A tracheal staging method, system, and computer-readable storage medium

Abstract

This specification provides a method and system for tracheal classification. The method includes: determining an initial tracheal classification mask based on a trained tracheal classification model and medical images; determining a tracheal centerline based on the medical images, and determining a tracheal centerline with an initial classification label based on the tracheal centerline and the initial tracheal classification mask; correcting the initial classification label using a preset method to obtain a target classification label for the tracheal centerline; and reconstructing the tracheal centerline with the target classification label to obtain a target tracheal classification mask.

Description

Technical Field

[0001] This specification relates to the field of image processing technology, and in particular to a tracheal classification method and system. Background Technology

[0002] In the respiratory system, the trachea is a core functional organ. Damage or lesions related to the trachea can seriously affect a patient's bodily functions. Examples include lung cancer, bronchial embolism, and pulmonary nodules. In the clinical diagnosis and treatment planning of lung diseases, the grading of tracheal anatomy is of paramount importance for lesion localization, lesion impact assessment, segmental and vascular division of the lungs, preoperative planning, and intraoperative navigation.

[0003] Therefore, it is hoped that a tracheal classification method and system can be proposed to improve the accuracy and overall orderliness of tracheal classification results. Summary of the Invention

[0004] This specification provides a method for tracheal classification. The method includes: determining an initial tracheal classification mask based on a trained tracheal classification model and medical images; determining a tracheal centerline based on the medical images, and determining a tracheal centerline with an initial classification label based on the tracheal centerline and the initial tracheal classification mask; correcting the initial classification label using a preset method to obtain a target classification label for the tracheal centerline; and reconstructing the tracheal centerline with the target classification label to obtain a target tracheal classification mask.

[0005] Another aspect of this specification provides a tracheal grading system. The system includes: a preliminary grading module for determining an initial tracheal grading mask based on a trained tracheal grading model and medical images; a centerline correction module for determining a tracheal centerline based on the medical images, and determining a tracheal centerline with an initial grading label based on the tracheal centerline and the initial tracheal grading mask; and correcting the initial grading label using a preset method to obtain a target grading label for the tracheal centerline; and a grading mask reconstruction module for reconstructing a target tracheal grading mask based on the tracheal centerline with the target grading label.

[0006] Another aspect of this specification provides a computer-readable storage medium that stores computer instructions, which, when read by a computer, execute the methods described above. Attached Figure Description

[0007] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein:

[0008] Figure 1 This is a schematic diagram illustrating an exemplary tracheotomy system application scenario according to some embodiments of this specification;

[0009] Figure 2 This is a schematic diagram of the modules of an exemplary tracheostratification system according to some embodiments of this specification;

[0010] Figure 3 This is a schematic flowchart of an exemplary tracheal triage method according to some embodiments of this specification;

[0011] Figure 4 This is a schematic diagram of an exemplary process for obtaining a tracheal centerline with a grading label, according to some embodiments of this specification.

[0012] Figure 5 This is a schematic flowchart illustrating the exemplary process of inter-segment correction of the tracheal centerline grading labels according to some embodiments of this specification;

[0013] Figure 6 This is a schematic diagram of an exemplary tracheal triage method according to some embodiments of this specification;

[0014] Figure 7 This is a schematic diagram of an exemplary target tracheal classification mask according to some embodiments of this specification;

[0015] Figure 8 This is a schematic diagram illustrating an exemplary method for obtaining a target tracheal classification mask according to some embodiments of this specification. Detailed Implementation

[0016] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this specification. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.

[0017] It should be understood that the terms “system,” “device,” “unit,” and / or “module” used herein are one way to distinguish different components, elements, parts, sections, or assemblies at different levels. However, if other terms can achieve the same purpose, they may be replaced by other expressions.

[0018] As indicated in this specification and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of expressly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0019] Flowcharts are used in this specification to illustrate the operations performed by the system according to embodiments of this specification. The related descriptions are provided to aid in a better understanding of the medical imaging methods and / or systems. It should be understood that preceding or subsequent operations are not necessarily performed precisely in sequence. Instead, steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0020] Figure 1 This is a schematic diagram illustrating an exemplary tracheotomy system application scenario according to some embodiments of this specification.

[0021] like Figure 1 As shown, the tracheal triage system 100 may include an imaging device 110, a processing device 120, one or more terminals 130, a storage device 140, and a network 150. The components in the tracheal triage system 100 may be connected in one or more of various ways. This is merely an example. Figure 1 As shown, imaging device 110 can be connected to processing device 120 via network 150. As another example, imaging device 110 can be directly connected to processing device 120, as indicated by the dotted double-headed arrows in the figure. As yet another example, storage device 140 can be directly connected to processing device 120. Figure 1 (Not shown in the diagram) or connected via network 150. As another example, one or more terminals 130 may be directly connected to processing device 120 (as shown by the dashed double-headed arrows connecting terminal 130 and processing device 120) or connected via network 150.

