A system and method for maintaining regular respiration in a patient undergoing radiotherapy for a tumor

Abstract

This invention belongs to the field of medical device technology, specifically a system and method for maintaining regular breathing in cancer radiotherapy patients. In this system, the ventilator includes a main control module, a data transmission module, a power management module, and a display module. The main control module includes a sensor data acquisition module, a respiratory waveform data storage unit, and an initial parameter setting module. The data acquisition unit includes a breathing mask and a respiratory data sensor. The breathing mask is connected to the ventilator's ventilation outlet. The respiratory data sensor is connected to the sensor data acquisition module. The sensor data acquisition module is connected to the respiratory waveform data storage unit via data transmission. The respiratory waveform data storage unit is connected to a deep learning unit via the data transmission module. The deep learning unit is also connected to the initial parameter setting module via the data transmission module. The main control module is connected to the power management module and the display module, respectively.

Description

Technical Field

[0001] This invention belongs to the field of medical device technology, specifically relating to a system and method for maintaining regular breathing in patients undergoing radiotherapy for tumors. Background Technology

[0002] Malignant lung tumors often cause respiratory failure in patients. In particular, as the nearly two-month radiotherapy course progresses, the cumulative radiation damage will increase the risk of respiratory rhythm disorder. This results in poor patient position repeatability during fractionated treatment, which to some extent affects the accuracy and effectiveness of treatment, and may even lead to off-target risk.

[0003] To address the aforementioned issues, existing respiratory motion monitoring and management products can be broadly categorized into two main types based on their operating principles:

[0004] (1) The main principle of the first type is to conduct respiratory training on lung tumor patients in advance so that their respiratory movements are controlled within a controllable range. Then, the dual-marker RPM box is placed on the patient's chest and abdomen. The maximum respiratory range in three dimensions during the treatment is monitored by an external infrared detection device. If the error exceeds the allowable error, a trigger signal is given to the linear accelerator to force it to stop irradiation until the respiratory movement error is restored to the allowable range.

[0005] (2) Another type of product mainly adopts autonomous breath-holding and free breathing technology. The method is to put a breathing mask and a visual video glasses on the patient. The device can obtain the patient's breathing movement pattern through the breathing mask and transmit it to the patient through the video glasses. Through the previous simulation training, the patient can control the breathing operation autonomously, thereby ensuring the smooth progress of the treatment process to a certain extent.

[0006] The shortcomings of the two existing types of respiratory monitoring devices are:

[0007] (1) The first type of infrared detection equipment is highly sensitive to factors such as the placement and tilt of the dual-marker RPM box, the incident direction of the examination room light, and the light intensity, which directly affect various parameters of the patient's respiratory waveform, thus significantly impacting treatment interruption and accuracy. In addition, frequent and irregular treatment interruptions undoubtedly prolong the treatment time for individual patients, increasing clinical treatment pressure and the potential for equipment malfunction.

[0008] (2) The deep inhalation and breath-holding therapy technique is not very suitable for lung tumor patients with respiratory insufficiency. In particular, as the treatment process progresses, it will aggravate the symptoms of respiratory insufficiency, thereby directly affecting subsequent treatment.

[0009] Furthermore, the two types of auxiliary treatment equipment mentioned above are relatively expensive, and radiotherapy centers in primary hospitals generally do not choose to equip them due to cost considerations. In addition, these technologies require a high level of experience and technical expertise from clinical physicists and therapists, involving issues such as quality control throughout the entire process, which limits their widespread application in primary hospitals.

[0010] The above methods are all passive monitoring methods, which cannot fundamentally improve the patient's respiratory movement patterns and effectively avoid treatment interruptions or radiation damage caused by positional deviations due to occasional breathing difficulties or coughing. Summary of the Invention

[0011] The purpose of this invention is to provide a system for maintaining regular breathing in patients undergoing radiotherapy for tumors. This system can collect data on the patient's treatment sessions before treatment, and use a deep learning model to output the expected working mode of the ventilator and various important setting parameters, thereby guiding the patient's respiratory movements during the current treatment, maintaining the regularity of the patient's respiratory movements, and successfully completing the entire treatment cycle. This fundamentally solves the problems of accuracy and efficacy of radiotherapy for lung cancer patients with respiratory insufficiency.

