Method and system for ventilation system monitoring

Abstract

Methods and systems are provided for a ventilation system (100). In one example, a method includes obtaining (504, 602), with the ventilation system, one or more patient ventilation parameter images of a patient while the patient is performing mechanical ventilation; obtaining one or more reference ventilation parameter images; processing each patient ventilation parameter image and each reference ventilation parameter image with at least one comparison model to characterize at least one feature in each patient ventilation parameter image and each reference ventilation parameter image, including converting each patient ventilation parameter image and each reference ventilation parameter image to a binary mask; identifying (508, 606) a deviation between the patient ventilation parameter image and the corresponding reference ventilation parameter image based on the at least one feature; and in response to the identification, outputting (510, 608) a notification indicative of the deviation.

Description

Technical Field

[0001] The implementation scheme of the subject matter disclosed in this article involves assisted ventilation for subjects. Background Technology

[0002] During events in which subjects (such as patients undergoing surgery or requiring anesthesia) require ventilatory support, ventilation systems can be used to provide the necessary gas exchange in the lungs. Ventilation systems can have relatively complex configurations for delivering oxygen to and removing carbon dioxide from the lungs of a subject, and can rely on sensor-based microprocessor-based controls, valves, flow controllers, and various other components. Therefore, mechanical ventilation can be provided to the subject, and the flow of gas in and out of the subject can be monitored and controlled by the ventilation system and one or more clinicians (such as anesthesiologists). Summary of the Invention

[0003] In one embodiment, a method for a ventilation system includes: acquiring one or more patient ventilation parameter images of a patient while the patient is receiving mechanical ventilation using the ventilation system; acquiring one or more reference ventilation parameter images; processing each patient ventilation parameter image and each reference ventilation parameter image using at least one comparison model to characterize at least one feature in each patient ventilation parameter image and each reference ventilation parameter image, the processing including converting each patient ventilation parameter image and each reference ventilation parameter image into a binary mask; identifying a deviation between the patient ventilation parameter image and the corresponding reference ventilation parameter image based on at least one feature in each patient ventilation parameter image and each reference ventilation parameter image; and outputting a notification indicating the deviation in response to the identification.

[0004] It should be understood that the above brief description is provided to introduce selected concepts further described in the detailed embodiments in a simplified form. This is not intended to identify key or essential features of the claimed subject matter, the scope of which is uniquely defined by the claims following the detailed embodiments. Furthermore, the claimed subject matter is not limited to specific implementations that address any shortcomings mentioned above or in any part of this disclosure. Attached Figure Description

[0005] The invention will be better understood by referring to the following description of non-limiting embodiments, in which:

[0006] Figure 1 An example of a medical system used to provide respiratory support to a subject is shown.

[0007] Figure 2 It shows Figure 1 A diagram of the healthcare system.

[0008] Figure 3It shows including Figures 1 to 2 A block diagram of the ventilation system of a medical system.

[0009] Figure 4 This is a flowchart illustrating an advanced method for monitoring ventilation systems.

[0010] Figure 5 This is a flowchart illustrating a method for monitoring the ventilation system using initial patient waveforms and vital capacity measurements relative to annotated images via multiple comparative models.

[0011] Figure 6 This is a flowchart illustrating a method for monitoring the ventilation system using subsequent patient waveform and vital capacity measurement images relative to a baseline image via multiple comparative models.

[0012] Figure 7 This is a flowchart illustrating a method for confirming a patient's readiness to be weaned off the ventilation system.

[0013] Figure 8 It is being executed Figures 4 to 7 An exemplary graphical user interface that can be displayed on the equipment of the ventilation system during the method.

[0014] Figures 9 to 16 An exemplary image of vital capacity measurement is shown that can be analyzed via the comparative model disclosed herein.

[0015] Figures 17 to 21 An exemplary waveform image is shown that can be analyzed via the comparative model disclosed herein. Detailed Implementation

[0016] The following describes various implementations of ventilation systems, which may include anesthesia machines and ventilators. During certain procedures, or based on certain patient conditions, patients may be unable to breathe spontaneously and achieve adequate gas exchange, thus requiring mechanical ventilation via a ventilation system. During mechanical ventilation, various patient and machine parameters are measured and displayed for one or more clinicians to monitor the patient, such as respiratory rate (RR), positive end-expiratory pressure (PEEP), and fractional oxygen inspired (FOS). I O2), exhaled oxygen fraction (F E O2), respiratory gas flow rate (F), tidal volume (V) T Temperature (T), airway pressure (P) awSome parameters can be displayed as time-series data in waveform form, such as pressure, flow rate, and volume waveforms. Additionally, some parameters can be combined into spirometry plots or images, such as pressure / flow rate and flow / volume spirometry images, which are also displayed. Clinicians supervising the patient can monitor the displayed parameters to determine if the patient is stable and if the ventilation system is functioning as expected. If the clinician notices any deviation from the expected parameters, they can initiate various actions, such as adjusting the ventilation system settings.

[0017] However, the above approach assumes that all clinicians can correctly interpret all the parameters displayed. Particularly in the case of spirometry images, many clinicians cannot correctly interpret each type of spirometry image and may fail to recognize changes in spirometry images that indicate potential problems, such as laryngeal mask air leakage, kinked endotracheal tubes, or other issues. Furthermore, even for clinicians who can correctly interpret all displayed parameters, it is practically impossible to continuously monitor all parameters and identify deviations indicating the aforementioned problems, as parameter changes can occur slowly (and therefore be difficult to detect) or may manifest differently for different types of patients and different respiratory system settings. For example, a clinician may not know or remember the baseline / normal values ​​for each parameter for a given patient under a given set of ventilation system settings, and therefore may be unable to identify when any parameter changes, let alone when it changes to the extent that action is recommended. Since most clinicians, especially supervising anesthesiologists or other trained physicians, may supervise multiple patients simultaneously, the aforementioned problems can be further complicated, and clinicians may not be able to detect problems quickly enough.

[0018] Therefore, the embodiments disclosed herein address these issues by performing automatic comparisons between predefined baseline or benchmark real ventilation parameter images (e.g., waveform and vital capacity measurement images) and current patient ventilation parameter images using one or more comparison models. These comparison models are configured to identify and quantify various features in the ventilation parameter images and output any detected changes in those features. At the initial stage of mechanical ventilation (e.g., when the patient is initially intubated), benchmark real waveform and vital capacity measurement images can be obtained from a database of annotated waveform and vital capacity measurement images reflecting real or simulated waveform and vital capacity measurement images generated for patients with different demographic characteristics (age, sex, body size, etc.) and different respiratory system settings. The annotated waveform and vital capacity measurement images can be annotated to reflect whether the lung function of the real or simulated patient is satisfactory and whether the ventilation system exhibits normal or declining performance. One or more comparison models can be used to compare a patient’s initial waveform and vital capacity measurement images with annotated waveform and vital capacity measurement images matched for demographic and ventilation system parameters to identify whether there are any significant deviations between the patient’s initial waveform and vital capacity measurement images and the annotated waveform and vital capacity measurement images, which may indicate, for example, a problem with the initial ventilation system settings.

[0019] After obtaining the patient's initial waveform and vital capacity measurement images, assuming no deviation between the initial images and matched annotated images, the initial images can be saved as baseline images for comparison with subsequent images. If any deviation is detected, one or more clinicians can be notified via displayed alerts, for example, prompting them to adjust ventilation system settings. Previous baseline images can be replaced with current images as a new comparison method (assuming the changed ventilation system settings reflect the ideal situation for the patient). Each time the baseline images are replaced, previous images can be moved to local or remote storage to allow for further refinement of one or more comparison models.

[0020] One or more comparison models can also be configured to detect whether the patient is breathing spontaneously or has stopped breathing. If one or more comparison models detect that the patient is breathing spontaneously, a weaning protocol can be initiated. The weaning protocol may include analyzing patient video / image frames via one or more of the comparison models to determine whether the patient is able to follow commands, and performing rule-based checks on patient monitoring parameters such as blood oxygen saturation and heart rate. If the patient is able to follow commands and the parameters pass the rule-based checks, one or more clinicians may be notified to recommend weaning the patient.

[0021] Therefore, the above method allows for the monitoring of multiple parameters in mechanically ventilated patients, with simultaneous monitoring of multiple parameters at a higher rate than that achievable solely by relying on clinicians. The patient's own waveform and spirometry images can be used as a baseline to detect subtle or rare deviations in the waveform and / or spirometry images. This allows for automated monitoring of patient status and the identification of any changes in patient status, which will be noticed by the clinician monitoring the patient. The baseline images can be automatically updated whenever ventilation system settings change, allowing for the detection of changes in patient status based on the most appropriate image, reducing unnecessary alarms and improving monitoring efficiency.

[0022] Before further discussing the systems and methods for automatically monitoring ventilation systems and patients receiving mechanical ventilation via these systems, a general description of medical systems configured to provide ventilation support is provided. The accompanying drawings illustrate diagrams of functional blocks for various embodiments. Functional blocks do not necessarily indicate divisions between hardware circuits. Thus, for example, one or more functional blocks (e.g., processors or memory) may be implemented in a single piece of hardware (e.g., a general-purpose signal processor or a block of random access memory, a hard disk, etc.) or in multiple pieces of hardware. Similarly, a program may be a standalone program, may be included as a subroutine in an operating system, may be a function in an installed software package, etc. It should be understood that the various embodiments are not limited to the arrangements and tools shown in the drawings.

[0023] For example, in Figures 1 to 2 The image shows a medical system 10, which may include an anesthesia machine 14 with a ventilator 16. Therefore, the medical system 10 may be referred to as a ventilator and / or anesthesia machine. The ventilator 16 may have suitable connectors, such as a first connector 18 and a second connector 20, for connection to an inspiratory branch 22 and an expiratory branch 24 of a breathing circuit 26 leading to a patient 12, wherein the inspiratory branch 22, the expiratory branch 24, the breathing circuit 26, and the patient 12 are as follows: Figure 2 As shown. The ventilator 16 and the breathing circuit 26 can cooperate to deliver breathing gas to the patient 12 via the inspiratory branch 22 and receive the gas exhaled by the patient 12 via the expiratory branch 24.

[0024] The ventilator 16 can also be equipped with a manual resuscitator 28, such as Figure 1 As shown, this is used for manual ventilation of patient 12. In one example, the manual resuscitator 28 may be a bag valve mask (BVM), which may also be referred to as a bag. For example, the manual resuscitator 28 may be filled with a respiratory gas, such as oxygen, anesthetic gas, etc., and is manually squeezed by a clinician (not shown) to deliver the respiratory gas to patient 12. Using the manual resuscitator 28, or “bagging the patient,” allows the clinician to manually and / or immediately control the delivery of respiratory gas to patient 12. The clinician may also sense the patient 12’s breathing and / or lungs 30 (e.g., lungs) based on sensory feedback during the operation of the manual resuscitator 28. Figure 2 The ventilator 16 may also provide a bag-to-ventilation (BTV) switch 32 for switching and / or alternating between manual and automatic (e.g., mechanical) ventilation when the manual resuscitator 28 is provided.

[0025] The medical system 10, particularly the processing terminal 36 and / or processing subsystem 58 of the medical system 10 (which may be implemented on the anesthesia machine 14), can be accessed from, for example... Figure 2 The sensor 34, shown as being associated with the ventilator 16 and the anesthesia machine 14, receives input for its subsequent processing. The processed input may be displayed on a monitor 38 (e.g., on the anesthesia machine 14). Representative data received from the sensor 34 may include, for example, inspiratory time (T). I ), expiratory time (T) E ), natural exhalation time (T) EXH ), respiratory rate (RR), inspiratory time to expiratory time (I:E) ratio, positive end-expiratory pressure (PEEP), and fractional oxygen inhaled (F). I O2), exhaled oxygen fraction (F E O2), respiratory gas flow rate (F), tidal volume (V) T Temperature (T), airway pressure (P) aw ), arterial oxygen saturation level (S a O2), blood pressure (BP), pulse rate (PR), pulse oxygen level (S) p O2), exhaled CO2 level (F ET CO2), concentration of inhaled anesthetics (C I Reagent), concentration of inhaled anesthetic (C) E Reagent), arterial blood oxygen partial pressure (P) a O2), partial pressure of carbon dioxide in the artery (P) a(CO2), etc. Therefore, sensor 34 may be included in the breathing circuit 26, ventilator 16, anesthesia machine 14, and / or on patient 12. For example, patient 12 may be connected to a patient monitor that includes features for detecting BP, PR, and S. p O2 sensors, such as Figure 3 As shown below and explained in more detail.