[0022] Imaging device 110 can be used to scan a target object within a detection area to obtain scan data (e.g., medical images, etc.) of the target object. In some embodiments, the target object may include biological and / or non-biological objects. For example, the target object may include specific parts of the body, such as the head, chest, abdomen, etc., or combinations thereof. As another example, the target object may be a human-made component of living or non-living organic and / or inorganic matter. In some embodiments, the scan data related to the target object may include projection data of the target object, one or more scan images, etc.

[0023] In some embodiments, imaging device 110 may include a non-invasive bioimaging device for disease diagnosis or research purposes. For example, imaging device 110 may include a single-modal scanner and / or a multimodal scanner. A single-modal scanner may include, for example, an ultrasound scanner, an X-ray scanner, a computed tomography (CT) scanner, a magnetic resonance imaging (MRI) scanner, an ultrasound examination device, a positron emission tomography (PET) scanner, an optical coherence tomography (OCT) scanner, an ultrasound (US) scanner, an intravascular ultrasound (IVUS) scanner, a near-infrared spectroscopy (NIRS) scanner, a far-infrared (FIR) scanner, etc. A multimodal scanner may include, for example, an X-ray imaging-magnetic resonance imaging (X-MRI) scanner, a positron emission tomography-X-ray imaging (PET-X-ray) scanner, a single-photon emission computed tomography-magnetic resonance imaging (SPECT-MRI) scanner, a positron emission tomography-computed tomography (PET-CT) scanner, a digital subtraction angiography-magnetic resonance imaging (DSA-MRI) scanner, etc. The scanners described above are for illustrative purposes only and are not intended to limit the scope of this specification. As used herein, the term "imaging modality" or "modality" broadly refers to imaging methods or techniques for collecting, generating, processing, and / or analyzing imaging information of a target object.

[0024] In some embodiments, data acquired by imaging device 110 (e.g., medical images, etc.) may be transmitted to processing device 120 for further analysis. Additionally or alternatively, data acquired by imaging device 110 may be sent to terminal device (e.g., terminal 130) for display and / or to storage device (e.g., storage device 140) for storage.

[0025] The processing device 120 can process data and / or information acquired and / or extracted from the imaging device 110, the terminal 130, the storage device 140, and / or other storage devices. For example, the processing device 120 can acquire medical images from the imaging device 110, determine an initial tracheal classification mask based on the medical images, determine a tracheal centerline with an initial classification label based on the tracheal centerline and the initial tracheal classification mask, correct the initial classification label based on a preset method to obtain a target classification label for the tracheal centerline, and obtain a target tracheal classification mask based on the tracheal centerline with the target classification label.

[0026] In some embodiments, the processing device 120 may be a single server or a group of servers. The server group may be centralized or distributed. In some embodiments, the processing device 120 may be local or remote. In some embodiments, the processing device 120 may be implemented on a cloud platform. By way of example only, the cloud platform may include private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, internal cloud, multi-tiered cloud, etc., or any combination thereof.

[0027] In some embodiments, the processing device 120 may be implemented on a computing device. In some embodiments, the processing device 120 may be implemented on a terminal (e.g., terminal 130). In some embodiments, the processing device 120 may be implemented on an imaging device (e.g., imaging device 110). For example, the processing device 120 may be integrated into terminal 130 and / or imaging device 110.

[0028] Terminal 130 can be connected to imaging device 110 and / or processing device 120 for inputting / outputting information and / or data. For example, a user can interact with imaging device 110 through terminal 130 to control one or more components of imaging device 110 (e.g., inputting patient information, etc.). As another example, imaging device 110 can output generated medical images and / or tracheal classification masks to terminal 130 for display to the user.

[0029] In some embodiments, terminal 130 may include mobile device 131, tablet computer 132, laptop computer 133, etc., or any combination thereof. In some embodiments, mobile device 131 may include smart home device, wearable device, smart mobile device, virtual reality device, augmented reality device, etc., or any combination thereof. In some embodiments, one or more terminals 130 may remotely operate imaging device 110. In some embodiments, terminal 130 may operate imaging device 110 via wireless connection. In some embodiments, one or more terminals 130 may be part of processing device 120. In some embodiments, terminal 130 may be omitted.

[0030] Storage device 140 may store data and / or instructions. In some embodiments, storage device 140 may store data acquired from terminal 130 and / or processing device 120. For example, storage device 140 may store medical images, initial tracheal grading masks, tracheal centerlines with initial grading labels, target tracheal grading masks, etc. In some embodiments, storage device 140 may store data and / or instructions, and processing device 120 may execute or use the data and / or instructions to perform the exemplary methods described herein.