[0012] To achieve the above objectives, the present invention adopts the following technical solution:

[0013] A system for maintaining regular breathing in cancer radiotherapy patients, the system comprising:

[0014] Ventilator, data acquisition unit, deep learning unit;

[0015] The ventilator includes a main control module, a wireless or wired data transmission module, a power management module, and a display module;

[0016] The main control module includes a sensor data acquisition module, a respiratory waveform data storage module, and an initial parameter setting module;

[0017] The data acquisition unit includes a breathing mask and a breathing data sensor;

[0018] The breathing mask is connected to the ventilation outlet of the ventilator; the breathing data sensor is connected to the sensor data acquisition module; the sensor data acquisition module is connected to the breathing waveform data storage via data transmission; the breathing waveform data storage is connected to the deep learning unit via a wireless or wired data transmission module; the deep learning unit is also connected to the initial parameter setting module via a wireless or wired data transmission module; the main control module is connected to the power management module and the display module respectively.

[0019] Preferably, the main control module further includes a PEEP control module for setting the PEEP value.

[0020] Preferably, the deep learning unit includes a deep learning server, which embeds a convolutional network model and related extended models that deepen the network and enhance the functionality of the convolutional module.

[0021] The deep learning models used in this invention are all open source, meaning they are freely available resources or can be purchased commercially. The deep learning units and related deep learning servers can also be purchased directly.

[0022] This invention adds a data transmission module, a storage module, and an external deep learning module to the ventilator, expanding the application of the ventilator in the clinical application environment of radiotherapy.

[0023] The data transmission module and storage module can be purchased commercially.

[0024] The present invention also provides a method for maintaining regular breathing in patients undergoing radiotherapy for tumors, the method comprising the following steps:

[0025] 1) Data collection:

[0026] The data acquisition unit collects respiratory waveform data of tumor radiotherapy patients during the positioning mold making, CT positioning and fractionated radiotherapy process, stores the respiratory waveform data in the respiratory waveform data storage device, and periodically transfers it to the deep learning unit. At the same time, it classifies and stores the data by tumor radiotherapy patients.

[0027] 2) Deep learning training:

[0028] The deep learning unit is trained to process the respiratory waveform data of tumor radiotherapy patients, and outputs the trained respiratory waveform and quantified ventilator-recognizable parameters as the initial settings of the ventilator. At the same time, the parameters are tested using a respiratory simulation test device. The difference between the output waveform of the ventilator and the output waveform of the respiratory simulation test device is compared to perform deep optimization of the deep learning model and obtain the ventilator-recognizable parameters after deep learning optimization.

[0029] 3) Respiratory maintenance:

[0030] The ventilator uses parameters optimized by deep learning to assist cancer radiotherapy patients with their breathing during radiotherapy. It also collects respiratory waveform data during the current radiotherapy session and stores it in the deep learning unit as the basis for data analysis for the next radiotherapy session.

[0031] In this invention, the respiratory waveform data refers to respiratory waveform data that can be stored in text format with time as the horizontal axis and respiratory data as the vertical axis, including pressure volume loop, peak pressure, plateau pressure, positive end-expiratory pressure, air resistance, compliance, tidal volume, total frequency, and spontaneous frequency within the sampling interval.

[0032] In this invention, the parameters that the ventilator can recognize include tidal volume, target pressure, respiratory rate, inspiratory-to-expiratory ratio, end-expiratory pressure, pressure rise time, oxygen concentration, trigger pressure, and trigger flow rate.

[0033] In this invention, the respiratory simulation testing device includes a simulated lung.

[0034] This invention utilizes waveform data and initial parameters obtained from machine learning model training to set up a portable ventilator during the radiotherapy cycle for lung cancer patients. This assists the patient's respiratory movements during radiotherapy, ensuring that the patient's respiratory movements are within a relatively reasonable range. It avoids positional errors caused by arrhythmic breathing, which could lead to off-target effects and additional irradiation of normal lung tissue or vital organs.

[0035] The technical solution is mainly divided into three parts: data acquisition and transmission to the server, machine learning model training and transmission of training results to the ventilator to complete initialization, and the application of the ventilator at the treatment end to assist the patient in completing the treatment.

[0036] This invention utilizes machine learning methods to solve the problem of setting various ventilator parameters suitable for the respiratory function status of lung tumor patients at the time of treatment; solves the problem of data interaction between the ventilator's built-in storage device and the computing network via wired and wireless methods; optimizes the assisted breathing function and improves its portability;

[0037] This invention addresses the reality of respiratory dysfunction in lung cancer patients and the potential for continuous deterioration of respiratory function as radiotherapy progresses. By utilizing the continuously changing respiratory waveform data of lung cancer patients during the treatment cycle, a machine learning model is established to predict the changing patterns of this waveform. This allows for the dynamic setting of initial ventilator parameters and assists patients in limiting their respiratory movements within a reasonable range during treatment. It also improves patients' respiratory function to a certain extent, ensuring they can complete the long radiotherapy process, ultimately improving the accuracy and efficacy of tumor radiotherapy. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the system architecture of the present invention;

[0039] Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0040] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0041] Example 1

[0042] like Figure 1 As shown, a system for maintaining regular breathing in patients undergoing radiotherapy for tumors includes:

