[go: up one dir, main page]

WO2024108409A1 - Procédé et système d'imagerie en quatre dimensions sans contact basés sur un signal respiratoire de surface en quatre dimensions - Google Patents

Procédé et système d'imagerie en quatre dimensions sans contact basés sur un signal respiratoire de surface en quatre dimensions Download PDF

Info

Publication number
WO2024108409A1
WO2024108409A1 PCT/CN2022/133611 CN2022133611W WO2024108409A1 WO 2024108409 A1 WO2024108409 A1 WO 2024108409A1 CN 2022133611 W CN2022133611 W CN 2022133611W WO 2024108409 A1 WO2024108409 A1 WO 2024108409A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
body surface
data
phase
respiratory
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2022/133611
Other languages
English (en)
Chinese (zh)
Inventor
张艺宝
董正坤
余疏桐
李莎
华凌
刘宏嘉
吴昊
李俊禹
卢子红
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Cancer Hospital
Original Assignee
Beijing Cancer Hospital
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Cancer Hospital filed Critical Beijing Cancer Hospital
Priority to CN202280005277.6A priority Critical patent/CN115996668B/zh
Priority to PCT/CN2022/133611 priority patent/WO2024108409A1/fr
Publication of WO2024108409A1 publication Critical patent/WO2024108409A1/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/08—Measuring devices for evaluating the respiratory organs
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—2D [Two Dimensional] image generation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration

Definitions

  • the present invention is applicable to but not limited to various application scenario technical fields such as radiotherapy, diagnosis, intervention, puncture, etc., and in particular relates to a non-contact four-dimensional imaging method and system based on four-dimensional body surface respiratory signals.
  • Respiratory motion not only causes image artifacts, but also reduces the accuracy of image-guided diagnosis and treatment. For example, dose delivery errors in radiotherapy can lead to tumor control failure and normal organ damage, and errors in needle insertion position and time can lead to pathological biopsy failure or damage to important organs. Even if the body's range of motion is limited by various external constraints during image scanning, respiratory motion can still cause movement and deformation of internal organs.
  • the existing detection methods mainly include: methods based on pressure sensors, such as contact abdominal pressure belts, which require additional devices to be attached to the patient, and the detection signal is 1D data that changes with time; methods based on the reflection signal of reflective blocks placed on the chest and abdomen to detect the movement amplitude and frequency in the dorsal and ventral directions, which is also contact-based and requires additional devices to be attached to the patient, and the detection signal is 1D data that changes with time; based on real-time projection image segmentation of anatomical markers, such as the position change of the diaphragm, each imaging field/field of view requires the diaphragm to be included.
  • one-dimensional motion signals may oversimplify complex motion models, and contact devices have limitations in terms of ease of use and system compatibility; diaphragm image segmentation methods have many restrictions on imaging sites and fields, and for real-time guidance of respiratory gated radiotherapy, this method may involve additional imaging radiation doses or expensive and complex MR equipment integration.
  • the main purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a non-contact four-dimensional imaging method and system based on four-dimensional body surface respiratory signals.
  • the present invention provides a non-contact four-dimensional imaging method based on four-dimensional body surface respiratory signals.
  • the optical body surface data of the patient is synchronously acquired; the respiratory signal is extracted based on the acquired optical body surface data; the patient image data is grouped based on the respiratory phase information in the respiratory signal to reconstruct a four-dimensional image.
  • the optical body surface data is collected using a device that can obtain three-dimensional coordinate information including but not limited to plane and depth, and some may also include color information or reflection intensity information.
  • a certain phase of the image is taken as the reference phase of the image, all optical surface data are aligned with the reference phase of the image, the deformation after alignment and the displacement of the reference phase are calculated in turn to obtain a displacement matrix, the displacement matrix is subjected to dimensionality reduction processing and the component with the largest contribution is selected, for example, using dimensionality reduction methods such as principal component analysis, factor analysis, and independent component analysis to obtain an optical surface respiratory signal.
  • the image data is segmented, and the surface information of the segmented body surface is saved, and a region of interest is selected on the segmented body surface to obtain segmented body surface data; the optical body surface data is registered with the segmented body surface data.
  • Deep learning technology is used to improve the image quality of direct grouping reconstruction.
  • Deep learning technology includes deep neural networks, or recursive neural networks, or convolutional neural networks, or generative adversarial networks, etc.
  • a multi-channel neural network is used to input multiple adjacent phases of the patient as prior data at one time to learn the motion association between adjacent phases, obtain a phase-by-phase training model, and use the phase-by-phase training model to improve image quality.
  • the image with improved quality is obtained and used to obtain the displacement vector field reconstruction.
  • a certain phase group among the N groups of images is used as the reference phase, and the reference phase image is sequentially aligned to the other N-1 groups of phases to obtain N-1 displacement vector fields, where N is an integer greater than 1.
  • the present invention provides a non-contact four-dimensional imaging system based on a four-dimensional body surface respiratory signal, comprising: an acquisition unit, used to synchronously acquire the optical body surface data of the patient during the process of acquiring the patient's image; and a processing unit, used to extract the respiratory signal based on the acquired optical body surface data; and grouping the patient image data based on the respiratory phase information in the respiratory signal to reconstruct a four-dimensional image.
  • the system includes a device capable of collecting plane and depth information of optical body surface data; some devices may also include color information or reflection intensity information.
  • the processing unit takes a certain phase of the image as the reference phase of the image, aligns all optical surface data with the reference phase of the image, calculates the deformation after alignment and the displacement of the reference phase in sequence, obtains a displacement matrix, performs dimensionality reduction processing on the displacement matrix, such as principal component analysis, factor analysis, independent component analysis and other dimensionality reduction methods, to obtain an optical surface respiratory signal.
  • the processing unit is used to segment the image data, save the surface information of the segmented body surface, select a region of interest on the segmented body surface to obtain segmented body surface data; and align the optical body surface data with the segmented body surface data.
  • the processing unit is used to improve the image quality of direct group reconstruction using deep learning technology based on the image data and respiratory signals collected before treatment as a training set.
  • the deep learning technology includes a deep neural network, a recursive neural network, a convolutional neural network, a generative adversarial network, etc.
  • the processing unit is used to adopt a multi-channel neural network, input multiple adjacent phases of the patient as prior data at one time, learn the motion association between adjacent phases, obtain a phase-divided training model, and use the phase-divided training model to improve image quality.
  • the processing unit is used to obtain an image with improved quality, which is used to obtain a displacement vector field reconstruction, and a certain phase group among the N groups of images is used as a reference phase group, and the reference phase image is sequentially aligned to the other N-1 groups of phase groups to obtain N-1 displacement vector fields, where N is an integer greater than 1.
  • the present invention proposes a non-contact four-dimensional imaging method and system based on four-dimensional body surface respiratory signals, which can be used for respiratory phase division and image reconstruction in imaging processes including but not limited to four-dimensional CT, four-dimensional cone beam CT, four-dimensional nuclear magnetic resonance, etc., and can also be used for scenes such as radiotherapy and puncture guided by respiratory gating.