[0026] Now for specific reference Figure 2 The ventilator 16 delivers breathing gas to the patient 12 via a breathing circuit 26. Therefore, the breathing circuit 26 includes an inspiratory branch 22 and an expiratory branch 24. In some examples, one end of each of the inspiratory branch 22 and the expiratory branch 24 is connected to the ventilator 16, while the other end is connected to a Y-connector 40, which can then be connected to the patient 12 via the patient branch 42. An interface 43 may also be provided to secure the patient 12's airway to the breathing circuit 26 and / or prevent gas leakage therefrom.

[0027] The ventilator 16 may also include electronic control circuitry 44 and / or pneumatic circuitry 46. Specifically, various pneumatic elements of the pneumatic circuitry 46 can deliver breathing gas to the lungs 30 of the patient 12 via the inspiratory branch 22 of the breathing circuit 26 during inhalation. During exhalation, breathing gas can be expelled from the lungs 30 of the patient 12 and enter the expiratory branch 24 of the breathing circuit 26. This process can be iteratively activated by the electronic control circuitry 44 and / or pneumatic circuitry 46 in the ventilator 16, which can establish various control parameters, such as the number of breaths per minute administered to the patient 12, tidal volume (V... T Maximum pressure, etc., can characterize the mechanical ventilation supplied by the ventilator 16 to the patient 12. Therefore, the ventilator 16 can be microprocessor-based and can incorporate suitable memory operations to control the exchange of gases in the lungs within the breathing circuit 26 connected to the patient 12 and the ventilator 16, as well as between the patient and the ventilator.

[0028] Various pneumatic elements of the pneumatic circuit 46 may also include a pressurized gas source (not shown) that can operate via a gas concentrating subsystem (not shown) to provide breathing gas to the lungs 30 of the patient 12. The pneumatic circuit 46 may provide breathing gas directly to the lungs 30 of the patient 12, as may be used in chronic and / or intensive care applications, or the pneumatic circuit 46 may provide a driving gas to compress a bellows 48 containing breathing gas (e.g., Figure 1 (As shown). Subsequently, the bellows 48 can supply breathing gas to the lungs 30 of the patient 12, such as in anesthesia applications. In either case, breathing gas can repeatedly pass from the inspiratory branch 22 through to the Y-connector 40 and to the patient 12, and then return to the ventilator 16 via the Y-connector 40 and the expiratory branch 24.

[0029] exist Figures 1 to 2 In the illustrated embodiment, one or more sensors in sensor 34, when placed in breathing circuit 26, can also provide feedback signals back to the electronic control circuitry 44 of ventilator 16 via a feedback loop. More specifically, the signal in the feedback loop can be, for example, proportional to the gas flow and / or airway pressure in the patient branch 42 of the lung 30 leading to patient 12. The concentrations of inhaled and exhaled gases (e.g., oxygen (O2), carbon dioxide (CO2), nitrous oxide (N2O), and inhaled anesthetics), flow rates (including, for example, vital capacity measurements), and gas pressure levels can be captured by sensor 34, as well as the duration of the time period between when ventilator 16 allows patient 12 to inhale and exhale and when the patient's natural inspiratory and expiratory flow ceases.

[0030] Therefore, the electronic control circuit 44 of the ventilator 16 can also control the display of digital and / or graphical information (such as...) from the breathing circuit 26 on the monitor 38 of the medical system 10. Figures 1 to 2 (as shown), and other patient 12 and / or system 10 parameters from other sensors 34 and / or processing terminal 36 (such as... Figure 1 (As shown). In other embodiments, various components may also be integrated and / or separated as needed and / or desired.

[0031] Processing subsystem 58 can be located in Figure 1 The processing terminal 36 may include various electronic components, such as hardware, for receiving and transmitting signals and processing signals. For example, the processing subsystem 58 may include a controller (e.g., a processor) configured to receive signals from sensors 34 (which may include pressure sensors, flow sensors, sensors monitoring the state of valves and switches, etc.) and, in response to the sensor signals, send control signals to actuators of the medical system (such as pressure regulators, valves, and switches, etc.). The controller may be a microcomputer, including a microprocessor unit, input / output ports, and electronic storage media (including non-transitory memory) for storing executable programs and parameter setting values. The controller may be programmed with computer-readable data representing instructions executable to perform the methods described herein and other variations contemplated but not specifically listed.

[0032] Processing subsystem 58 can also coordinate and / or control, for example Figure 2The elements depicted include a ventilator setting signal 54 and a ventilator control signal 56. The ventilator setting signal 54 and the ventilator control signal 56 can be used to control the operation of the ventilator 16 via electronic control circuitry 44 and pneumatic circuitry 46. The processing subsystem 58 can also be configured to output signals to display, control / display alarm 60 on monitor 38 and / or similar devices, and / or control / generate / display operator interface 62, which may include a graphical user interface (GUI) displayed on monitor 38, and one or more input devices 64, all of which are provided as needed and / or desired and appropriately interconnected.

[0033] To illustrate, it is described Figures 1 to 2 The components may include various other components that can be integrated and / or separated based on requirements and / or expectations. Other components may be provided, such as one or more power supplies for medical systems 10 and / or anesthesia machines 14 and / or ventilators 16, etc. (not shown).

[0034] Figure 3 A block diagram of a ventilation monitoring system 100 including a medical system 10 is shown. As described above, the medical system 10 includes a monitor 38 (e.g., a display device), an input device 64, a sensor 34, and a processing subsystem 58. The processing subsystem 58 can be configured to execute instructions stored in a memory 102 to implement the various functions described above and the methods disclosed herein. The memory 102 may store a clinical information processing module 104, a waveform processing module 106, an inference module 108, and an output generation module 110. The medical system 10 may also include a data storage device including a model library 112, baseline images 114, and annotated images 116.

[0035] Within the ventilation monitoring system 100, the medical system 10 can be operatively / communically coupled to various devices, either directly or via a network such as network 140. Figure 3As shown, multiple care provider devices 130 may be included as part of network 140 and / or directly coupled to healthcare system 10, from first care provider device 134, second care provider device 136, and up to nth care provider device 138. Each care provider device may include a processor, memory, communication module, user input device, display (e.g., screen or monitor), and / or other subsystems, and may take the form of a desktop computing device, laptop computing device, tablet computer, smartphone, or other device. Each care provider device may be adapted to send and receive encrypted data and display medical information (including medical images in suitable formats such as Digital Imaging and Communications in Medicine (DICOM) or other standards). Care provider devices may be located locally within the same healthcare facility as healthcare system 10 and substantially fixed in an appropriate location (e.g., at a nurses' station or in a patient's room), and / or located locally or remotely within the healthcare facility and configured to move with the care provider (e.g., a care provider's mobile device).

[0036] Medical system 10 may be communicatively coupled to one or more external patient devices 120. External patient devices 120 may include a patient monitor 122, a camera 124, and / or a speaker 126. Patient monitor 122 may include sensors configured to measure patient status, which are not included as part of medical system 10, such as sensors configured to measure one or more of blood oxygen saturation measurements (SpO2), heart rate (HR), and pulse rate (PR). Although external patient devices 120 are shown as being directly coupled to medical system 10, it should be understood that one or more external patient devices may be communicatively coupled to medical system 10 via network 140. Medical system 10 may also be communicatively coupled to data storage devices located separately from medical system 10, such as local data storage device 150 and / or remote data storage device 160.

[0037] The devices disclosed herein, such as care provider devices and / or aspects of medical system 10, may each include a communication module, a memory, and a processor to store and execute the methods disclosed herein and to send and receive communications, graphical user interfaces, medical data, and other information. For example, medical system 10 may include a communication module 118.

[0038] Each communication module facilitates the transmission of electronic data within and / or between one or more systems. Communication via the communication module can be implemented using one or more protocols. In some examples, communication via the communication module occurs according to one or more standards (e.g., Digital Imaging and Communication in Medicine (DICOM), Health Class 7 (HL7), ANSI X12N, etc.). The communication module can be a wired interface (e.g., data bus, Universal Serial Bus (USB) connection, etc.) and / or a wireless interface (e.g., radio frequency, infrared, near field communication (NFC), etc.). For example, the communication module can use any past, present, or future communication protocol (e.g., wired local area network (LAN), wireless LAN, wide area network (WAN), etc.) It communicates via USB 2.0, USB 3.0, etc.

[0039] Each memory (including memory 102, memory of each care provider device, local data storage device 150, and remote data storage device 160) may include one or more memory structures, such as optical memory devices, magnetic memory devices, or solid-state memory devices, for storing programs and routines executed by a processor to implement the various functionalities disclosed herein, and for storing annotated waveform images, annotated spirometry images, baseline waveform images, and baseline spirometry images, as discussed in more detail below. The memory may include any desired type of volatile and / or non-volatile memory, such as, for example, static random access memory (SRAM), dynamic random access memory (DRAM), flash memory, read-only memory (ROM), etc. The processor may be, for example, any suitable processor, processing unit, or microprocessor. The processor may be a multiprocessor system and therefore may include one or more additional processors that are identical or similar to each other and communicatively coupled via an interconnect bus.

[0040] As used herein, the terms “sensor,” “system,” “unit,” or “module” can include hardware and / or software systems that operate to perform one or more functions. For example, a sensor, module, unit, or system can include a computer processor, controller, or other logic-based device that performs operations based on instructions stored on a tangible, non-transitory computer-readable storage medium, such as computer memory. Alternatively, a sensor, module, unit, or system can include a hardwired device that performs operations based on the device’s hardwired logic. The various modules or units illustrated in the figures can represent hardware that operates based on software or hardwired instructions, software that instructs the hardware to perform operations, or a combination thereof.

[0041] The terms "system," "unit," "sensor," or "module" can include or represent hardware and associated instructions (e.g., software stored on a tangible, non-transitory computer-readable storage medium, such as a computer hard disk drive, ROM, RAM, etc.) that perform one or more of the operations described herein. Hardware can include electronic circuitry that includes and / or is connected to one or more logic-based devices, such as microprocessors, processors, controllers, etc. These devices can be readily available devices that are appropriately programmed or instructed to perform the operations described herein according to the instructions described above. Additionally or alternatively, one or more of these devices may be hardwired to logic circuitry to perform these operations.

[0042] One or more of the devices described herein may be implemented via the cloud or other computer networks. For example, model store 112 and inference module 108 are shown residing on medical system 10, but it should be understood that aspects of medical system 10 (e.g., model store 112 and / or inference module 108) may be implemented on remote devices (e.g., servers) or distributed across multiple devices, such as across multiple servers.

[0043] As will be explained in more detail below, the ventilation monitoring system 100 can be configured to monitor patient ventilation parameter images, such as waveform images and vital capacity measurement images, generated by the medical system 10 for a patient undergoing mechanical ventilation. Monitoring may include comparing the patient ventilation parameter images with annotated, demographically and system-set matched ventilation parameter images to identify any deviations between the patient ventilation parameter images and expected (e.g., baseline true) ventilation parameter images. For example, ventilation waveform and vital capacity measurement images can be collected for a broad patient population ranging from neonates to adults under different ventilation settings, such as different airway resistance, peak inspiratory pressure (Ppeak), plateau pressure (Plat), positive end-expiratory pressure (PEEP), tidal volume (TV), respiratory rate (RR), inspiratory time to expiratory time (I:E) ratio, etc. These waveform and vital capacity measurement images can be annotated by a professional anesthesiologist / clinician and stored in an image repository, such as in the cloud or on a server. Annotated waveform and vital capacity measurement images can be categorized based on different patient lung conditions and breathing circuit conditions, such as good lungs and good breathing circuit (e.g., no problems such as leaks, endotracheal tube (ET tube) blockage, expiratory limb blockage, etc.), good lungs but defective breathing circuit, good lungs during weaning and sufficient tidal volume during spontaneous breathing, and no spontaneous breathing. These annotated waveform and vital capacity measurement images can be used in the initial phase of ventilation support to detect any deviation from the expected patient condition by comparing the patient's waveform and vital capacity measurement images with the annotated waveform and vital capacity measurement images via a comparison model suite.

[0044] For example, when initiating mechanical ventilation on a patient, the anesthesiologist may set the ventilation settings of medical system 10 based on the patient's condition (e.g., based on patient information such as age, sex, height, and weight, and based on the patient's diagnosis, reason for mechanical ventilation, etc.). Medical system 10 may include a clinical information processing module 104, which may be configured to receive and process patient information from various sources (e.g., electronic medical record systems, hospital information systems, and / or user input via input device 64) to determine the patient's condition. Waveform processing module 106 may generate patient waveform images and spirometry loop images for a given set of ventilation settings at specified time intervals (e.g., generating a new waveform and spirometry image every two minutes). Inference module 108 may invoke one or more comparison models that can be stored in model store 112 to compare the patient waveform and spirometry images with appropriate images from annotated waveform and spirometry images. In some examples, the annotated waveform and vital capacity measurement images may be stored locally on the medical system 10 as part of the annotated image 116, or the annotated waveform and vital capacity measurement images may be stored remotely and accessed during comparison.