[0043] Ventilator, data acquisition unit, deep learning unit;

[0044] The ventilator includes a main control module, a wireless or wired data transmission module, a power management module, and a display module;

[0045] The main control module includes a sensor data acquisition module, a respiratory waveform data storage module, and an initial parameter setting module;

[0046] The data acquisition unit includes a breathing mask and a breathing data sensor;

[0047] The breathing mask is connected to the ventilation outlet of the ventilator; the breathing data sensor is connected to the sensor data acquisition module; the sensor data acquisition module is connected to the breathing waveform data storage via data transmission; the breathing waveform data storage is connected to the deep learning unit via a wireless or wired data transmission module; the deep learning unit is also connected to the initial parameter setting module via a wireless or wired data transmission module; the main control module is connected to the power management module and the display module respectively.

[0048] The main control module also includes a PEEP control module, which is used to set the PEEP value.

[0049] The deep learning unit includes a deep learning server, which embeds a convolutional network model and related extended models that deepen the network and enhance the functionality of the convolutional module.

[0050] like Figure 2 As shown, a method for maintaining regular breathing in cancer radiotherapy patients includes the following steps:

[0051] 1) Data collection:

[0052] Respiratory waveform data of tumor radiotherapy patients are collected during the creation of positioning molds, CT positioning, and fractionated radiotherapy. The respiratory waveform data is stored in the memory of the control module and periodically transferred to the deep learning unit. At the same time, the respiratory waveform data of tumor radiotherapy patients is collected and stored in the built-in storage device and periodically transferred to the model training server via wired / wireless means. The data is categorized and stored by tumor radiotherapy patients. In this way, a large amount of respiratory waveform data of lung tumor patients throughout the entire treatment cycle is gradually formed.

[0053] The respiratory waveform data refers to respiratory waveform data (which can be stored in text format, openable with Notepad) with time as the horizontal axis and respiratory data as the vertical axis. This data includes pressure, flow rate, and tidal volume waveforms within the sampling interval, including PV loop, peak pressure, plateau pressure, peep, air resistance, compliance, tidal volume, total frequency, and spontaneous frequency. The file header must include basic patient information (name, gender, age, and medical record ID or treatment ID associated with radiotherapy-related imaging data), acquisition time, data sampling rate and scaling ratio, as well as current ventilator settings.

[0054] 2) Building a deep learning model:

[0055] A deep learning model is used to process the respiratory waveform data of tumor radiotherapy patients stored on the model training server. The trained respiratory waveform and quantified ventilator-recognizable parameters are output as the initial settings to guide the ventilator's assisted breathing function. The respiratory simulation test equipment is used to test according to the above parameters, and the difference between the ventilator output waveform and the model output waveform is compared to perform in-depth optimization of the training model.

[0056] The deep learning model employs Convolutional Neural Networks (CNN) models and related extended models that deepen the network and enhance the functionality of convolutional modules for machine learning. Simultaneously, based on the actual situation, 3D localization CT images of the patient are incorporated to extract imaging features such as the patient's entire lungs and tumor lesion size, which are then used in model building and optimization.

[0057] The ventilator identifiable parameters refer to all parameters that a portable imaging ventilator supports for ventilation functions, such as tidal volume, target pressure, respiratory rate, inspiratory-expiratory ratio, end-expiratory pressure, pressure rise time, oxygen concentration, trigger pressure, and trigger flow rate.

[0058] Breathing simulation test equipment: A device used to simulate a patient's breathing movements, such as a test lung. The purpose is to generate a set of simulated breathing waveforms to test the stability and reliability of the entire system.

[0059] For example, IMT Analytics' simulated lung products provide a simple and efficient method for accurately and reliably testing the functionality and precision of ventilators and anesthesia machines. Due to their strong adaptability, these simulated lung products are well-suited for use as part of training to simulate different types of ventilation. For instance, they can be used to simulate insufficient ventilation due to accidental leaks, or to test the accuracy of ventilators for premature infants.

[0060] The ventilator output waveform and the model output waveform refer to the actual breathing waveform generated by simulating the lungs, and the waveform data (predicted waveform data) generated at the ventilator output end under the intervention of the corresponding ventilator setting parameters output by the deep learning network.

[0061] Both types of waveform data can be used for the stability and reliability of the entire system, as well as for the accuracy testing and debugging of deep learning models (parameter tuning of deep learning models).

[0062] 3) Maintain regular breathing in patients undergoing radiotherapy for tumors:

[0063] During actual treatment, the patient's respiratory waveform data was collected at each stage following the same steps. This data was used to train and validate the model, further optimizing the network structure and related weights. During treatment, the patient wore a breathing mask and was connected to a portable ventilator. The parameters were set according to the model training parameters to assist the patient's breathing throughout the treatment process. Respiratory waveform data for each treatment session was collected and stored on the model training server, serving as the basis for data analysis in the next radiotherapy session.