  • the images are sorted and classified according to the respiratory phases to facilitate reconstruction and display, and four-dimensional images including but not limited to four-dimensional cone beam CT images are realized.
  • the method obtains respiratory phase information from the optical body surface and realizes non-contact respiratory phase grouping.
  • the dynamic display and monitoring of the anatomical structure in the body is realized through four-dimensional imaging, which is conducive to improving the accuracy of clinical diagnosis and treatment, and there is no need to attach a contact device to the patient, which is suitable for a wider range of body parts, imaging modalities and clinical application scenarios. It forms good complementarity with the prior art in terms of applicable but not limited to radiotherapy, diagnosis, intervention, puncture and other application scenarios, data dimensions and technical indicators, economic costs, etc.
  • FIG1 is a schematic diagram of a non-contact four-dimensional imaging system based on four-dimensional body surface respiratory signals
  • FIG2 is a flow chart of a non-contact four-dimensional imaging method based on four-dimensional body surface respiratory signals
  • FIG3 is a schematic diagram of simulated body surface point cloud data generated based on simulated phantom CT
  • FIG4 is a schematic diagram of ROI selection based on simulated body surface point cloud data generated by simulated phantom CT;
  • FIG. 1 Schematic diagram of ground truth signal
  • FIG. 6 Schematic diagram of optical surface signal
  • Fig. 7 Diagram of diaphragm signal
  • Fig. 8 Schematic diagram of the transverse plane of the reconstruction results of the MLEM (Maximum Likelihood Expectation Maximization algorithm) for the three types of male simulated data;
  • Fig. 9 Schematic diagram of the coronal plane reconstruction results of the MLEM (Maximum Likelihood Expectation Maximization algorithm) for the three types of male simulated data;
  • Figure 10 Schematic diagram of the sagittal plane reconstruction results of the MLEM (Maximum Likelihood Expectation Maximization algorithm) for the three types of male simulated data.
  • system and unit used herein are a method for distinguishing different components, elements, parts, parts or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.
  • system and unit can be implemented by software or hardware, and can be a physical or virtual name for the functional part.
  • the present invention uses a flow chart to illustrate the operations performed by the system according to an embodiment of the present invention. It should be understood that the preceding or following operations are not necessarily performed precisely in order. On the contrary, the various steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes. The technical solutions in the various embodiments may be combined with each other to achieve the purpose of the present invention.
  • Embodiment 1 is a diagrammatic representation of Embodiment 1:
  • a non-contact four-dimensional imaging system based on a four-dimensional body surface respiratory signal comprises: an acquisition unit for synchronously acquiring optical body surface data of the patient during the process of acquiring the patient's image; a processing unit for extracting the respiratory signal based on the acquired optical body surface data; and grouping the patient's image data based on the respiratory phase information in the respiratory signal to reconstruct a four-dimensional image.
  • the present invention can be used for respiratory phase division and image reconstruction in imaging processes such as four-dimensional CT, four-dimensional cone-beam CT, and four-dimensional nuclear magnetic resonance (not limited to these modalities).
  • Respiratory signals are extracted based on the collected optical body surface data to generate four-dimensional images including but not limited to four-dimensional cone-beam CT images.
  • the method directly obtains respiratory phase information based on the optical body surface, and realizes projection grouping by phase without the need for additional contact sensors.
  • the motion data extracted by this system has a higher dimension, a non-contact advantage, is suitable for imaging the entire chest and abdomen, has better compatibility with other systems, is low cost, and does not involve additional imaging radiation doses and complex system integration in the real-time guidance process of respiratory gated radiotherapy.
  • Embodiment 2 is a diagrammatic representation of Embodiment 1:
  • the present invention provides a non-contact four-dimensional imaging method based on a four-dimensional body surface respiratory signal.
  • the optical body surface data of the patient is synchronously acquired; a respiratory signal is extracted based on the acquired optical body surface data; the patient image data is grouped based on the respiratory phase information in the respiratory signal to reconstruct a four-dimensional image.
  • This method corresponds to the first embodiment, has the same beneficial effects, and will not be described in detail.
  • the present invention is applicable to a variety of application scenarios such as radiotherapy, diagnosis, intervention, and puncture, and the method is the same.
  • a CBCT image of the patient is taken.
  • the image can be 3D, and then AI complements the sparse reconstruction, or it can be a sufficient projection of 4D.
  • the patient's optical surface data is synchronously acquired, and the respiratory signal is acquired based on the optical surface data acquired during the CBCT acquisition process.
  • the human body's respiratory process is divided into 10 phases according to the amplitude of inhalation and exhalation, that is, the inhalation state is divided into In 0%, In 20%, In 40%, In 60%, In 80%, In 100%, and the corresponding exhalation state is divided into Out 20%, Out 40%, Out 60%, Out 80%.
  • In 100% is the end state of inhalation
  • In 0% is the end state of exhalation.
  • the patient image data is grouped based on the respiratory phase information in the respiratory signal to reconstruct a four-dimensional image.
  • There are many current methods for extracting respiratory signals including using a respiratory belt sensor to monitor respiratory signals, using motion signals of external or built-in devices as respiratory signals, etc.
  • the present invention proposes a method for obtaining real-time respiratory signals based on four-dimensional optical body surface data acquired by a depth camera, and grouping the relative image data based on breathing to obtain the 4DCBCT of the day. And the image quality is improved by deep learning technology.
  • the present embodiment is divided into 10 phases, which are not limited to 10 phases in practice, and can also be divided into other number of phases, such as other 2-100 arbitrary integers.
  • the images of the previous treatment can also be used; the images taken are not limited to 3D C BCT images, but also include 4D cone beam CT, MR, CT and other images.
  • the optical body surface data of the present invention is derived from an external depth camera.
  • the body surface data and images can be realized by software.
  • the two sets of image acquisition systems can be realized by internal communication.
  • the present invention also proposes a solution: using the depth camera to track the movement of components during image data acquisition. This solution does not require any additional hardware equipment, is easy to promote, and has good reliability.
  • the only condition that needs to be met is that there are components that are easy to identify and rotate synchronously in the area where the camera collects images, such as flat-panel detectors, X-ray sources, accelerator racks or heads in four-dimensional cone-beam CT imaging.