[0045] If a significant deviation is observed between the patient's waveform and spirometry images and the annotated waveform and spirometry images, a notification describing the problem and (where known) the possible causes of the problem can be output to the anesthesiologist based on this deviation. For example, the output generation module 110 can generate a pop-up message that can be displayed on one or more of the monitor 38 and / or care provider devices (such as via the graphical user interface 135). Additionally, the patient's waveform and spirometry images can be sent, for example, to a designated storage location (e.g., local data storage device 150 and / or remote data storage device 160) to be used as annotated waveform and spirometry images. When a significant deviation is observed between the patient's waveform and spirometry images and the annotated waveform and spirometry images, the anesthesiologist can take actions to correct the observation, such as adjusting the settings of the medical system 10, replacing leaking components, or disabling the ET (Electronic Toll Collection). After addressing these issues, newly generated patient waveform and spirometry images with updated ventilation settings are classified as baseline images, and the baseline images, along with the ventilation settings, are timestamped and stored in memory (e.g., in baseline image 114). Following this step, subsequent patient waveform and spirometry images are compared to the baseline waveform and spirometry images via one or more comparison models. Image generation and comparison with baseline images continue at specific time intervals until the patient's ventilation support is removed (e.g., until the patient is discharged from medical system 10).

[0046] Whenever a change is observed between the baseline images and a new set of waveform and spirometry images, a notification / alarm is sent to the anesthesiologist, showing the deviation in the images and (where known) the possible cause of the problem. The anesthesiologist can take action and resolve the issue by changing any ventilation settings, and the change in ventilation settings can be identified from the waveform and spirometry images. These waveform and spirometry images are now the new baseline and stored in memory (e.g., in baseline image 114). The previous baseline images and the patient waveform and spirometry images that triggered the change in ventilation settings are moved to local and / or remote data storage devices. This process continues, and if (e.g., via a comparison model) a patient's spontaneous breathing effort with sufficient tidal volume is observed, a weaning protocol can be executed (e.g., via inference module 108), which includes checking patient monitor data such as SpO2, blood pressure, etc., based on predetermined rules. If the patient monitor data passes the check, an alarm can be output to notify the anesthesiologist that the patient is ready to be weaned. If the anesthesiologist approves, another aspect of the weaning protocol can be implemented, including (e.g., via speaker 126) outputting an audio command instructing the patient to open their eyes or perform another task that can be recorded via camera 124. The video from the camera can be evaluated via a comparative model to determine if the patient is able to follow the audio command. Different patient facial images and videos can be stored (e.g., in local and / or remote data storage devices) and used to train the comparative model to detect changes in the patient's face.

[0047] In current ventilation systems, various patient / ventilation data, such as pressure waveforms, flow waveforms, and volume waveforms, are presented to clinicians. Pressure, flow, and / or volume data can be combined to generate spirometry loops. Interpretation of spirometry loops requires deep knowledge and strong clinical expertise. Currently, there is no available periodic baseline comparison data in ventilation systems to interpret these spirometry loops. Continuous monitoring of these waveforms and spirometry images is time-consuming and labor-intensive, and it is difficult to identify changes in ventilation parameters that could impact patient care. Therefore, the implementation scheme for continuous monitoring of patient waveforms and spirometry images disclosed herein is configured to assist clinicians in interpreting waveforms and spirometry images, diagnosing and identifying causes affecting ventilation, such as airway leaks, airway obstruction, changes in lung compliance, patient spontaneous breathing effort, intrinsic PEEP (autoPEEP), etc. The method disclosed herein monitors ventilation throughout the process and alerts clinicians whenever changes are detected in the waveforms and spirometry loops.

[0048] Briefly switch to Figure 8Examples of patient ventilation parameter images that can be generated and evaluated as disclosed herein are shown. One or more sensors of the ventilation monitoring system 100 (e.g., one or more sensors in sensor 34) can be configured to measure patient airway pressure, airway flow, and airway volume over time. The output from one or more sensors can be represented as time-series data, referred to as waveform images. Figure 8 An exemplary pressure waveform image 802 depicting airway pressure over time, an exemplary flow waveform image 804 depicting airway flow over time, and an exemplary volume waveform image 806 depicting airway volume over time are shown. Figure 8 Each waveform image shown depicts two respiratory cycles, but it should be understood that more or fewer respiratory cycles may also be depicted. The output from one or more sensors may be represented as spirometry images, including an exemplary pressure / volume spirometry image 808 that depicts airway volume as a function of airway pressure and an exemplary flow / volume spirometry image 810 that depicts airway flow rate as a function of airway volume. Figure 8 The lung capacity measurement image shown depicts the pressure / volume and flow / volume within a respiratory cycle.

[0049] return Figure 3 The ventilation monitoring system 100 can be used to monitor a patient's ventilation parameters during mechanical ventilation (e.g., when the patient is intubated and receiving respiratory support from the medical system 10), including pressure, flow, and volume waveforms, as described above and Figure 8 The pressure / volume and flow / volume vital capacity measurement loops shown are used to identify deviations from expected ventilation parameters using one or more comparative models. Medical system 10 may include instructions stored in memory (e.g., memory 102) that can be executed by one or more processors (e.g., processing subsystem 58) to implement [the desired ventilation parameters]. Figures 4 to 7 The methods described are shown and described below.

[0050] Figure 4 An advanced method 400 for monitoring patient ventilation is shown. Method 400 can be used with ventilation monitoring systems (such as...) Figure 3 This is achieved through a ventilation monitoring system 100, which includes a ventilation system (such as...). Figure 1 and Figure 2 (Medical system 10). Method 400 can be implemented according to instructions stored in non-transitory memory and executed by one or more processors (such as the memory and processor of medical system 10). At 402, method 400 includes performing an initial ventilation parameter analysis upon initiation of ventilation. (See below for reference.) Figure 5A more detailed description of initial ventilation parameter analysis is provided. In short, initial ventilation parameter analysis may include generating initial patient ventilation parameter images (e.g., waveform images and vital capacity measurement images) from the outputs of one or more sensors of the ventilation system once mechanical ventilation has been initiated with initial ventilation settings, as explained above and as... Figure 8 As shown. For example, waveform images may include pressure, flow, and volume waveforms depicting airway pressure, flow, and volume over time, respectively. Vital capacity measurement images may include pressure / volume and flow / volume vital capacity measurement loops, depicting airway volume as a function of pressure and airway flow as a function of volume, respectively, where each vital capacity measurement loop depicts data for one respiratory cycle. The initial patient ventilation parameter images can be compared (via one or more comparison models) to matched baseline true / annotated ventilation parameter images to determine if any deviations exist in the initial patient ventilation parameter images, indicating that adjustments to the initial ventilation settings may be necessary. If any deviations are detected, a notification can be output to the clinician monitoring the patient, allowing for adjustments to the ventilation settings. If no deviations are detected, the initial patient ventilation parameter images can be categorized as baseline images reflecting the expected / baseline waveforms and vital capacity measurement status / output for the patient. The baseline images can then be used for continued patient monitoring.

[0051] Therefore, at 404, method 400 includes continuing to monitor subsequent patient ventilation images upon instruction, which will be referenced below. Figure 6 To explain in more detail: After evaluating the initial waveform and spirometry images as described above and obtaining the patient's baseline images, the ventilation system can continue to generate waveform and spirometry images at a specified rate (e.g., once per minute, once every two minutes, etc.). Each set of waveform and spirometry images can be compared to the baseline images via one or more comparison models to determine if any (significant) deviations are detected. If any deviations are detected, a notification can be output to the clinician monitoring the patient, allowing adjustments to the ventilation settings.

[0052] The process of continuing to monitor patient ventilation parameter images may include determining whether any detected deviations (e.g., waveform or vital capacity measurement deviations) are due to spontaneous breathing, as indicated at 406. As will be explained in more detail below, as the patient's condition improves and / or as the procedures performed on the patient end (and therefore the need for mechanical ventilation decreases), the patient may begin to breathe spontaneously, which can be observed in waveform and vital capacity measurement images. Detection of spontaneous breathing can trigger the activation of a weaning protocol that guides clinicians in making decisions about when to wean the patient off mechanical ventilation.

[0053] At 408, method 400 determines whether a weaning protocol has been activated (e.g., due to the detection of spontaneous breathing). If the weaning protocol has not been activated, method 400 returns to 404 to continue monitoring the patient's ventilation parameter images upon indication. If the weaning protocol has been activated, method 400 proceeds to 410 to execute the weaning protocol, which will be referenced below. Figure 7 A more detailed description follows. The weaning protocol may include rule-based checks of patient parameters (e.g., SpO2, PR, etc.) and (e.g., via the ventilation system) automatic determination of whether the patient is able to follow commands. If the weaning protocol indicates that the patient may be ready to be weaned, a notification may be output, allowing the clinician monitoring the patient to wean the patient as needed. Method 400 then concludes.

[0054] Figure 5 This is a flowchart illustrating a method 500 for performing initial ventilation parameter analysis on a patient undergoing mechanical ventilation. Method 500 can be performed using a ventilation monitoring system (such as...) Figure 3 This is achieved through a ventilation monitoring system 100, which includes a ventilation system (such as...). Figure 1 and Figure 2 (Medical system 10). Method 500 may be implemented according to instructions stored in non-transitory memory and executed by one or more processors (such as the memory and processor of medical system 10). In some examples, method 500 may be implemented as part of method 400, such as at 402 of method 400.

[0055] At 502, method 500 includes obtaining ventilation system settings and patient information. As a non-limiting example, ventilation system settings may include whether mechanical ventilation is in pressure-target mode, volume-target mode, or flow-target mode; setting FiO2; setting EtO2; setting mechanical ventilation; setting inhaled anesthetics; and / or setting exhaled anesthetics. Patient information may include the patient's age, sex, weight, and height. Additionally, ventilation system settings and / or patient information may include measured ventilation parameters such as airway resistance, peak inspiratory pressure, plateau pressure, PEEP, tidal volume, RR, and inspiratory to expiratory time (I:E) ratio.

[0056] At 504, method 500 includes generating initial patient ventilation parameter images, such as one or more waveform images and / or one or more vital capacity measurement images. A first iteration of method 500 may include generating initial waveform and vital capacity measurement images (e.g., first waveform and vital capacity measurement images generated upon initiation of mechanical ventilation or once mechanical ventilation has been initiated and is operating in a stable state). The waveform and vital capacity measurement images may be generated based on the output of sensors from the ventilation system, such as sensor 34. For example, the sensor may output airway pressure, flow rate, and volume as time-series data. The waveform images may be generated by taking snapshots of the time-series data (e.g., generated by waveform processing module 106). The waveform images may capture pressure, flow rate, and volume within one or more respiratory cycles, respectively. The vital capacity measurement images may be generated by plotting volume as a function of pressure and flow rate as a function of volume for a respiratory cycle and generating a plotted image (e.g., via waveform processing module 106). The vital capacity measurement image can be generated from the same sensor data captured in the waveform image, or from sensor data output within a threshold time period (e.g., within one or two respiratory cycles) of the sensor data captured in the waveform image.

[0057] At point 506, the (initial) waveform and vital capacity measurement images are compared with selected annotated images (e.g., baseline real images) via one or more comparison models. The selected annotated images can be obtained from an image repository (such as...) that stores a collection of annotated images. Figure 3 The annotated images in the image repository (116) are obtained. The annotated images in the image repository can be waveform and spirometry images that capture expected or baseline true pressure, flow, and volume for various patient demographics and ventilation system settings. The annotated images in the image repository can also be waveform and spirometry images obtained from the ventilation system during ventilation of previous patients (where the previous patients cover all age ranges, sexes, and body types, and the ventilation systems cover various ventilation system settings). In some examples, the annotated images in the image repository may include waveform and spirometry images generated with an artificial lung. The annotated images in the image repository may be annotated with patient information, ventilation system settings, and expert annotations indicating the patient's lung condition (good or poor) and ventilation system condition (system without defects or system with one or more defects, where defects are marked).

