acquisition module aM configured to control a distance imaging sensor Sri to acquire at least one raw image including at least a portion of the patient's torso, wherein each point in the raw image represents the distance between the distance imaging sensor Sri and the patient.

[0154] The system may include a communication module that controls image acquisition and reception of raw images via a physical or wireless connection between the distance imaging sensor Sri and a system for transmitting acquisition commands.

[0155] According to one embodiment, the system includes a distance imaging sensor Sri. A processor implementing the surface generation module and / or computing module can be connected to the distance imaging sensor via a USB 3.0 bus.

[0156] In a preferred embodiment, the distance imaging sensor Sri is a time-of-flight camera. The ToF camera can be a pulsed light camera using a pulsed light sensor or a continuous-wave modulated light camera using a continuous-wave modulated sensor. The ToF camera can combine one or more laser beams (possibly mounted on a rotating mechanism) with a 2D photodetector array and a time-to-digital converter to produce an array of 1-D or 2-D depth values. The ToF camera can utilize laser diodes or LEDs of different wavelengths (particularly infrared or near-infrared).

[0157] According to one embodiment, the time-of-flight camera uses infrared illumination.

[0158] Several ToF cameras are actually available, such as the CamCube PMD with PMD (Photonic Mixer Device) technology, the SwissRanger 4000, or Microsoft's Kinect V1 sensor. According to one embodiment, the ToF camera is the Kinect v2 RGB-D camera, which uses multiple modulation frequencies (10–130 MHz) to achieve an excellent trade-off between depth accuracy and phase unfolding.

[0159] As described above, in a preferred embodiment, the distance imaging sensor Sri is placed in front of the patient's chest, specifically fixed to the ceiling above the patient. This arrangement of the distance imaging sensor Sri avoids introducing additional objects into the patient's surrounding environment. This is advantageous because an unobstructed environment allows for easier access for healthcare personnel during patient interventions.

[0160] According to one embodiment, the system includes a calibration module configured to apply a rotation matrix to the original image to align the patient's chest in the original image with the xy-plane of a time-of-flight camera. The calibration module is particularly configured to implement steps of the method of the invention related to the calibration steps described in the above embodiments.

[0161] According to one embodiment, the system includes a surface generation module sgM.

[0162] According to one embodiment, the surface generation module sgM is configured to perform preliminary operations on the original image, including filtering the original image to remove noise originating from other objects in the scene.

[0163] According to one embodiment, the surface generation module sgM is configured to generate a surface image of at least a portion of the patient's torso surface through surface interpolation of the original image. The surface image can be obtained using a base spline function via this module. The surface generation module sgM is particularly configured to implement the steps of the method of the invention related to the surface image generation step 120 described in the above embodiments.

[0164] According to one embodiment, the system includes a computation module cM configured to calculate trunk respiratory parameters such as tidal volume and respiratory rate based on morphological changes on surface images. In this embodiment, the computation module cM is configured to calculate a time-varying respiratory signal as the spatial average of the difference between depth values ​​of a surface image at a given time and depth values ​​of a reference surface image within a given region of interest defined on the subject's trunk, and to calculate tidal volume as the respiratory signal multiplied by the maximum value of the surface area of ​​the region of interest over one respiratory cycle.

[0165] The calculation module cM is specifically configured to implement the method of the present invention involving steps 130 and 140 for estimating tidal volume as described in the above embodiments.

[0166] According to one embodiment, the calculation module cM is configured to further calculate the respiratory rate based on the detection of the inspiratory peak in the respiratory signal. The calculation module cM is specifically configured to perform the calculation operation to obtain the respiratory rate according to the method embodiment described above with respect to step 150 in this specification.

[0167] According to one embodiment, the system includes an output module configured to provide the respiratory parameters as output. The output information may be represented as visual or auditory output, indicating results such as lung volume, respiratory rate, and / or tidal volume. The output module may be a display or microphone that is wirelessly connected or not connected to the system.