  • the basic idea of this solution is to select a suitable feature area on a component that rotates randomly, calculate the position parameters of the feature area in the image, and combine the change of the parameter with a pre-calibrated conversion function to obtain the rack rotation angle that corresponds to the image captured by the depth camera in real time, thereby realizing the time synchronization of the camera and image acquisition.
  • the conversion function can use a suitable linear function.
  • one embodiment is to track the position of the elliptical contour on the flat-panel detector in the blue channel of the three optical primary colors RBG when the camera collects data to ensure synchronization.
  • the calibration of the conversion function should be completed first.
  • the specific position of the elliptical contour is determined by using threshold segmentation and ellipse fitting, and the distance from the characteristic center of the ellipse to the upper left corner of the image is calculated. The multiple results at the same position are averaged as the characteristic distance parameter corresponding to the angle.
  • depth cameras can also obtain depth information of the photographed object, that is, three-dimensional position and size information.
  • the entire computing system then obtains three-dimensional data of the environment and the object. When combined with time, four-dimensional data can be formed. This information can be used for human body tracking and three-dimensional reconstruction.
  • the present invention is not limited to the use of a depth camera to collect optical body surface data, and all devices that can obtain body surface plane and depth information can be used.
  • the optical body surface respiratory signal is obtained and four-dimensional image reconstruction is performed.
  • a test is performed based on the XCAT simulation phantom.
  • the 1320 simulated CT data generated by XCAT are segmented, and the segmented body surface is saved in the form of point cloud data.
  • the region of interest (ROI) is selected from the chest and abdomen of the body surface point cloud.
  • the size of the ROI is 32x32.
  • the point cloud form of the simulation phantom is shown in Figure 3, and the ROI display is shown in Figure 4.
  • the body surface data is three-dimensional coordinate information. This embodiment uses the point cloud data form to save the body surface data, but the present invention is not limited to the use of the point cloud data form.
  • All forms of expression of the set data that can express a set of vectors in a three-dimensional coordinate system are acceptable; for example, grid data, etc. can also store surface information, and the actual operation may be similar to point cloud. For example, ply stores grid data, which itself includes both point cloud and surface.
  • Point cloud processing of simulated data is simpler than that of real data.
  • ROI of point cloud data and preprocessing of point cloud data will be very important.
  • This part of preprocessing includes noise reduction of point cloud data and selection of downsampling algorithm.
  • Point cloud downsampling is to resample the point cloud according to certain sampling rules. The purpose is to reduce the density of the point cloud while ensuring that the overall geometric features of the point cloud remain unchanged, thereby reducing the amount of data and algorithm complexity of related processing.
  • the subsequent processing is mainly to perform point cloud registration on the selected ROI point cloud data in ply format.
  • the registration method used in the present invention is to register point cloud data based on Gaussian mixture model.
  • the core idea of this registration framework is to use Gaussian distribution to represent the input point set mixture model.
  • the point set registration problem is restated as the registration problem of two Gaussian mixture models (GMM), that is, by minimizing the statistical difference between the two Gaussian mixtures, which is intuitively expressed as calculating the L2 distance between the two Gaussian mixtures, and obtaining the L2 value feedback registration effect through continuous iteration to obtain the optimal solution.
  • GMM Gaussian mixture models
  • Registration refers to the alignment of two or more images of the same target in spatial position.
  • the first phase of the CT data is used as the reference phase
  • two Gaussian mixture models (GMM) point cloud registration methods are used to register all subsequent optical surface data with the reference phase, and the displacement matrix of all points of the registered deformation ROI and the reference ROI is calculated in turn, and the displacement matrix is subjected to principal component analysis (PCA).
  • PCA principal component analysis
  • the first principal component is the respiratory signal of the optical body surface data required by the present invention, which can be used to reduce noise and divide the respiratory phase.
  • using the first principal component as the respiratory signal of the optical body surface data is not the only solution. It is also possible to use the first principal component plus the second principal component, or even the first principal component plus more principal components.
  • Principal component analysis of the displacement matrix is a dimensionality reduction processing method, the purpose of which is to select the component with the largest contribution. In addition to using principal component analysis, dimensionality reduction methods such as factor analysis and independent component analysis can also be used to obtain the optical body surface respiratory signal.
  • Gaussian noise reduction is used to reduce the noise of the raw signal of the point cloud data to make the curve smoother, which is more conducive to the realization of normalization.
  • the selection of ROI does not need to specifically distinguish between the abdomen and the chest.
  • the ROI selection range is too large, it will not only affect the registration speed, but also affect the accuracy of curve extraction due to the large difference in the movement amplitude of the chest and abdomen. Therefore, the selection of real data ROI should pay special attention to the division of regions.
  • K-mean clustering can be performed on the displacement matrix to distinguish between abdominal data and chest data.
  • the solution of the present invention also solves the technical problem that there will be obvious artifacts when the number of projections is insufficient, and uses deep learning technology, including but not limited to: deep neural network (DNN), recursive neural network (RNN), convolutional neural network (CNN), and generative adversarial network (GAN) to improve the image quality of time-phase.
  • DNN deep neural network
  • RNN recursive neural network
  • CNN convolutional neural network
  • GAN generative adversarial network
  • the diaphragm signal or the respiratory signal obtained based on the optical surface information proposed by the present invention is used to divide the projection into phases and reconstruct the 4DCBCT with sparse artifacts.
  • Time-phase training of unpaired 4DCT and 4DCBCT (the training set can be updated and expanded regularly based on the update of hospital data, and male and female physiological structure differences should be taken into consideration, so that male and female should be trained separately) improves the quality of image reconstruction directly grouped for current patients, and real-time image quality improvement can be achieved on the day of treatment for the current patient.
  • the improvement of image quality through deep learning technology makes four-dimensional cone-beam CT reconstruction under conventional cone-beam CT scanning possible.
  • time-phase training can also be used to improve image quality after the solution may cause detail loss, so as to improve image details.