[0064] In practice, patients require mechanical ventilation to assist in regulating their breathing. This involves the following points:

[0065] 1. Before each treatment, use a ventilator with data acquisition and transmission capabilities to collect respiratory waveform data (for a specific time period, such as 5-10 minutes during a quiet breathing state).

[0066] This process generates two sets of data: ① a dataset of respiratory waveforms before treatment (5-10 minutes) ② a set of initial ventilator settings;

[0067] 2. The data is transmitted to a machine deep learning server (wired or wireless) and outputs the initial settings parameters required by the ventilator (suitable for the patient's respiratory state under the current conditions). During the treatment, the ventilator assists the patient in completing respiratory movements throughout the entire treatment period, maintains their rhythmic breathing, and avoids accidental irradiation caused by large respiratory fluctuations (avoiding additional radiation damage). This process generates a set of data: a dataset of the patient's respiratory waveforms during the treatment period.

[0068] The three sets of data—pre-treatment respiratory waveform dataset, initial ventilator setting parameter set, and patient respiratory waveform dataset during treatment—can be used for further training and testing (validation) of the deep learning model. Optimization mainly refers to network structure optimization and adjustment of relevant weights, while testing refers to testing the prediction accuracy.

[0069] All aspects not described in detail in this invention can be covered using conventional technical knowledge in the field.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the embodiments, those skilled in the art should understand that modifications or equivalent substitutions to the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A system for maintaining regular breathing in patients undergoing radiotherapy for tumors, characterized in that, The system includes: Ventilator, data acquisition unit, deep learning unit; The ventilator includes a main control module, a data transmission module, a power management module, and a display module; The main control module includes a sensor data acquisition module, a respiratory waveform data storage module, and an initial parameter setting module; The data acquisition unit includes a breathing mask and a breathing data sensor; The breathing mask is connected to the ventilation outlet of the ventilator; the breathing data sensor is connected to the sensor data acquisition module; the sensor data acquisition module is connected to the breathing waveform data storage via data transmission; the breathing waveform data storage is connected to the deep learning unit via the data transmission module; the deep learning unit is also connected to the initial parameter setting module via the data transmission module; the main control module is connected to the power management module and the display module respectively. The data acquisition unit is used to collect respiratory waveform data of tumor radiotherapy patients during the process of positioning mold making, CT positioning and fractionated radiotherapy. The respiratory waveform data is stored in the respiratory waveform data storage device and periodically transferred to the deep learning unit. At the same time, it is categorized and stored by tumor radiotherapy patients. The deep learning unit is trained to process the respiratory waveform data of tumor radiotherapy patients, and outputs the trained respiratory waveform and quantified ventilator-recognizable parameters as the initial settings of the ventilator. At the same time, the parameters are tested using a respiratory simulation test device. The difference between the output waveform of the ventilator and the output waveform of the respiratory simulation test device is compared to perform deep optimization of the deep learning model and obtain the ventilator-recognizable parameters after deep learning optimization. The ventilator uses parameters that can be recognized by the ventilator after deep learning optimization to assist the breathing of tumor radiotherapy patients during radiotherapy. It also collects respiratory waveform data during the current radiotherapy and stores it in the deep learning unit as the basis for data analysis of the next radiotherapy. Respiratory waveform data refers to respiratory waveform data that can be stored in text format with time as the horizontal axis and respiratory data as the vertical axis. It includes pressure-volume loop, peak pressure, plateau pressure, positive end-expiratory pressure, air resistance, compliance, tidal volume, total frequency, and spontaneous frequency within the sampling interval. The parameters that the ventilator can recognize include tidal volume, target pressure, respiratory rate, inspiratory-to-expiratory ratio, end-expiratory pressure, pressure rise time, oxygen concentration, trigger pressure, and trigger flow rate.

2. The system for maintaining regular breathing in tumor radiotherapy patients according to claim 1, characterized in that, The main control module also includes a PEEP control module, which is used to set the PEEP value.

3. A system for maintaining regular breathing in a tumor radiotherapy patient according to claim 1 or 2, characterized in that, The data transmission module is a wireless or wired data transmission module.

4. The system for maintaining regular breathing in tumor radiotherapy patients according to claim 1, characterized in that, The deep learning unit includes a deep learning server, which embeds a convolutional network model and related extended models that deepen the network and enhance the functionality of the convolutional module.

5. The system for maintaining regular breathing in a tumor radiotherapy patient according to claim 1, characterized in that, The respiratory simulation testing equipment includes a simulated lung.

Images