[0168] According to one embodiment, the system is a data processing system such as a dedicated circuit or a general-purpose computer device (i.e., a processor), configured to receive raw images and perform the operations described in the above embodiments. The computer device may include a processor and a computer program. The data processing system may include, for example, one or more servers, a motherboard, a processing node, a personal computer (portable or non-portable), a personal digital assistant, a smartphone, a smartwatch, a smart bracelet, a mobile phone, other mobile devices having at least one processor and memory, and / or other devices providing one or more processors at least partially controlled by instructions.

[0169] According to one embodiment, the computer device includes a network connection capable of remotely implementing the method according to the invention.

[0170] In one example, the input (i.e., raw data) is sent to a data processing unit such as an integrated processor, where software and hardware are used to process the signals.

[0171] Outputs (breathing characteristics, alarms) are transmitted to a computing platform or user system (computer or tablet) via wireless or wired networks.

[0172] Example

[0173] The invention is further illustrated by the following examples.

[0174] Example 1 - Patient simulator model:

[0175] Materials and Methods

[0176] The patient simulator model was connected to a ventilator capable of delivering volumes of 100 to 500 ml at different respiratory rates (12 to 50). The model had two operating modes: a symmetrical mode allowing simultaneous airflow into both lungs and an asymmetrical mode allowing airflow into only one lung (the right lung). The first mode was used to assess respiratory rate (RR) and tidal volume (Vt) of the model's trunk, while the second mode was used to assess regional lung function. Recordings from a total of 21 models over 10 minutes were analyzed, and respiratory measurements acquired per minute were calculated. The estimated parameters were compared to reference parameters (ventilator settings).

[0177] result

[0178] The results obtained through analysis of the model records are reported. Figure 7In (a) and (b), the scatter plots and regression lines illustrate the degree of agreement between the reference and estimated RR. The correlation coefficient shows a high value (r = 0.99; p < 0.001), with a low deviation (0 bps) and bias (±1.86 bps). The mean error between the reference and estimated RR shows a value of 1.6 ± 0.9 bps, with a minimum of 0 bps and a maximum of 3 bps.

[0179] Figure 8 (a) and (b) show the correlation between the reference tidal volume and the estimated tidal volume. Scatter plots and linear regression indicate high consistency (r = 0.99; p < 0.001), with a low deviation (7.90 ml) and an acceptable bias (±22.03 ml). The mean error is 18.0 ± 14.5 ml, with a minimum and maximum difference of 1 ml and 45 ml, respectively.

[0180] Figure 9 The model's asynchronous mode (allowing air to pass through only the right lung) is shown for the assessment of regional volume. The ventilator settings were set to 250 ml and 22 cpm.

[0181] The volume curves show that the air delivered by the ventilator was completely inhaled into the right lung. Although a slight fluctuation appeared in the left lung volume curve due to noise in the depth measurements, the signal had a very low amplitude and no obvious pattern. For all tests performed, the right chest contributed 87.2% to the total volume.

[0182] Example 2 - Patient:

[0183] Materials and Methods

[0184] This example involves a clinical assessment of 16 mechanically ventilated patients (10 men and 6 women) admitted to the ICU at Brest University Hospital.

[0185] Invasive ventilation procedures typically involve the use of sedatives and anesthetics to ensure patient safety and optimize air exchange. This helps prevent desynchronization with the ventilator and intolerance to endotracheal intubation. Depending on the level of sedation or consciousness, spontaneous ventilation may not be maintained. The commonly used ventilation mode is assisted controlled ventilation (ACV), where a patient uses pressure support ventilation (PSV) to assist and support their spontaneous breathing. Ventilator settings are defined based on the patient's physiology and pathology. Table 1 summarizes the patient's physiological characteristics.