  • Phase-by-phase training can realize a deep learning network for motion compensation: a multi-channel neural network is designed to input multiple adjacent phases of the patient's 4DCT as priors at one time to learn the motion association between adjacent phases and obtain a phase-by-phase training model. Based on the above model, sparse reconstruction artifacts are eliminated and the details of the image area of interest are supplemented and corrected.
  • Solution 2 Phase-by-phase deep learning plus motion compensation: Apply deep learning technology to improve the image quality of direct group reconstruction.
  • the deep learning improvement part can refer to the solution given in the above solution 1.
  • the improved image can be used to obtain the displacement vector field (DVF) and for motion compensation reconstruction.
  • the acquisition of DVF can be achieved by using isotropic total variation registration (pTV-reg).
  • pTV-reg isotropic total variation registration
  • phase 0 is used as the reference phase
  • the reference phase image is sequentially registered to the other 9 phases to obtain 9 groups of DVF, which are used to realize motion correction in the forward and backward projection process in iterative reconstruction of different phases. It is not limited to dividing into 10 phases. In practice, it can be divided into N groups, where N is an integer greater than 1.
  • the image quality can be improved by using deep learning techniques such as generative adversarial networks.
  • the optical surface respiratory signals including the tumor area at the end of expiration (EOE) stage can be registered to the deep neural network.
  • a generative adversarial network including a generator and a discriminator is used to train the generative adversarial network using the patient's 4DCT prior data samples. After the training, the generator with determined parameters is extracted as the artifact removal model. It is not limited to the optical surface respiratory signals at the end of expiration, but can also be the optical surface respiratory signals that are divided into various phases according to the amplitude of inspiration and expiration during the respiratory process.
  • the samples are not limited to the patient's 4DCT prior data, but can also be various types of 4D data such as 4DCBCT.
  • the 4DCBCT obtained in each treatment based on the method proposed by the present invention will be used as a training set. Combined with the optical body surface data during the treatment process, real-time image guidance can be used for treatment work.
  • the estimated displacement vector field (DVF) is used for real-time detection of the lesion centroid and target area contour to obtain the final four-dimensional cone-beam CT image.
  • This patent is based on the respiratory signal obtained from the optical body surface data collected during the 3D image acquisition process, divides the projection into phases based on the respiratory signal, and reconstructs a 4D image. In the simulation stage, the simulated optical body surface signal is compared and evaluated with other signals.
  • the signals of the extracted optical body surface data are evaluated and compared with the input diaphragm signal true value (input signal), and the accuracy and feasibility of the body surface signal extraction based on registration are explored.
  • the evaluation indicators are the fast Fourier transform (fft) of the normalized signal to extract the spectrum, the comparison of the calculation cycle with the input diaphragm signal true value (input signal), and the peak and trough errors between the extracted optical body surface data signal and the true signal.
  • the signal evaluation shows that the stability of the optical body surface signal extraction proposed by the present invention is no less than that of the diaphragm signal extraction, and is even closer to the input signal.
  • the diaphragm signal (Diaphragm signal) extracted in the previous article, the simulated optical surface signal (Optical Surface signal) and the input diaphragm signal true value (Ground truth) are used to reconstruct and compare the reconstruction results, as shown in Figures 5, 6 and 7.
  • the reconstruction algorithm adopts the MLEM (Maximum Likelihood Expectation Maximization algorithm), and the evaluation indicators are mainly RMSE, SSIM, and PSNR.
  • the true value is the full projection reconstruction result generated by the simulated 4DCT mentioned above.
  • the image quality indicators used in the present invention include the root mean square error (RMSE), structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR) of the image.
  • RMSE root mean square error
  • SSIM structural similarity index
  • PSNR peak signal-to-noise ratio
  • ⁇ x and ⁇ y are the means of images x and y.
  • ⁇ x and ⁇ y are the variances x and y.
  • ⁇ xy is the covariance of x and y.
  • h1 and h2 are two constants.
  • L is the range of pixel values in x and y.
  • the PSNR of a given image is defined as:
  • Figure 5 is a schematic diagram of the ground truth signal
  • Figure 6 is a schematic diagram of the optical surface signal
  • Figure 7 is a schematic diagram of the diaphragm signal
  • Figures 8 to 10 are schematic diagrams of the reconstruction results of the MLEM (Maximum Likelihood Expectation Maximization algorithm) for the three types of male simulated data.
  • MLEM Maximum Likelihood Expectation Maximization algorithm
  • the RMSE is less than 3, the SSIM is greater than 0.94, and the PSNR is between 35-39; and the quality evaluation of the diaphragm signal reconstructed image is not much different from that of the optical surface signal reconstructed image; it can be seen that the reconstruction of 4D images based on the optical surface signal can achieve results comparable to those obtained using the signal obtained by internal motion.
  • Figures 8 to 10 are male data, but the present invention is applicable to the reconstruction of respiratory images of men and women.
  • the method of the present invention directly obtains respiratory phase information from the optical body surface, and realizes projection grouping by phase without the need for a contact sensor.
  • this method can not only be used for projection data that does not contain diaphragm signals in the projection, but the optical body surface can be collected in real time during treatment.
  • the optical body surface information has a higher dimension and does not require a contact auxiliary device.
  • a non-contact four-dimensional imaging method and system based on four-dimensional body surface respiratory signals provided by the present invention can be based on existing equipment, does not rely on contact sensors, and can effectively estimate four-dimensional cone-beam CT images from conventional cone-beam CT scans. It can be used for respiratory phase division and image reconstruction in imaging processes including but not limited to four-dimensional CT, four-dimensional cone-beam CT, four-dimensional nuclear magnetic resonance, etc. This method will not introduce a higher radiation dose and is expected to be used in existing clinical treatment processes.
  • the 4DCBCT of the day based on the 3DCBCT before treatment and the signal collected from the optical body surface is synthesized for treatment.
  • beneficial effects may be any one or a combination of the above, or may be any other possible beneficial effects.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Pathology (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Animal Behavior & Ethology (AREA)
  • Surgery (AREA)
  • Medical Informatics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Mathematical Physics (AREA)
  • Pulmonology (AREA)
  • Physiology (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Abstract