[0058] Annotated images selected for comparison with patient waveform and vital capacity measurement images can be chosen based on patient information and ventilation system settings obtained at 502, such that the selected annotated images match the patient and ventilation system settings. For example, annotated images that demographically match the patient can be selected (e.g., if the patient is a male in his 50s who is 5 feet 10 inches tall and weighs 170 pounds, the selected annotated image could be a previous patient who was also male, in his 50s, and had similar height and weight). From the demographically matched annotated images, the final selected images can be chosen based on ventilation system settings that are the same or similar (which may include the same or similar measured ventilation parameters). To the extent possible, images of the same type of ventilation parameters can be selected, such that if the patient's ventilation parameter images include pressure, flow, and volume waveform images and pressure / volume and flow / volume vital capacity measurement images, then the selected annotated / benchmark real images that include pressure, volume, and flow waveform images and pressure / volume and flow / volume vital capacity measurement images are selected.

[0059] One or more comparison models (such as those stored in model store 112) may be configured to detect various anomalies in waveform and spirometry images, such as fault bias, prolonged hysteresis, invalid triggering, double triggering, open-loop circulation, volume increase and decrease, etc. One or more comparison models may detect these anomalies by comparing patient waveform and spirometry images with selected annotated waveform and spirometry images (e.g., comparing a patient pressure waveform image with an annotated / benchmark true pressure waveform image; comparing a patient pressure / volume spirometry image with an annotated / benchmark true pressure / volume spirometry image; etc.) and determining the degree of bias / difference of one or more features. One or more comparison models may use color segmentation to extract only useful information, such as plotted data (e.g., curves, lines, loops) and associated axes, from various ventilation parameter images (e.g., patient ventilation parameter images and annotated / benchmark true ventilation parameter images), which may include converting the ventilation parameter images into corresponding binary masks (explained in more detail below). Then, edge segmentation, adaptive thresholding, standard deviation determination, bit manipulation, edge detection, contour detection, density-based clustering, and / or combinations thereof can be applied to a binary mask via one or more comparison models to detect one or more features of the patient ventilation parameter image and quantify the features, such as the area enclosed by curves / loops, the number of intersections on the x-axis and / or y-axis, the angles and / or curvatures of various lines, etc. Each feature can be compared with the corresponding features in the annotated image to determine the differences / biases in one or more features and quantify the degree of bias. Thresholding can be applied to ignore minor biases, and the specific threshold applied can depend on the feature. The following is about... Figures 9 to 21 Further details are presented regarding one or more comparative models and one or more features analyzed via these models. It should be understood that the ventilation parameter images processed by the comparative models may be digital versions of ventilation parameter images, including one or more numerical values ​​for each pixel (e.g., color values, brightness values, etc.), rather than versions of the ventilation parameter images actually displayed and illustrated herein (e.g., where pixel values ​​have been converted to color / brightness). In some examples, in addition to analyzing waveform and vital capacity measurement images, method 500 may also include comparisons between other ventilation system parameters of the patient (e.g., Ppeak, Plat, PEEP, FiO2, etc. as instantaneous or average values) and corresponding ventilation system parameters associated with selected annotated images.

[0060] At 508, method 500 includes determining whether any bias is detected. A bias can be a difference between at least one feature / parameter of a patient (e.g., one or more features of patient waveforms and spirometry images) analyzed by one or more comparison models and at least one feature / parameter in an annotated image. If a detected difference satisfies a condition relative to a threshold, the detected difference may be identified as a bias only. For example, the area enclosed by a loop of a given patient spirometry image (e.g., pressure / volume spirometry image) can be calculated via at least one comparison model of one or more comparison models and compared with the area enclosed by a loop of a corresponding annotated spirometry image. If the difference between the area of ​​the patient spirometry image and the area of ​​the annotated spirometry image exceeds a threshold (e.g., at least 10%), the area difference can be identified as a bias. Conversely, if the difference between the area of ​​the patient spirometry image and the area of ​​the annotated spirometry image is less than a threshold (e.g., less than 10%), the area difference may not be identified as a bias.

[0061] If a deviation is detected, method 500 proceeds to 510 to output a notification indicating the deviation, and, if known, outputs the possible cause of the deviation. The notification may be displayed on a display of the ventilation system (e.g., monitor 38) and / or on one or more care provider devices (e.g., displayed on care provider device 134 via a graphical user interface 135). For example, if the closed area of ​​a patient flow / volume spirometry image is significantly smaller than the area of ​​a selected annotated flow / volume spirometry image, and the peak pressure of a patient pressure / volume spirometry image is significantly higher than that of a selected annotated pressure / volume spirometry image, the possible cause of the reduced closed area might be a kinked endotracheal tube. The notification may include an indication that a deviation has been detected, a quantification of the deviation (percentage of each feature or actual deviation), and / or the possible cause.

[0062] When a deviation is detected and a notification is output, the clinician monitoring the patient can choose to take action to resolve the deviation. In an example where the endotracheal tube may be kinked, the clinician can examine the endotracheal tube and, if kinked, repair it. In this way, in response to the output notification, the clinician can adjust one or more ventilation system settings. Therefore, method 500 can loop back to 502 to continue acquiring ventilation settings and patient information (since ventilation settings may have changed), generating subsequent waveform and vital capacity measurement images (at a specified rate, such as every two minutes), and comparing the subsequent waveform and vital capacity measurement images with selected annotated images via one or more comparison models to determine whether the deviation persists or has been resolved. If the deviation persists, a notification can continue to be output (and in some examples, the notification can be updated if the deviation changes due to a change in ventilation settings).

[0063] If no deviation is detected at 508 (whether the deviation was not detected in the initial waveform and spirometry images, or the initial deviation has been corrected and no longer exists), then method 500 proceeds to 512 to store the current patient waveform and spirometry images as the patient's baseline images (e.g., in baseline image 114), which will be used as the basis for comparisons for any further waveform and spirometry image analysis (see below for details). Figure 6 (Explanation). It should be understood that in some examples, when a notification indicating a deviation is output, the clinician can choose to ignore or dismiss the notification. Therefore, if user input is received after a deviation notification is output (e.g., at 510) that dismisses the notification or otherwise indicates that the clinician does not intend to adjust any ventilation settings based on the notification, the current patient waveform and spirometry images can be saved as baseline images, even if they deviate from the selected annotated images. Because the selected annotated images are from other patients or based on simulated ventilation (with an artificial lung), the selected annotated images may not always reflect the expected or ideal situation for an actual patient. Therefore, the user input dismissing the notification can serve as an indication that the current patient waveform and spirometry images should be used as a future baseline. Similarly, if a notification is output at 510, but the deviation is not resolved after a threshold time amount (e.g., five minutes, ten minutes), the current patient waveform and spirometry images can be saved as baseline images. Method 500 then ends.

[0064] Figure 6 This is a flowchart illustrating a method 600 for performing continuous ventilation parameter analysis on a patient undergoing mechanical ventilation. Method 600 can be performed using a ventilation monitoring system (such as...) Figure 3 This is achieved through a ventilation monitoring system 100, which includes a ventilation system (such as...). Figure 1 and Figure 2 (Medical system 10). Method 600 may be implemented according to instructions stored in non-transitory memory and executed by one or more processors (such as the memory and processor of medical system 10). In some examples, method 600 may be implemented as part of method 400, such as at 404 of method 400.

[0065] At 602, method 600 includes obtaining new patient ventilation parameter images (e.g., one or more waveform and / or vital capacity measurement images) at specified intervals (such as once per minute or once every two minutes). This can be as described above regarding... Figure 4 and Figure 5 New patient waveform and spirometry images are obtained / generated as explained. At 604, the new patient waveform and spirometry images are compared with stored patient baseline images via one or more comparison models. This can be done as described above regarding... Figure 5 The explanation and the following about Figures 9 to 21 A more detailed explanation is needed to perform the comparison.

[0066] At 606, method 600 determines whether any deviation is detected between the new patient waveform and spirometry images and the baseline image. As previously explained, one or more comparison models can be deployed to process the new patient waveform and spirometry images and the baseline image to segment / isolate and quantify various features of the new patient waveform and spirometry images and the baseline image. Any changes in the analyzed features in the new patient waveform and spirometry images relative to the baseline image can be identified and compared with corresponding thresholds to determine whether the changes are significant enough to be classified as deviations.

[0067] If a deviation is detected, method 600 proceeds to 608 to output a notification indicating the deviation and its possible causes (if known). The notification may be output on one or more displays, such as the monitor 38 and / or one or more care provider devices. At 610, new patient waveform and spirometry images continue to be acquired at specified intervals, and at 612, the new patient waveform and spirometry images are again compared to the patient's baseline images via a comparison model.

[0068] At 614, method 600 determines whether a change in ventilation settings has been detected. A change in ventilation settings can be detected based on new patient waveform and spirometry images relative to a baseline image, and / or it can be detected based on user input received at the ventilation system indicating an actual change in settings (e.g., a change in airflow parameters, anesthetic concentration, FiO2 setting, etc.). For example, when a deviation is detected but subsequent patient waveform and spirometry images return to being the same as or similar to the baseline images, it can be inferred that the ventilation system settings have changed.

[0069] If no change in ventilation settings is detected, method 600 returns to 608 to continue outputting a notification, obtaining new patient waveform and vital capacity measurement images, and comparing the new patient waveform and vital capacity measurement images with the baseline image. This process can be repeated until a change in ventilation settings is detected, or until the clinician removes the notification, or until a threshold amount of time has elapsed since the initial output of the notification. If a change in ventilation settings is detected at 614, method 600 proceeds to 616 to move the current baseline image to a remote or local data device (e.g., remote data device 160 or local data device 150), and at 618 saves the new patient waveform and vital capacity measurement images as the baseline image (e.g., in baseline image 114). When ventilation system settings are changed, it can be assumed that the waveform and vital capacity measurement images generated after the change in ventilation system settings reflect the new ideal / default parameters for the patient, and the baseline image is updated accordingly. However, instead of discarding the previous baseline image, it can be saved in a new storage location where the baseline image can be used to further refine one or more comparison models. Method 600 then returns to 602 to continue monitoring subsequent patient waveforms and vital capacity measurement images.

[0070] Returning to 608, if no deviation is detected between the new patient waveform and vital capacity measurement images and the baseline images, method 600 proceeds to 620 to compare the new patient waveform and vital capacity measurement images with selected annotated images via one or more comparison models. The selected annotated images may be similar to those described above regarding... Figure 5The selection of annotated images is based on patient information and current ventilation system settings, chosen from stored annotated images (e.g., from annotated image 116). The selected annotated image may be used with one or more comparison models to detect outlier conditions (rather than deviations from the patient's baseline). Outlier conditions may include respiratory arrest and / or spontaneous breathing. Therefore, at 622, method 600 may include determining whether respiratory arrest has been identified. For example, respiratory arrest may be identified based on the presence of a flat line in a waveform image. If respiratory arrest has been identified, method 600 proceeds to 624 to output a notification indicating that the patient has ceased breathing, at which point method 600 may proceed to 610 to continue acquiring patient waveform and vital capacity measurement images and comparing them with baseline images to determine whether a change in ventilation settings has been detected.

[0071] If respiratory cessation is not detected, method 600 proceeds to 626 to determine whether spontaneous breathing has been detected based on new patient waveform and vital capacity measurement images relative to the selected annotated images. For example, spontaneous breathing may be detected based on, for example, a small negative deflection (e.g., a pressure drop) in a pressure waveform image. The following is about... Figure 21 Additional information regarding the detection of spontaneous breathing is provided. If spontaneous breathing is not detected, method 600 returns to 602 to continue monitoring subsequent patient waveform and spirometry images. If spontaneous breathing is detected, method 600 proceeds to 628 to output a notification indicating that the patient is breathing spontaneously (e.g., output to a display such as monitor 38 and / or one or more care provider devices). At 630, method 600 determines whether a command to initiate a weaning protocol has been received. For example, a notification that the patient is breathing may include a request to initiate a weaning protocol. Upon seeing the notification that the patient is breathing spontaneously, the clinician monitoring the patient may enter user input to command the ventilation system to initiate a weaning protocol (e.g., selecting a "Yes" button or other user interface element). If no command to initiate a weaning protocol is received (e.g., the clinician dismisses the notification or selects a "No" button), method 600 proceeds to 616. If a command to initiate a weaning protocol is received, method 600 proceeds to 632 to initiate the weaning protocol, which will be explained in more detail below. In other examples, the ventilation system may automatically initiate a weaning protocol upon detecting that the patient is breathing spontaneously. Then, method 600 ends.