[0186]

[0187] Table 1

[0188] During the assessment phase, respiratory parameters were calculated using the system's automated, continuous measurements. A total of 216 records were analyzed, assuming a monitoring time of 10 to 20 minutes per patient. The estimated RR and Vt parameters were compared to reference values ​​provided by the ventilator (considered the gold standard). However, a reference for the zone volume (Vr) parameter was unavailable because it could not be measured by the ventilator. Comparisons between the estimated and reference parameters were based on linear correlation using Pearson analysis. The Pearson correlation coefficient is a statistical measure of the strength of a linear relationship between paired data. It requires that the variables be normally distributed, continuous, and have a linear relationship. The normality of the data was checked using the D'Agostino and Pearson combined normality test.

[0189] In addition, the mean error (mean absolute difference between reference and estimated values ​​± standard deviation (std)) is calculated to assess the accuracy and reliability of this method.

[0190] Finally, to evaluate the pose estimation algorithm in a specific clinical ICU setting, the success / failure rate of the algorithm in detecting patients' chests was measured. A standard method for measuring the success / failure rate is to calculate the number of successful / failed detections over the total number of runs.

[0191] result

[0192] The records of 216 ICU patients were analyzed. Figure 10 The degree of agreement between the estimated RR and the reference RR was reported. Scatter plots and linear regression showed a high correlation (r = 0.95; p < 0.0001) with low deviation (0.39 bps) and bias (±1.7 bps).

[0193] RR estimation results showed that 84.7% of the errors were less than or equal to 2 bps. Only 4.1% of the measurements had errors greater than 3 bps. The minimum and maximum biases were 0 and 5 bps, respectively. The mean error calculated from the total patient records showed a value of 1.3 ± 1.1 bps. Vt estimation in the patient population, such as... Figure 11 As shown, this demonstrates a strong correlation with the reference value (r = 0.90; p < 0.0001), with low deviation (-5.3 ml) and bias (< ± 23.7 ml).

[0194] These results indicate that 69.0% of the estimated Vt values ​​were less than 25 ml. Only 6.4% of the measurements were greater than 35 ml. The mean error calculated from the total patient records was 19.6 ± 14.2 ml.

[0195] To assess the degree of change in the measurement, Figure 12 and Figure 13 The mean error (mean difference ± standard deviation) calculated based on the records for each patient was reported separately.

[0196] Figure 12 The mean error of the RR estimate for each patient was less than 3 bps. The minimum and maximum standard deviations were 0.4 and 1.49 bps, respectively. The minimum error and standard deviation (1.1 ± 0.4 bps) were observed in the patient 1 record.

[0197] Vt estimates for each patient, such as Figure 13 As shown, the error for each patient was less than 28 ml. The minimum and maximum standard deviations were 5.0 and 18.2 ml, respectively. The minimum error and standard deviation (8.5 ± 6.3 ml) were observed in the patient 1's record, while a larger error and standard deviation (26.7 ± 17.2 ml) were observed in the patient 11's record.

[0198] Other results from regional surveillance are shown in the figure below. However, quantitative analysis was not possible due to a lack of clinically available non-invasive equipment for regional surveillance. Figure 14 This shows an example of monitoring the left and right lung regions in two ICU patients. Figure 14 (a) The volume-time curves of the left lung (black curve) and right lung (gray curve) show a good volume distribution match, while Figure 14 (b) shows the uneven volume distribution between the left and right sides.

[0199] Figure 15 Another example shown illustrates regional monitoring of the chest and abdominal areas in two ICU patients. Figure 15 In (a), the volume-time curves are synchronous, while Figure 15 (b) shows asynchronous movement between the thoracic cavity and the abdomen.