La présente invention divulgue un procédé et un système d'imagerie à quatre dimensions sans contact basés sur un signal respiratoire de surface à quatre dimensions. Le procédé consiste : pendant le processus de collecte d'images d'un patient, à collecter de manière synchrone des données de surface optique du patient ; à extraire un signal respiratoire sur la base des données de surface optique collectées ; et à grouper des données d'image du patient sur la base d'informations de phase respiratoire dans le signal respiratoire et à reconstruire une image en quatre dimensions. Dans la présente invention, les informations de phase respiratoire sont acquises à partir d'une surface optique, de telle sorte qu'un regroupement de phase respiratoire sans contact soit mis en œuvre ; par comparaison avec l'état de la technique, des données de détection présentent une dimension plus élevée et un affichage et une surveillance dynamiques de structures anatomiques intracorporelles sont également mis en œuvre au moyen d'une imagerie en quatre dimensions, de telle sorte qu'une amélioration de la précision du diagnostic et du traitement cliniques soit facilitée ; en outre, il n'est pas nécessaire de fixer des appareils de contact aux corps de patients, la présente invention est donc appropriée pour une plus large plage de parties de corps, de modalités d'imagerie et de scénarios d'application clinique.
PCT/CN2022/133611 2022-11-23 2022-11-23 Procédé et système d'imagerie en quatre dimensions sans contact basés sur un signal respiratoire de surface en quatre dimensions Ceased WO2024108409A1 (fr)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202280005277.6A CN115996668B (zh) 2022-11-23 2022-11-23 一种基于四维体表呼吸信号的四维成像方法和系统
PCT/CN2022/133611 WO2024108409A1 (fr) 2022-11-23 2022-11-23 Procédé et système d'imagerie en quatre dimensions sans contact basés sur un signal respiratoire de surface en quatre dimensions