[0072] Therefore, at the start of mechanical ventilation, a set of ventilation parameter images (e.g., pressure, flow, and volume waveform images, and pressure / volume and flow / volume vital capacity measurement images) can be captured for the patient and compared using at least one comparison model with corresponding annotated / benchmark images selected based on the patient and ventilation parameters (e.g., patient age, height, weight, and sex, and ventilation system settings). This process can be used to “finalize” the patient’s baseline images reflecting the current target ventilation waveform and the vital capacity measurement loop required by the patient. For example, if the initial ventilation parameter images deviate from the annotated / benchmark images, the ventilation system settings can be adjusted. Once the baseline images are captured, they can be compared with subsequent patient ventilation parameter images using at least one comparison model. Any deviation from the baseline images can trigger a notification output, allowing clinicians to adjust the ventilation system settings as instructed. During and / or after the procedure, clinicians can annotate any baseline or other ventilation parameter images and store them in a database for future reference (e.g., along with other annotated / benchmark images). As will be explained in more detail below, one or more comparison models can characterize / quantify various features of ventilation parameter images (both patient and annotated / baseline real images), such as the area enclosed by curves, the number of x and / or y intersections, line angles, etc. The ventilation parameter images, along with the quantified features, can be stored locally (e.g., on the ventilation system) and transmitted to local and / or remote data storage devices. In some examples, once an image has been characterized by one or more comparison models, all quantified features can be saved and used for any future comparisons. For example, a given baseline image (e.g., a pressure / volume vital capacity measurement loop) can be characterized by one or more comparison models to determine the area enclosed by curves / loops in the image, as well as other features. Whenever a given baseline image is compared to a subsequent patient ventilation image (e.g., a subsequent pressure / volume vital capacity measurement image), the quantified features can be retrieved from memory for comparison, rather than repeatedly processing the baseline image through one or more comparison models, thus reducing processing requirements.

[0073] Furthermore, the one or more comparative models disclosed herein allow for the characterization and quantification of various features of ventilation parameter images, and can compare the characterization / quantification features of a reference ventilation parameter image (e.g., a patient baseline image or an annotated / benchmark real image) with the characterization / quantification features of the current patient ventilation parameter image. Several benefits can be achieved by evaluating and comparing ventilation parameter images (e.g., rather than just evaluating sensor data). For example, some characterization features (such as the area enclosed by a curve / loop in a spirometry image) are not identifiable in the sensor data used to generate the ventilation parameter images. Additionally, the ventilation system (e.g., a ventilator and / or anesthesia machine) may not be configured to store raw sensor data, but can be configured to store some ventilation parameter images (e.g., the last five images of each of the various waveforms and spirometry images disclosed herein). Therefore, by analyzing the ventilation parameter images, changes in the patient's condition can be detected using resources already available on the ventilation system. Furthermore, the one or more comparative models disclosed herein may have lower resource requirements (e.g., requiring less memory and / or processing power) than other types of models (e.g., artificial intelligence models). Overall, the method for evaluating ventilation parameter images disclosed in this paper provides a mechanism for comparing current patient ventilation parameters with expected ventilation parameters, which utilizes existing resources available on the ventilation system in a memory and processor-efficient manner.

[0074] Figure 7 This is a flowchart illustrating method 700 for implementing a weaning protocol for a mechanically ventilated patient. Method 700 can be performed using a ventilation monitoring system (such as...) Figure 3 This is achieved through a ventilation monitoring system 100, which includes a ventilation system (such as...). Figure 1 and Figure 2 (Medical system 10). Method 700 may be implemented according to instructions stored in non-transitory memory and executed by one or more processors (such as the memory and processor of medical system 10). In some examples, method 700 may be implemented as part of method 400 and / or method 600, such as at 410 of method 400 and / or at 632 of method 600.

[0075] At 702, method 700 includes outputting an audio command via a loudspeaker (such as a loudspeaker (e.g., loudspeaker 126)). The audio command may include instructions relating to the patient, such as instructing the patient to open their eyes, close their eyes, and / or blink in a specific pattern (e.g., blink twice rapidly). During and after the output of the audio command, a patient video is recorded via a camera (e.g., camera 124), as indicated at 704. The patient video may include the patient in the field of view and may be color or monochrome video. At 706, weaning parameters are obtained, and weaning rules are applied to the weaning parameters to perform a rule-based check of the patient's condition. Weaning parameters may include the patient's SpO2, blood pressure, PR, tidal volume, FiO2, EtCO2, and / or other parameters measured by the ventilation system and / or patient monitors. Weaning rules may include threshold ranges for each weaning parameter that indicate or suggest that the patient is not anesthetized and is able to breathe without assistance.

[0076] At 708, method 700 determines whether the patient is able to comply with an audio command and whether the offline parameters pass a rule-based check. One or more comparison models may include image models trained to identify whether a patient is able to comply with an audio command. For example, the image model may be a neural network trained to identify whether the patient's eyes are open or closed in each image frame of a patient video, and the method may determine whether the patient is able to comply with an audio command based on whether (and when) the patient's eyes are open and closed in the patient video relative to the time the audio command is issued. If each offline parameter is within a specified threshold range for that parameter, the offline parameter passes a rule-based check. For example, a rule-based check may include determining whether the patient is making a spontaneous breathing effort exceeding a threshold number of continuous breaths (e.g., five breaths), whether SpO2 exceeds a threshold (e.g., 96%), and whether EtCO2 is within a threshold range (e.g., 35 mm Hg to 45 mm Hg). If all rules / parameters are satisfied, the offline parameters pass a rule-based check.

[0077] If the patient is able to comply with the audio commands and the weaning parameters do indeed pass the rule-based check, method 700 proceeds to 710 to output a notification (e.g., for display on monitor 38 and / or one or more care provider devices) to monitor the patient's weaning response. The notification may alert the clinician monitoring the patient that the patient is ready to be weaned, and the clinician can initiate the weaning and extubation process, at which point method 700 may terminate. If the patient is unable to comply with the audio commands, or if the weaning parameters do not pass the rule-based check, method 700 proceeds to 712 to move frames from the patient video to a remote or local storage device to further refine the image model. Method 700 then proceeds to 602 to continue monitoring the patient's waveform and vital capacity images, as the patient does not show any signs of being ready to be weaned.

[0078] Figure 8 An exemplary graphical user interface 800 is shown that can be displayed on a monitor of the ventilation system, such as monitor 38, during the execution of the methods described above. The graphical user interface 800 can display pressure waveform images 802, flow waveform images 804, volume waveform images 806, pressure / volume vital capacity measurement images 808, and flow / volume vital capacity measurement images 810. The graphical user interface 800 can also display other anesthetic parameters 812 (such as inhaled and exhaled anesthetic concentrations) and other ventilation parameters 814 (e.g., PEEP, tidal volume). The graphical user interface 800 can also display notifications generated and output during the execution of the methods described above. For example, a notification 816 can be displayed indicating that a deviation has been detected in the flow / volume vital capacity measurement image, possibly due to a kinked endotracheal tube.

[0079] Figure 9 The first set of vital capacity measurement images is shown, which can be assessed to determine changes in a patient's ventilation status. The first set of vital capacity measurement images can be performed in accordance with the above description... Figures 4 to 7 One or more of the methods described herein are generated using a medical system 10 and via one or more comparison models (e.g., Figure 3 The analysis is performed using a comparative model (based on a model library). The first set of spirometry images includes a first spirometry image 910 and a second spirometry image 920. The first spirometry image 910 is a pressure / volume spirometry image, depicting the patient's airway volume as a function of the patient's airway pressure during a respiratory cycle. Because a respiratory cycle includes inhalation and exhalation, the first spirometry image 910 can include a first loop 912 of patient pressure and volume. The characteristics of the first loop 912 can be evaluated by one or more comparative models to establish a baseline for the patient, since the first spirometry image 910 can be an initial image capturing the initial state of the subject.

[0080] The second spirometry image 920 is also a pressure / volume spirometry image and includes a second loop 922 capturing patient pressure and volume following the first spirometry image 910. Features of the second loop 922 can be evaluated by one or more comparative models and compared with features of the first loop 912 to determine whether the patient's condition has changed relative to an initial / baseline condition. For illustration, the first loop 912 is also included in the second spirometry image 920 to illustrate changes in patient condition.

[0081] To evaluate the characteristics of the first loop 912 and the second loop 922, loops can be extracted from the spirometry image using a first comparison model. The first comparison model can be configured to perform color segmentation by grouping pixels in the spirometry image that share common characteristics such as color. To perform color segmentation, the first comparison model can convert the image (e.g., the second spirometry image 920) to a suitable color space, such as hue, saturation, and value (HSV), and convert the image (in the selected color space) into a binary mask, where pixels in, for example, a selected color range are set to a value of 1, and all other pixels are set to a value of 0. Thus, color segmentation can extract loops or waveforms and axes. After extraction, binary closure (e.g., dilation followed by erosion) can be used to fill any spaces (e.g., areas where the extraction of loops or axes is uneven). The finally extracted image can be used for downstream analysis via an additional comparison model, such as determining areas, intersections, etc., as explained in more detail below. The first comparison model can have adjustable parameters, including HSV selection and a binary mask (e.g., a color range used to set the value of each pixel in the binary mask, which may depend on the colors present in the ventilation parameter image). By converting the ventilation parameter image into a corresponding binary mask and performing downstream computer vision / image processing techniques on the binary mask via one or more comparison models, memory requirements can be reduced by storing the binary mask instead of the complete image, and the processing requirements of the comparison model when performing computer vision / image processing techniques can be reduced since the complete image does not have to be processed.

[0082] Figure 10 A second set of vital capacity images is shown, which can be assessed to determine changes in a patient's ventilation status. This second set of vital capacity images can be used in the procedures outlined above. Figures 4 to 7 One or more of the methods described herein are generated using a medical system 10 and via one or more comparison models (e.g., Figure 3 The comparison model in the model storage library was analyzed. The second set of vital capacity measurement images includes the first vital capacity measurement image 1010 and the second vital capacity measurement image 1020. Similar to Figure 9The second vital capacity measurement image 920, and each of the first vital capacity measurement images 1010 and 1020, includes two loops: a previous (e.g., baseline) loop and a current loop. The first vital capacity measurement image 1010 is a pressure / volume vital capacity measurement image, and the second vital capacity measurement image 1020 is a volume / flow vital capacity measurement image. Each of the first and second vital capacity measurement images 1010 and 1020 illustrates features indicating a kinked endotracheal tube. After color segmentation via a first comparison model, a second comparison model can determine the x-maximum point of each loop in the first vital capacity measurement image 1010, i.e., the highest airway pressure in the respiratory cycle. It can be seen from the first vital capacity measurement image 1010 that the x-maximum point of the current loop (shown as a solid line) is significantly higher than the x-maximum point of the previous loop (shown as a dashed line). The increase in maximum airway pressure indicates a kinked endotracheal tube, as the kink creates resistance to exhaled air. Similarly, after color segmentation via the first comparison model, the third comparison model can determine the area enclosed by each loop in the second vital capacity measurement image 1020. As shown in the second vital capacity measurement image 1020, the area of ​​the current loop (solid line) is significantly smaller than the area of ​​the previous loop (dashed line). The increased resistance of the kinked endotracheal tube leads to a reduction in the amount of air inhaled and exhaled.

[0083] The third comparison model can determine the area enclosed by the loop by employing color segmentation (which can also be determined by the first comparison model), edge segmentation, contour detection, and rendering. Adjustable parameters of the third comparison model include HSV selection and a binary mask (for color segmentation), thresholds and modes in edge detection, and modes and methods for contour detection. The third comparison model can use the binary mask as a cleanup tool (e.g., after edge segmentation, contour detection, and rendering, a binary mask can be applied to remove any features outside the resulting curve / loop). After color segmentation, the third comparison model can use edge segmentation (e.g., in the segmented image) to find the contour of the loop, ultimately creating a masked and highlighted image to calculate only the area enclosed by the curve / loop. In some examples, the loop can be segmented into, for example, a first portion above the x-axis and / or a second portion below the x-axis, and the area of ​​one or both portions can be calculated. This may include detecting the contour, then determining the area of ​​the portion above the x-axis, considering only the points forming the contour above the x-axis, and ultimately creating a mask to highlight only the area of ​​the curve above the x-axis. In some examples, the ratio of the upper portion (e.g., above the x-axis) to the total area can be determined. The total area can be calculated before only certain points are masked.

[0084] Figure 11 The third set of vital capacity images is shown, which can be assessed to determine changes in a patient's ventilation status. The third set of vital capacity images can be used in the procedures outlined above. Figures 4 to 7 One or more of the methods described herein are generated using a medical system 10 and via one or more comparison models (e.g., Figure 3 The comparison model in the model storage library was analyzed. The third group of vital capacity measurement images includes the first vital capacity measurement image 1110 and the second vital capacity measurement image 1120. Similar to Figure 9 The second vital capacity measurement image 920, each of the first vital capacity measurement image 1110 and the second vital capacity measurement image 1120 includes two loops: a previous (e.g., baseline) loop (shown as a dashed line) and a current loop (shown as a solid line). Both the first vital capacity measurement image 1110 and the second vital capacity measurement image 1120 are pressure / volume vital capacity measurement images. Each of the first vital capacity measurement image 1110 and the second vital capacity measurement image 1120 shows features indicating that the patient is breathing on a ventilator (e.g., lacking a relaxed state).