Claims

1. A computer-implemented method (100) for estimating respiratory parameters of a subject, wherein the method comprises: - receiving a set of acquisitions obtained from a range imaging sensor (Sri), the set of acquisitions comprising at least one raw image (110) of at least a portion of the subject's torso and a reference surface image, wherein each point of the at least one raw image represents a distance between the range imaging sensor (Sri) and the subject, and wherein the reference surface image comprises depth values; - for each raw image of the at least one raw image, generating a surface image (120) of at least a portion of the surface of the subject's torso by surface interpolation of the at least one raw image; - estimating a respiratory signal over time by: • for each generated surface image, computing a difference between depth values of the surface image at a given time of interest region (ROI) and depth values of the reference surface image, the region of interest (ROI) comprising at least a portion of the subject's torso; and • computing a spatial average of the computed differences; - estimating respiratory parameters (140) comprising at least a lung volume over time by multiplying the respiratory signal by the surface of the region of interest; and - providing the respiratory parameters as output.

2. The method of claim 1, further comprising estimating a tidal volume as a difference between a maximum and a minimum of the respiratory signal multiplied by the surface of the region of interest over one respiratory cycle.

3. The method of any one of claims 1 or 2, further comprising estimating a respiratory frequency (150) computed from a detection of an inspiration peak in the respiratory signal.

4. The method of any one of claims 1 or 2, wherein, The surface images are obtained using a basis spline function.

5. The method of any one of claims 1 or 2, further comprising a step of filtering the at least one raw image to remove noise originating from other objects in the scene (111).

6. The method of any one of claims 1 or 2, wherein, The at least one raw image is obtained from a range imaging sensor that is a time-of-flight (ToF) camera.

7. The method of claim 6, wherein, The method further comprises a calibration step of applying a rotation matrix to the at least one raw image in order to align the subject's torso in the at least one raw image with the xy plane of the time-of-flight camera.

8. The method of any one of claims 1 or 2, wherein, The at least one raw image is obtained from a range imaging sensor (Sri) placed in front of the subject's torso.

9. A computer program product comprising instructions and a computer program containing said instructions, characterized in that The instructions, when the program is executed by a computer, cause the computer to perform the steps of the method of any one of claims 1 to 8.

10. A non-transitory computer readable medium having stored thereon instructions and a computer program containing the instructions, characterized in that The instructions, when the program is executed by a computer, cause the computer to perform the steps of the method of any one of claims 1 to 8.

11. A system (S) for estimating respiratory parameters of a subject, comprising: - an input module (aM) configured to receive a set of acquisitions obtained from a range imaging sensor (Sri), said set of acquisitions comprising at least one raw image of at least a portion of a subject's torso and a reference surface image, wherein each point of said at least one raw image represents a distance between the range imaging sensor (Sri) and the subject, and wherein said reference surface image comprises depth values; - a surface generation module (sgM) configured to generate, for each raw image of said at least one raw image, a surface image of at least a portion of a surface of the subject's torso by surface interpolation of said at least one raw image; - a computation module (cM) configured to: • estimate a respiratory signal over time by: • for each generated surface image, computing a difference between depth values of the surface image at a given time of a given region of interest (ROI) and depth values of said reference surface image; • computing a spatial mean of the computed differences; • computing a respiratory parameter comprising at least a lung volume over time by multiplying the respiratory signal by the surface of the region of interest; - an output module configured to output said respiratory parameter.

12. The system of claim 11, wherein, The computation module is configured to also compute a respiratory frequency computed from detection of an inspiration peak in the respiratory signal.

13. The system of any one of claims 11 or 12, wherein, The system further comprises a range imaging sensor (Sri).

14. The system of any one of claims 11 or 12, wherein, The range imaging sensor (Sri) is a time-of-flight camera and the system further comprises a calibration module configured to apply a rotation matrix to said at least one raw image in order to align the subject's torso in said at least one raw image with an xy plane of the range imaging sensor (Sri).

15. The system of any one of claims 11 or 12, wherein, The computation module (cM) is further configured to compute a tidal volume as a difference between a maximum and a minimum of the respiratory signal multiplied by the surface of the region of interest over one respiratory cycle.

Images