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/133611 WO2024108409A1 (fr) 2022-11-23 2022-11-23 Procédé et système d'imagerie en quatre dimensions sans contact basés sur un signal respiratoire de surface en quatre dimensions

Publications (1)

Publication Number Publication Date
WO2024108409A1 true WO2024108409A1 (fr) 2024-05-30

Family

ID=85992530

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/133611 Ceased WO2024108409A1 (fr) 2022-11-23 2022-11-23 Procédé et système d'imagerie en quatre dimensions sans contact basés sur un signal respiratoire de surface en quatre dimensions

Country Status (2)

Country Link
CN (1) CN115996668B (fr)
WO (1) WO2024108409A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN119112107A (zh) * 2024-09-09 2024-12-13 合肥工业大学 轻量化的非接触式多参数生理检测方法及系统

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117372565B (zh) * 2023-12-06 2024-03-15 合肥锐视医疗科技有限公司 一种基于神经网络时相判别的呼吸门控ct成像方法
CN119606399B (zh) * 2024-10-10 2025-08-05 广州科莱瑞迪医疗器材股份有限公司 一种4d-ct图像重建方法

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017180097A1 (fr) * 2016-04-12 2017-10-19 Siemens Aktiengesellschaft Alignement déformable d'entrées peropératoire et préopératoire à l'aide de modèles de mélange génératifs et d'une déformation biomécanique
CN110613519A (zh) * 2019-09-20 2019-12-27 真健康（北京）医疗科技有限公司 动态配准定位装置及方法
CN111179409A (zh) * 2019-04-23 2020-05-19 艾瑞迈迪科技石家庄有限公司 一种呼吸运动建模方法、装置和系统
CN112154483A (zh) * 2020-07-15 2020-12-29 北京肿瘤医院(北京大学肿瘤医院) 一种利用光学体表运动信号合成实时图像的方法及系统
CN112435307A (zh) * 2020-11-26 2021-03-02 浙江大学 一种深度神经网络辅助的四维锥束ct影像重建方法
CN113574565A (zh) * 2021-02-07 2021-10-29 北京肿瘤医院(北京大学肿瘤医院) 一种利用光学体表运动信号合成实时图像的系统
CN114820855A (zh) * 2022-04-28 2022-07-29 浙江大学滨江研究院 一种基于患者4d-ct的肺部呼吸过程图像重构方法及装置