[0085] After color segmentation via the first comparison model and determination of the x-maximum point via the second comparison model, the fourth comparison model can determine the angle between the line connecting the x-maximum point and the lower right point of each loop. The first vital capacity measurement image 1110 shows early signs of the patient breathing on a ventilator for the current loop, while the second vital capacity measurement image 1120 shows late signs of the patient breathing on a ventilator for the current loop.

[0086] The fourth comparison model can employ color segmentation, edge segmentation, and contour detection, where the parameters of HSV selection and binary masking (e.g., for color segmentation), thresholding and mode in edge detection, and mode and method for contour detection are adjustable. The fourth comparison model can utilize binary masking as a cleanup tool. The fourth comparison model identifies the highest point on the x-axis and draws a line to a fixed point in the lower right corner. The deviation of this line from the positive x-axis direction is calculated. Similarly, the threshold can be set to a minimum to find the deviation from the baseline (e.g., the previous) image.

[0087] For example, as described above, lines have been placed on the first vital capacity measurement image and the second pacing image to show that the angle of the line for each current loop is significantly different from the angle of the corresponding line for each previous loop (e.g., showing that at the x-maximum / maximum airway pressure, the airway pressure of the current loop is higher than that of the previous loop, and the airway volume of the current loop at the maximum airway pressure is lower than that of the previous loop). Patient ventilation can be assessed by examining the angle of the lines. If the angle of the line is not within a threshold, a deviation is detected. If the angle is within a threshold, the x-maximum and y-maximum are determined. If the x-maximum and y-maximum are not within a threshold, a deviation is detected. Furthermore, starting from the origin, a deviation is detected if the line placed as described above intersects the curve at more than two locations, and the left side of the line covers a small area. Figure 11 In the example shown, the inward curve (marked with early signs and late signs) causes the line to intersect the curve at more than two locations.

[0088] Figure 12 The fourth set of vital capacity images is shown, which can be assessed to determine changes in a patient's ventilation status. The fourth set of vital capacity images can be used in the procedures outlined above. Figures 4 to 7 One or more of the methods described herein are generated using a medical system 10 and via one or more comparison models (e.g., Figure 3 The analysis was performed using a comparative model from the model storage library. The fourth group of vital capacity measurement images includes the first vital capacity measurement image 1210 and the second vital capacity measurement image 1220. Similar to... Figure 9 The second vital capacity measurement image 920, each of the first vital capacity measurement image 1210 and the second vital capacity measurement image 1220 includes two loops: a previous (e.g., baseline) loop (dashed line) and a current loop (solid line). The first vital capacity measurement image 1210 is a pressure / volume vital capacity measurement image, and the second vital capacity measurement image 1220 is a volume / flow vital capacity measurement image. The first vital capacity measurement image 1210 shows features indicating accidental single-lung intubation, and the second vital capacity measurement image 1220 shows features indicating leaks in the ventilation system, particularly leaks in the cuff or laryngeal mask.

[0089] Unexpected single-lung intubation can be detected based on a significant increase in pressure. This increase can be detected by identifying the point of maximum x-axis value and drawing a line from this point to a fixed point in the lower right corner (as described above). The deviation of the angle formed by this line and the positive x-direction is calculated. The basic principles of color segmentation can be used, with a threshold set to a minimum, to identify the deviation from previous images. The maximum x-axis value and the maximum y-axis value can also be compared.

[0090] Leaks in the ventilation system can be detected based on the output of a fifth and / or a sixth comparison model, each of which can be configured to generate an output indicating whether a vital capacity measurement loop (e.g., a volume / flow loop) is open or closed. After color segmentation via a first comparison model, the fifth comparison model can determine the intersections of loops in vital capacity measurement images such as a second vital capacity measurement image 1220. The fifth comparison model can be configured to perform color segmentation (or alternatively, color segmentation can be performed by the first comparison model), edge segmentation, contour detection, and DBSCAN. Adjustable parameters of the fifth comparison model include HSV selection and a binary mask (e.g., for color segmentation), thresholds and modes in edge detection, modes and methods for contour detection, minimum number of points in EPS and clustering, and a reference height for the x-axis. The fifth comparison model can use a binary mask as a cleanup tool.

[0091] Therefore, the fifth comparison model determines the points of intersection (POIs) of the loop with the x-axis after extracting the chromatic portion of the image. The POIs in the previous loop of the second pulmonary capacity measurement image 1220 may be located differently from the POIs in the current loop and / or be more or fewer in number, which can indicate leakage due to the current loop being an open loop. Alternatively, after color segmentation, the sixth comparison model can approximate the perimeter of the loop as a nearby polygon. If the polygon is open, leakage can be indicated. For example, a line can be drawn starting from the top of the loop, following the loop's contour using curve convexity. The start and end points of the line can be evaluated to determine whether the polygon is open or closed based on the distance between the start and end points.

[0092] Figure 13 Another vital capacity measurement image is shown, specifically vital capacity measurement image 1310, which can be evaluated to determine changes in the patient's ventilation status. Vital capacity measurement image 1310 can be used in the procedures described above regarding... Figures 4 to 7 One or more of the methods described herein are generated using a medical system 10 and via one or more comparison models (e.g., Figure 3 The analysis is performed using a comparison model from a model storage library. The spirometry image 1310 includes two loops: a previous (e.g., baseline) loop and a current loop. The spirometry image 1310 is a pressure / volume spirometry image. The spirometry image 1310 shows characteristics indicative of spontaneous breathing (e.g., negative airway pressure during a portion of the respiratory cycle), specifically, the previous / baseline loop is entirely on the positive side of the graph, while the current loop is included in a portion of the negative side of the graph.

[0093] Figure 14The fifth set of vital capacity images is shown, which can be assessed to determine changes in a patient's ventilation status. The fifth set of vital capacity images can be used in the procedures outlined above. Figures 4 to 7 One or more of the methods described herein are generated using a medical system 10 and via one or more comparison models (e.g., Figure 3 The analysis is performed using a comparative model from a model library. The fifth set of spirometry images includes a first spirometry image 1410 and a second spirometry image 1420. The first spirometry image 1410 may include a previous (e.g., baseline) loop, and the second spirometry image 1420 may include the current loop. Both the first spirometry image 1410 and the second spirometry image 1420 are pressure / volume spirometry images. The first spirometry image 1410 and the second spirometry image 1420 show features that can be evaluated by a seventh comparative model, which is configured to identify overexpansion or “wrinkling” (e.g., beak formation) of the spirometry loop (as indicated by marker points on the loop of the second spirometry image 1420).

[0094] The seventh comparison model can employ thresholding and contour detection, and can have adjustable parameters, including thresholds and patterns, contour retrieval patterns and methods, and the length and width of the rectangle. The seventh comparison model applies a curvature threshold of the loop as a portion of the rectangle, with a minimum length of 60 pixels and a minimum height of 15 pixels. The reference point is considered as the point with x_max. The size of the rectangle can be determined by the baseline / reference image. For example, the rectangle can be drawn based on the x-maximum and y-maximum points of the baseline / reference image, as shown in rectangle 1411. A rectangle 1421 of the same size can be placed on the second spirometry image 1420. The number of points outside the rectangle can be counted. A higher number of points indicates more pronounced hysteresis prolongation, while hysteresis prolongation in normal loops is less severe. The number of points in the region can be used to identify dilation. Hysteresis dilation indicates excessive pressure on the lungs, with lung dilation exceeding an acceptable threshold, which can cause barotrauma.

[0095] Figure 15 Two exemplary spirometry loops are shown, in which an exemplary volume / flow loop has been evaluated by an eighth comparison model to determine the number of intersections on the y-axis. The eighth comparison model may employ thresholding, DBSCAN, and contour detection. Adjustable parameters of the eighth comparison model may include thresholds and patterns, contour retrieval patterns and methods, reference values ​​for the x and y axes (e.g., the position of the y-axis), and acceptable point values, EPS, and the minimum number of points in the cluster.

[0096] The eighth comparison model is used to find the intersection with the y-axis, with the parameter to be adjusted being EPS in DBSCAN. The eighth comparison model selects an approximate distance to the y-axis calculated by finding the point at the minimum value (e.g., x_min) on the x-axis. For volume / flow loops, more than one POI on the y-axis, along with changes in the slope of the expiratory curve, indicates air accumulation due to potential airway obstruction.

[0097] For example, Figure 15 The first vital capacity measurement loop 1510 includes one intersection point on the y-axis, while the second vital capacity measurement loop 1520 includes two intersection points with the y-axis and an increased curvature of the expiratory curve, which is detected using the ninth comparative model, as explained below.

[0098] Figure 16 This shows the results after evaluation by the ninth comparative model. Figure 15 In the lung capacity measurement loop, the ninth comparison model is configured to determine the number of points along the line connecting the lowest point of the loop and the y-axis. The ninth comparison model can employ thresholding, DBSCAN, and contour detection, where adjustable parameters include the threshold and pattern, contour retrieval pattern and method, a reference value for the y-axis (position on the y-axis) and acceptable point values, EPS, and the minimum number of points in the cluster. The ninth comparison model compares the lowest point on the curve on the y-axis with the lowest intersection point of the curve on the y-axis. The number of points between lines defines the radius of curvature of the line / curve. The larger the number of points on the curve, the straighter the line.

[0099] For example, Figure 16 The lowest point of the first spirometry loop 1510 and the intersection point with the y-axis are marked on the second spirometry loop 1520. Points where each loop intersects the line drawn between the lowest point and the y-intercept are represented by gray dots on the loop. The first spirometry loop 1510 has more marked points than the second spirometry image, indicating that the portion of the curve between the lowest point (y_min) and the y-intercept of the first spirometry loop is straighter than the equivalent portion of the second spirometry loop.

[0100] Figure 17 The first set of waveform images 1710 is shown, which can be evaluated to determine changes in the patient’s ventilation status. Figure 17 Intermediate images generated by the tenth comparison model during the evaluation of the first set of waveform images 1710 are also shown. The first set of waveform images 1710 can be used to perform the above-mentioned... Figures 4 to 7 One or more of the methods described herein are generated using a medical system 10 and via one or more comparison models (e.g., Figure 3 The model storage library was compared with the model for analysis.

[0101] The first set of waveform images 1710 includes a normal / reference true image 1712 (which in some examples may be a baseline image) and an image showing the deviation (referred to as the deviation image 1714). A tenth comparison model can identify the common region between the curves of the normal image 1712 and the deviation image 1714. The tenth comparison model can employ adaptive thresholding and bit manipulation of the image pixels, where tunable parameters include a threshold and a mode, and cropping image values. The tenth comparison model identifies the overlapping region between the two curves (in this case, the curve of the normal image 1712 and the curve of the deviation image 1714). Identifying the overlapping region involves performing an OR operation on the two images formed by the normal image 1712 and the deviation image 1714 (in this case, the first image 1720 and the second image 1730, respectively) to form the final image 1740, and finding the region by calculating the number of black pixels in the final image 1740. To generate the two images (e.g., the first image 1720 and the second image 1730), Matplotlib can be applied to the normal image 1712 and the deviation image 1714, and the output can then be converted into an image. In addition, bitwise_or is used to generate the final image 1740.

[0102] Figure 18 This shows what can be applied to ventilation parameter images (such as...) Figure 17 The output of the eleventh comparison model (comprising the normal image 1712 and the deviated image 1714) is used. The eleventh comparison model can determine the point-to-point deviation and standard deviation between the two curves. The eleventh comparison model can employ adaptive thresholding, contour detection, and standard deviation, where the parameters of the threshold and mode, cropped image values, and contour retrieval mode and method are adjustable.

[0103] The eleventh comparison model uses point-to-point standard deviation and overall standard deviation, comparing two curves, such as the first curve 1810 and the second curve 1820, point-by-point and then calculating the pixel distance. Zero-padding the array ensures that the two curves are of equal length. Deviation can be detected by setting a tolerable standard deviation value, and any value exceeding this value can be considered a deviation. For example, the point-to-point deviation of the second curve 1820 relative to the first curve 1810 can be determined, and the overall standard deviation of the deviation between the two curves can be determined.