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2009004951A (ja) * 2007-06-20 2009-01-08 Fujifilm Corp 画像処理装置、方法、およびプログラム
CN102058936A (zh) * 2010-11-17 2011-05-18 山东省肿瘤防治研究院 一种用于胸部肿瘤放射治疗的呼吸控制定位装置
CN103222874B (zh) * 2012-01-31 2016-12-07 Ge医疗系统环球技术有限公司 拣选ct切片图像的方法和构建ct三维图像的方法
CN104305996A (zh) * 2014-10-15 2015-01-28 大连现代医疗设备科技有限公司 基于图像识别技术的呼吸监测方法及系统
JP6386402B2 (ja) * 2015-03-20 2018-09-05 トヨタ自動車株式会社 呼吸に伴う人間の体表面の変位を検出するセンサ装置
CN108451532B (zh) * 2018-04-28 2020-05-01 杜乙 呼吸信息的获取方法及系统
CN110390361B (zh) * 2019-07-25 2021-04-09 安徽工程大学 一种基于运动补偿学习的4d-cbct成像方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017180097A1 (fr) * 2016-04-12 2017-10-19 Siemens Aktiengesellschaft Alignement déformable d'entrées peropératoire et préopératoire à l'aide de modèles de mélange génératifs et d'une déformation biomécanique
CN111179409A (zh) * 2019-04-23 2020-05-19 艾瑞迈迪科技石家庄有限公司 一种呼吸运动建模方法、装置和系统
CN110613519A (zh) * 2019-09-20 2019-12-27 真健康（北京）医疗科技有限公司 动态配准定位装置及方法
CN112154483A (zh) * 2020-07-15 2020-12-29 北京肿瘤医院(北京大学肿瘤医院) 一种利用光学体表运动信号合成实时图像的方法及系统
CN112435307A (zh) * 2020-11-26 2021-03-02 浙江大学 一种深度神经网络辅助的四维锥束ct影像重建方法
CN113574565A (zh) * 2021-02-07 2021-10-29 北京肿瘤医院(北京大学肿瘤医院) 一种利用光学体表运动信号合成实时图像的系统
CN114820855A (zh) * 2022-04-28 2022-07-29 浙江大学滨江研究院 一种基于患者4d-ct的肺部呼吸过程图像重构方法及装置

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN119112107A (zh) * 2024-09-09 2024-12-13 合肥工业大学 轻量化的非接触式多参数生理检测方法及系统

Also Published As

Publication number Publication date
CN115996668B (zh) 2024-09-17
CN115996668A (zh) 2023-04-21

Similar Documents

Publication Publication Date Title
CN115996668B (zh) 一种基于四维体表呼吸信号的四维成像方法和系统
US9254112B2 (en) Respiratory interval-based correlation and processing of dynamic imaging data
CN110619635B (zh) 基于深度学习的肝细胞癌磁共振图像分割系统和方法
CN110163809A (zh) 基于U-net生成对抗网络DSA成像方法及装置
US8655040B2 (en) Integrated image registration and motion estimation for medical imaging applications
KR20140096919A (ko) 의료 영상 정합 방법 및 장치
CN119887773B (zh) 一种基于多模态图像融合的医疗图像识别处理系统及方法
CN111275669A (zh) 一种先验信息引导的四维锥束ct图像重建算法
Cheng et al. Airway segmentation and measurement in CT images
CN114565711B (zh) 基于深度学习的心脏图像重建方法及系统
CN113574565A (zh) 一种利用光学体表运动信号合成实时图像的系统
CN116128838A (zh) 基于双目视觉的呼吸检测系统、4d-ct图像重建系统和方法
CN106952285B (zh) 基于先验统计运动模型及自适应配准的肺部运动估计方法
Fischer et al. An MR-based model for cardio-respiratory motion compensation of overlays in X-ray fluoroscopy
Fu et al. Four-dimensional wide-field ultrasound reconstruction system with sparse respiratory signal matching
CN107292351B (zh) 一种结节的匹配方法及装置
Xue et al. Effective lung ventilation estimation based on 4D CT image registration and supervoxels
CN120088224B (zh) 一种基于cbct和多尺度小波注意力网络的牙齿图像处理系统
CN110279429A (zh) 四维超声重建方法及装置
CN120451203A (zh) 肺呼吸运动建模方法及相关装置
Zhou et al. Automated estimation of the upper surface of the diaphragm in 3-D CT images
Roy et al. MDL-IWS: multi-view deep learning with iterative watershed for pulmonary fissure segmentation
EP4082444B1 (fr) Sélection automatique de cadre de construction de modèle 3d
TWI773491B (zh) 肺功能鑑別方法
CN110858412A (zh) 基于图像配准的心脏冠脉cta模型建立方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22966108

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 22966108

Country of ref document: EP

Kind code of ref document: A1