[0104] Figure 19 A second set of waveform images 1910 is shown that can be evaluated to determine changes in the patient’s ventilation status. Figure 19 The output 1920 of the twelfth comparative model is also shown during the evaluation of the second set of waveform images 1910. The second set of waveform images 1910 can be performed in accordance with the above description. Figures 4 to 7One or more of the methods described herein are generated using a medical system 10 and via one or more comparison models (e.g., Figure 3 The analysis is performed using a comparison model from the model storage library. In the example shown, each waveform in the set of waveform images 1910 is a flow waveform.

[0105] The second set of waveform images 1910 includes a normal / reference true image 1912 (which in some examples may be a baseline image) and an image showing the deviation (referred to as the deviation image 1914). A twelfth comparison model can identify the number of intersections with the x-axis. The twelfth comparison model can employ edge detection, contour detection, and density-based clustering (DBSCAN), where adjustable parameters include a threshold and mode, a reference height for the x-axis in the image, DBSCAN parameters (EPS), a minimum number of elements in the cluster, and the distance of the points to be considered from the reference x-axis. The twelfth comparison model can identify and count the number of intersections of the waveforms on the x-axis (e.g., as shown in output 1920, where intersections are indicated by points placed on the x-axis). If the number of intersections is even, the waveform can be identified as having a deviation (specifically, gaps between loops, as indicated by the arrows in the waveform of the deviation image 1914); however, if the number of intersections is odd, the waveform can be identified as normal (as shown in the waveform of the normal / reference true image 1912). An increase in the number of intersections indicates that there may be gaps in the patient's respiratory cycle, as there is no airflow during inspiration. Ideally, each cycle should have 3 points of interest (POIs) based on input.

[0106] Figure 20 A third set of waveform images 2000 is shown that can be evaluated to determine changes in the patient’s ventilation status. Figure 20 The first output 2030 and the second output 2040 of the twelfth comparison model are also shown during the evaluation of the third set of waveform images 2000. The third set of waveform images 2000 can be used to perform the above-mentioned... Figures 4 to 7 One or more of the methods described herein are generated using a medical system 10 and via one or more comparison models (e.g., Figure 3 The analysis is performed using a comparison model from the model storage library. In the example shown, each waveform in the third group of waveform images 2000 is a flow waveform showing two consecutive respiratory cycles.

[0107] The third set of waveform images 2000 illustrates an example of patient-ventilator asynchrony, which can be seen on the flow-time waveform as a deviation from the baseline expiratory flow rate, where the decrease in airway pressure is small or not changed at all, and no breaths are delivered. For example, the first waveform image 2010 shows an invalid trigger where the expiratory flow rate is prolonged during the first illustrated respiratory cycle, and the subsequent inspiratory phase is skipped. Invalid triggers can be detected based on the number of intersections with the x-axis in each respiratory cycle (e.g., more than three intersections can indicate an invalid trigger), as shown by the intersections in the first output 2030, and by processing a portion of a cycle and the slope pattern of the curve. In another example, invalid triggers can be identified by determining the distance between peaks. For example, after color segmentation, contours are detected, and points located on a reference line are eventually masked. Clustering is performed to find only one point intersecting the reference line. Anomalies can be detected by finding the distance between two consecutive points. If this distance is greater than a threshold number of pixels, an invalid trigger exists.

[0108] For example, invalid triggers can be identified through a process that includes the following:

[0109] Step 1: Find the intersection point on the X-axis.

[0110] Step 2: For the first breathing cycle, find the first 3 points of interest (POIs) and the Y-max of the first two intersections.

[0111] Step 3: Check the next intersection and find its Y value, and compare it with Y-max. If it is less than Y-max, the expiratory flow rate is considered prolonged and the triggering is deemed invalid.

[0112] Step 4: Check the next intersection and find its Y value, and compare it with Y-max. If it equals Y-max (+ / - threshold), evaluate the next breathing cycle (starting from Step 2).

[0113] The second waveform image 2020 illustrates a double trigger, where no exhalation occurs between inspiratory airflows. A variation or extension of the twelfth comparison model can be used to identify the two closest points intersecting the x-axis and determine the distance between them. For example, the second output shows the two closest intersection points. The distance between these two intersection points can be calculated, and if this distance is less than a threshold, a double trigger can be identified.

[0114] Figure 21 A fourth set of waveform images 2100 is shown, which can be assessed to determine changes in the patient's ventilation status. The fourth set of waveform images 2100 can be used in the process described above regarding... Figures 4 to 7 One or more of the methods described herein are generated using a medical system 10 and via one or more comparison models (e.g., Figure 3The analysis was performed using a comparative model from the model storage library. The fourth set of waveform images 2100 includes pressure waveform image 2110, flow waveform image 2120, and volume waveform image 2130, each of which was obtained simultaneously from the same patient.

[0115] The fourth set of waveform images 2100 includes features indicating the patient's attempt at spontaneous breathing. For example, the patient initiates a mid-inspiratory breathing cycle in the first illustrated breathing cycle, as evidenced by the negative pressure deflection in pressure waveform image 2110, the decrease in flow rate (due to the patient's effort) in flow waveform image 2120, and the change in delivered volume due to the patient's inspiratory effort in volume waveform image 2130. Additionally, the patient initiates another breath, which causes the inspiratory valve to remain closed, thus no flow is provided to the subsequent breathing cycle. Therefore, negative pressure deflection is observed in pressure waveform image 2110, and missed triggering is observed in flow waveform image 2120 and volume waveform image 2130. Furthermore, the patient attempts to exhale during the subsequent inspiratory cycle, which generates positive pressure (shown on pressure waveform image 2110) and keeps the expiratory valve closed. Flow waveform image 2120 shows the expiratory flow rate under the patient's expiratory effort, and volume waveform image 2130 shows the decrease in tidal volume due to the patient's expiratory effort.

[0116] Spontaneous breathing can be detected using the comparative model disclosed herein, based on the detection of a small negative pressure before the next respiratory cycle. The process of detecting spontaneous breathing may include: acquiring a baseline image; rescaling the current image to the same size as the baseline image if necessary; comparing point-to-point deviations; and comparing overall deviations.

[0117] Therefore, the comparative model described above can be used to identify and quantify features of current patient ventilation parameter images and reference ventilation parameter images (e.g., patient baseline images or selected annotated / benchmark real images), and deviations between features in the current patient ventilation parameter image and corresponding features in the reference ventilation parameter image can trigger the output of one or more notifications to alert one or more clinicians monitoring the patient that the patient's status has changed. The comparative model can employ various thresholding, segmentation, and clustering techniques (such as color segmentation, edge segmentation, adaptive thresholding, standard deviation determination, bit manipulation, edge detection, contour detection, and density-based clustering) to identify and quantify features in the images, as explained in the examples above. While at least some features of the ventilation parameter images can be characterized / quantified using other techniques, the comparative model disclosed herein provides robust feature representations with minimal error compared to other techniques.

[0118] Comparative models can also be used to identify and quantify further features. For example, the area under the curve (AUC) for each inspiratory and expiratory phase can be determined on a volume waveform image. The inspiratory volume should equal the expiratory volume, and the difference between these volumes can indicate air leakage in the system or inherent positive end-expiratory pressure (PEEP) (e.g., automated PEEP or air retention). Additionally, flow waveform images can provide useful information about a patient's exhalation. The shape of the expiratory limb of the flow waveform is influenced by airflow resistance and lung compliance. In cases where obstructive processes lead to higher airway resistance, the flow waveform will show a decrease in PEF and a prolonged time for the expiratory curve to return to the zero flow baseline.

[0119] Pressure waveform images and pressure-volume spirometry images (e.g., pressure-volume) can provide information about airway compliance. Regardless of the mode in which Pplat is measured (e.g., pressure-targeted mode, volume-controlled mode, or flow-targeted control), a large difference between PIP and Plat, as seen in pressure waveforms, indicates greater airway resistance, as exemplified in severe bronchospasm. Furthermore, in a normally compliant lung, pressure increases at a constant rate, and volume also increases constantly. This corresponds to the pressure index and the slope of the straight line in the section of the pressure / volume spirometry circuit where volume increases during expiration. A decreasing slope is consistent with a stress index less than 1, giving it a downward-curving shape. This indicates that lung compliance and re-expansion improve with increasing lung volume. When lung compliance deteriorates with increasing volume, the slope increases, resulting in an upward-curving shape, or a stress index greater than 1. This can indicate lung overexpansion, which can be detected using the seventh comparison model and... Figure 14 As shown in the image.

[0120] By evaluating flow-volume vital capacity (CV) images, especially by assessing the expiratory limbs, several important pieces of information can be obtained about the airflow in and out of the lungs. For example, the flow-volume loop can be used to measure peak expiratory flow. A lower peak expiratory flow indicates the presence of potential obstruction, such as that seen during bronchoconstriction.

[0121] It should be understood that some of the features discussed above may be specific to spirometry images, while others may be specific to waveform images. Both waveform and spirometry images may also share other common features. Therefore, when processing a particular ventilator parameter image to characterize / quantify the features in that ventilator parameter image, some, but not all, of the comparison models described above can be invoked (e.g., the comparison model applied to a given ventilator parameter image can be selected based on the image type, such as waveform or spirometry image). Furthermore, each possible feature can be characterized for a given ventilator parameter image (e.g., as described above), or only selected features can be characterized. Therefore, a comparison between a current patient ventilator parameter image and a corresponding reference ventilator parameter image may include using at least one comparison model to characterize at least one feature of each image. Furthermore, it should be understood that the individual comparison models described above can be appropriately combined to form a larger model configured to characterize multiple features in a given ventilator parameter image. However, by providing information as described above… Figures 9 to 21 The individual comparison models, each configured to perform image processing tasks to facilitate feature representation, can achieve various benefits such as performing parallel processing on images of given ventilation parameters, updating or adding new comparison models as needed without affecting the performance of the remaining comparison models, and the ability to invoke only those comparison models deemed suitable for a given image of ventilation parameters.

[0122] The advantage of using at least one comparison model to compare each patient's ventilation parameter image among multiple patient ventilation parameter images with a corresponding reference ventilation parameter image is that deviations from the reference ventilation parameter image can be automatically detected. This allows a deviation alert to be issued to one or more clinicians to adjust the ventilation system settings (if indicated). Since patients may require ventilation for hours, days, or even weeks, another advantage of automatically detecting deviations from the reference ventilation parameter image is that deviations that might be missed if only a clinician were monitoring the patient's ventilation parameter images can be detected. The reference ventilation parameter image can be a baseline real image obtained from another patient, or it can be a previous image of the patient. In each case, the reference ventilation parameter image can represent the patient's target or ideal ventilation parameters, and deviations from this target or ideal ventilation parameter can indicate a problem with the patient or the ventilation system.

[0123] This disclosure also provides support for a method for a ventilation system, the method comprising: acquiring one or more patient ventilation parameter images of a patient while the patient is undergoing mechanical ventilation using the ventilation system; acquiring one or more reference ventilation parameter images; processing each patient ventilation parameter image and each reference ventilation parameter image using at least one comparison model to characterize at least one feature in each patient ventilation parameter image and each reference ventilation parameter image, the processing comprising converting each patient ventilation parameter image and each reference ventilation parameter image into a binary mask; identifying a deviation between the patient ventilation parameter image and the corresponding reference ventilation parameter image based on at least one feature in each patient ventilation parameter image and each reference ventilation parameter image; and outputting a notification indicating the deviation in response to the identification. In a first embodiment of the method, the one or more reference ventilation parameter images are from one or more previous patients and / or an artificial lung, and wherein the one or more reference ventilation parameter images are selected from a database of baseline real ventilation parameter images based on patient information and / or the settings of the ventilation system when the patient is undergoing mechanical ventilation. In a second embodiment of the method (optionally including the first embodiment), one or more patient ventilation parameter images include one or more current patient ventilation parameter images of the patient, and wherein each reference ventilation parameter image is a corresponding baseline ventilation parameter image of the patient obtained prior to obtaining one or more current patient ventilation parameter images. In a third embodiment of the method (optionally including one or both of the first and second embodiments), the method further includes: storing a previous patient ventilation parameter image as a corresponding baseline ventilation parameter image in response to the absence of deviation between a previous patient ventilation parameter image and a selected ventilation parameter image, the selected ventilation parameter image being obtained from a database of reference true ventilation parameter images based on the patient's patient information and / or ventilation system settings. In a fourth embodiment of the method (optionally including one or more or each of the first to third embodiments), one or more patient ventilation parameter images include one or more of pressure waveform images, flow waveform images, and volume waveform images. In a fifth embodiment of the method (optionally including one or more or each of the first to fourth embodiments), one or more patient ventilation parameter images include one or more of pressure / volume vital capacity measurement images and flow / volume vital capacity measurement images. In a sixth embodiment of the method (optionally including one or more or each of the first to fifth embodiments), identifying bias includes identifying that the patient is breathing spontaneously, and in response, executing a weaning protocol to determine whether the patient is ready to be weaned off mechanical ventilation.In a seventh embodiment of the method (optionally including one or more or each of the first to sixth embodiments), performing the offline procedure includes: outputting an audio command to the patient; recording a video of the patient during the output of the audio command; and analyzing the video via a comparison model selected from at least one comparison model to determine whether the patient complies with the audio command. In an eighth embodiment of the method (optionally including one or more or each of the first to seventh embodiments), the method further includes: obtaining one or more additional patient ventilation parameter images of the patient; processing each additional patient ventilation parameter image and a corresponding reference ventilation parameter image using at least one comparison model to characterize at least one feature in each additional patient ventilation parameter image and each corresponding reference ventilation parameter image; determining, based on at least one feature in each additional patient ventilation parameter image and each corresponding reference ventilation parameter image, that the settings of the ventilation system have been changed; and, in response to this determination, setting one or more additional patient ventilation parameter images as one or more reference ventilation parameter images. In a ninth embodiment of the method (optionally including one or more or each of the first to eighth embodiments), at least one comparison model is configured to perform color segmentation, edge segmentation, adaptive thresholding, standard deviation determination, bit manipulation, edge detection, contour detection, density-based clustering, and combinations thereof. In some examples, a ventilation system is configured to perform the method and optionally one or more or each of the first to ninth embodiments.

[0124] This disclosure also provides support for a ventilation system comprising: a memory storing instructions; and one or more processors configured to execute the instructions to: acquire one or more patient ventilation parameter images of a patient while the patient is receiving mechanical ventilation using the ventilation system; acquire one or more reference ventilation parameter images; compare each patient ventilation parameter image with a corresponding reference ventilation parameter image using at least one comparison model, the comparison including comparing at least one feature in each patient ventilation parameter image with a corresponding feature in the corresponding ventilation parameter image, characterizing each feature using at least one comparison model after converting each patient ventilation parameter image and each reference ventilation parameter image into a corresponding binary mask; and outputting a notification indicating the deviation and a possible cause of the deviation in response to determining, based on the comparison, that at least one of the one or more patient ventilation parameter images deviates from the corresponding reference ventilation parameter image. In a first embodiment of the system, the one or more patient ventilation parameter images include one or more of pressure waveform images, flow waveform images, volume waveform images, pressure / volume vital capacity measurement images, and flow / volume vital capacity measurement images. In a second embodiment of the system (optionally including the first embodiment), one or more reference ventilation parameter images are from one or more previous patients and / or artificial lungs, and wherein the one or more reference ventilation parameter images are selected from a database of baseline true ventilation parameter images based on the patient's patient information and / or the ventilation system settings when the patient is undergoing mechanical ventilation. In a third embodiment of the system (optionally including one or both of the first and second embodiments), one or more patient ventilation parameter images include one or more current patient ventilation parameter images of the patient, and wherein each reference ventilation parameter image is a corresponding baseline ventilation parameter image of the patient obtained prior to obtaining one or more current patient ventilation parameter images. In a fourth embodiment of the system (optionally including one or more or each of the first to third embodiments), instructions are also capable of being executed to store a previous patient ventilation parameter image as a corresponding baseline ventilation parameter image in response to no deviation between the previous patient ventilation parameter image and the selected ventilation parameter image, the selected ventilation parameter image being obtained from a database of baseline true ventilation parameter images based on the patient's patient information and / or the ventilation system settings.

[0125] This disclosure also provides support for a method comprising: acquiring a patient ventilation parameter image of a patient using a ventilation system while the patient is receiving mechanical ventilation; comparing the patient ventilation parameter image with a baseline ventilation parameter image of the patient obtained prior to acquiring the patient ventilation parameter image, based on output from at least one comparison model, wherein at least one of the at least one comparison model is configured to convert the patient ventilation parameter image and the baseline ventilation parameter image into corresponding binary masks using color segmentation; identifying a deviation between the patient ventilation parameter image and the baseline ventilation parameter image based on the comparison; outputting a notification indicating the deviation in response to the determination; acquiring a subsequent patient ventilation parameter image of the patient using the ventilation system while the patient continues receiving mechanical ventilation; and determining, based on the subsequent patient ventilation parameter image, that one or more settings of the ventilation system have been changed, and replacing the baseline ventilation parameter image with the subsequent patient ventilation parameter image in response. In a first embodiment of the method, the method further includes: storing the previous patient ventilation parameter image as a baseline ventilation parameter image in response to determining via at least one comparison model that there is no deviation between the previous patient ventilation parameter image and the selected ventilation parameter image, the selected ventilation parameter image being obtained from a database of baseline real ventilation parameter images based on the patient's patient information and / or ventilation system settings. In a second embodiment of the method (optionally including the first embodiment), the patient ventilation parameter image includes a pressure waveform image, a flow waveform image, a volume waveform image, a pressure / volume vital capacity measurement image, or a flow / volume vital capacity measurement image. In a third embodiment of the method (optionally including one or both of the first and second embodiments), comparing a patient ventilation parameter image with a baseline ventilation parameter image using at least one comparison model includes: performing one or more of color segmentation, edge segmentation, adaptive thresholding, standard deviation determination, bit manipulation, edge detection, contour detection, and density-based clustering using at least one comparison model to quantify one or more features in the patient ventilation parameter image; comparing each of the one or more features in the patient ventilation parameter image with a corresponding feature in the baseline ventilation parameter image; and based on the comparison, identifying a deviation between the patient ventilation parameter image and the baseline ventilation parameter image based on at least a threshold amount that at least one of the one or more features differs from its corresponding feature. In a fourth embodiment of the method (optionally including one or more or each of the first to third embodiments), the patient ventilation parameter image depicts a curve of a first patient ventilation parameter as a function of time or as a function of a second patient ventilation parameter, and one or more features include one or more of the following: the area enclosed by a loop in the curve, the number of intersections with the x-axis and / or y-axis of the curve, the hysteresis extension of the curve, the curvature of a segment of the loop in the curve, and the angle of another segment of the loop in the curve.In some examples, the ventilation system is configured to perform the method and optionally perform one or more or each of the first to fourth embodiments.

[0126] As used herein, elements or steps listed in the singular and beginning with the word "a" or "an" should be understood to not exclude a plurality of said elements or steps unless such exclusion is explicitly stated. Furthermore, references to "one embodiment" of the invention are not intended to be construed as excluding the existence of additional embodiments that also include the referenced features. Moreover, unless explicitly stated to the contrary, embodiments that "comprise," "include," or "have" elements or multiple elements having a particular characteristic may include additional such elements that do not have that characteristic. The terms "comprise" and "in" are used as concise linguistic equivalents to the corresponding terms "comprising" and "wherein". Furthermore, the terms "first," "second," and "third," etc., are used merely as notations and are not intended to impose numerical requirements or a particular order of position on their objects.

[0127] This written description uses examples to disclose the invention, including the best mode, and also enables those skilled in the art to practice the invention, including making and using any apparatus or system and performing any included methods. The scope of patentability of the invention is defined by the claims, but may include other examples that would occur to those skilled in the art. Such other examples are intended to fall within the scope of the claims if they have structural elements that are not indistinguishable from the literal language of the claims, or if they include equivalent structural elements that have minor differences from the literal language of the claims.

Claims

1. A method for a ventilation system, the method comprising: While the patient is receiving mechanical ventilation using the ventilation system, one or more patient ventilation parameter images of the patient are obtained using the ventilation system (504, 602); Obtain one or more reference ventilation parameter images; Each patient ventilation parameter image and each reference ventilation parameter image are processed using at least one comparative model to characterize at least one feature in each patient ventilation parameter image and each reference ventilation parameter image, the processing including converting each patient ventilation parameter image and each reference ventilation parameter image into a binary mask; Based on at least one feature in each patient ventilation parameter image and each reference ventilation parameter image, identify the deviation between (506, 508, 604, 606) patient ventilation parameter images and corresponding reference ventilation parameter images; and In response to the identification, a notification indicating the deviation is output (510, 608).

2. The method of claim 1, wherein the one or more reference ventilation parameter images are from one or more previous patients and / or with an artificial lung, and wherein the one or more reference ventilation parameter images are selected from a database of baseline real ventilation parameter images based on the patient's patient information and / or the settings of the ventilation system when the patient was undergoing mechanical ventilation.

3. The method of claim 1, wherein the one or more patient ventilation parameter images include one or more current patient ventilation parameter images of the patient, and wherein each reference ventilation parameter image is a corresponding baseline ventilation parameter image of the patient obtained prior to obtaining the one or more current patient ventilation parameter images.

4. The method of claim 3, further comprising storing (512, 618) the previous patient ventilation parameter image as a corresponding baseline ventilation parameter image in response to the absence of deviation between the previous patient ventilation parameter image and the selected ventilation parameter image, the selected ventilation parameter image being obtained from a database of baseline true ventilation parameter images based on the patient's patient information and / or the settings of the ventilation system.

5. The method of claim 1, wherein the one or more patient ventilation parameter images include one or more of pressure waveform images, flow waveform images, and volume waveform images.

6. The method of claim 1, wherein the one or more patient ventilation parameter images include one or more of pressure / volume vital capacity measurement images and flow / volume vital capacity measurement images.

7. The method of claim 1, wherein identifying the deviation includes identifying (626) that the patient is breathing spontaneously, and in response, executing a weaning protocol (700) to determine whether the patient is ready to be weaned off the mechanical ventilation.

8. The method of claim 7, wherein performing the offline scheme comprises: Output an audio command (702) to the patient; record (704) a video of the patient during the output of the audio command; The video is analyzed via a comparison model selected from the at least one comparison model to determine whether the patient complies with the audio command.

9. The method according to claim 1, further comprising: Obtain one or more additional patient ventilation parameter images of the patient (610); Each additional patient ventilation parameter image and the corresponding reference ventilation parameter image corresponding to each additional patient ventilation parameter image are processed using the at least one comparison model to characterize at least one feature in each additional patient ventilation parameter image and each corresponding reference ventilation parameter image. Based on at least one feature of each additional patient ventilation parameter image and each corresponding reference ventilation parameter image, it is determined (614) that the settings of the ventilation system have been changed; as well as In response to the determination, the one or more additional patient ventilation parameter images are set (618) to the one or more reference ventilation parameter images.

10. The method of claim 1, wherein the at least one comparison model is configured to perform color segmentation, edge segmentation, adaptive thresholding, standard deviation determination, bit manipulation, edge detection, contour detection, density-based clustering, and combinations thereof.

11. A ventilation system (100), the ventilation system comprising: Memory (102), the memory stores instructions; and One or more processors (58) are configured to execute the instructions in order to: While the patient is receiving mechanical ventilation using the ventilation system, one or more patient ventilation parameter images of the patient are obtained using the ventilation system (504, 602); Obtain one or more reference ventilation parameter images; Each patient ventilation parameter image is compared with a corresponding reference ventilation parameter image using at least one comparison model (506, 604), the comparison comprising comparing at least one feature in each patient ventilation parameter image with a corresponding feature in the corresponding ventilation parameter image, each feature being characterized by the at least one comparison model after each patient ventilation parameter image and each reference ventilation parameter image have been converted into a corresponding binary mask; as well as In response to determining, based on the comparison, that at least one of the patient ventilation parameter images is deviated from the corresponding reference ventilation parameter image, a notification (510, 608) indicating the deviation and the possible cause of the deviation is output.

12. The ventilation system of claim 11, wherein the one or more patient ventilation parameter images include one or more of pressure waveform images, flow waveform images, volume waveform images, pressure / volume vital capacity measurement images, and flow / volume vital capacity measurement images.

13. The ventilation system of claim 11, wherein the one or more reference ventilation parameter images are from one or more previous patients and / or artificial lungs, and wherein the one or more reference ventilation parameter images are selected from a data repository (116) of baseline real ventilation parameter images based on the patient's patient information and / or the settings of the ventilation system when the patient is undergoing mechanical ventilation.

14. The ventilation system of claim 11, wherein the one or more patient ventilation parameter images include one or more current patient ventilation parameter images of the patient, and wherein each reference ventilation parameter image is a corresponding baseline ventilation parameter image of the patient obtained prior to obtaining the one or more current patient ventilation parameter images.

15. The ventilation system of claim 14, wherein the instruction is further capable of storing (512) the previous patient ventilation parameter image as a corresponding baseline ventilation parameter image in response to the absence of deviation between the previous patient ventilation parameter image and the selected ventilation parameter image, the selected ventilation parameter image being obtained from a data repository (116) of a baseline true ventilation parameter image based on the patient's patient information and / or the settings of the ventilation system.

Images