[0031] In some embodiments, storage device 140 may include mass storage devices, removable storage devices, volatile read-write memory, read-only memory (ROM), and any combination thereof. Exemplary mass storage devices may include disks, optical disks, solid-state drives, etc. Exemplary removable storage devices may include flash drives, floppy disks, optical disks, memory cards, compact disks, magnetic tapes, etc. Exemplary volatile read-write memory may include random access memory (RAM). In some embodiments, storage device 140 may be implemented on a cloud platform. In some embodiments, storage device 140 may be part of processing device 120.

[0032] Network 150 may include any suitable network that can facilitate the exchange of information and / or data within the tracheostratification system 100. In some embodiments, one or more components of the tracheostratification system 100 (e.g., imaging device 110, one or more terminals 130, processing device 120, or storage device 140) may communicate with one or more other components of the tracheostratification system 100 to transmit information and / or data. In some embodiments, network 150 may be any type of wired or wireless network or a combination thereof. For example, network 150 may be and / or include public networks (e.g., the Internet), private networks (e.g., local area networks (LANs), wide area networks (WANs), etc.), wired networks (e.g., Ethernet), wireless networks (e.g., 802.11 networks, Wi-Fi networks, etc.), cellular networks (e.g., LTE networks), Frame Relay networks, virtual private networks (“VPNs”), satellite networks, telephone networks, routers, hubs, switches, server computers, and / or any combination thereof. In some embodiments, network 150 may include one or more network access points.

[0033] It should be noted that the above description of the tracheal triage system 100 is for illustrative purposes only and is not intended to limit the scope of this specification. Various modifications and variations can be made based on this specification by those skilled in the art. However, these changes and modifications do not depart from the scope of this specification. For example, the imaging device 110, processing device 120, and terminal 130 may share a single storage device 140, or they may each have their own storage devices.

[0034] Anatomically, the trachea is characterized by complex and varied morphological features such as shape, length, and radius. If divided into zones based on tracheal diameter and lung function, the tracheal branching levels can reach more than 10 grades (from the main trachea and main bronchi to the terminal bronchioles). Manual tracheal grading is not only highly demanding on the user's imaging experience but also extremely time-consuming, with the time increasing with finer grading. In some embodiments, image processing-based methods can be used for tracheal tree grading. However, such methods require complex preprocessing and feature extraction operations and heavily rely on prior image knowledge and statistical features, resulting in low tracheal grading efficiency and robustness, and are prone to misclassification and insufficient grading granularity. In some embodiments, the high generalization and learnability of deep learning can be utilized to extract high-level features from medical images through neural network models, thereby achieving tracheal grading. However, existing deep learning-based methods can only classify larger bronchi at the subsegment level and above (i.e., tracheal grading results of level 4 or above), lacking effective processing for finer, lower-level tracheal branches. Moreover, these methods typically rely on raw medical images and tracheal segmentation masks, lacking contextual guidance, resulting in low accuracy and uneven boundary delineation between different levels of tracheal branches in the obtained tracheal grading results.

[0035] This specification provides a method and system for tracheal classification based on medical images. Based on the tracheal centerline and an initial tracheal classification mask obtained using a trained tracheal classification model, a tracheal centerline with an initial classification label is determined. The initial classification label of the tracheal centerline is then corrected to obtain a tracheal centerline with a target classification label. Reconstruction is then performed based on this target classification centerline to obtain the target tracheal classification mask. This method not only utilizes the high robustness and generalization of deep learning technology but also incorporates additional contextual information such as tracheal radius variations and relative features with lung lobes, resulting in a relatively accurate and richly layered initial tracheal classification mask. Furthermore, by combining prior knowledge of the tracheal tree anatomy and a centerline tracing algorithm to progressively optimize the initial tracheal classification mask, a more accurate and ordered target tracheal classification mask can be obtained.

[0036] Figure 2 This is a schematic diagram of the modules of an exemplary tracheostratification system according to some embodiments of this specification.

[0037] like Figure 2 As shown, in some embodiments, the tracheal triage system 200 may include a preliminary triage module 210, a centerline correction module 220, and a triage mask reconstruction module 230. In some embodiments, the tracheal triage system 200 may be integrated into the imaging device 110 or the processing device 120.

[0038] The preliminary grading module 210 can be used to determine an initial tracheal grading mask. In some embodiments, the preliminary grading module 210 can determine the initial tracheal grading mask based on a trained tracheal grading model and medical images. In some embodiments, the preliminary grading module 210 can determine a tracheal segmentation mask and a lung lobe segmentation mask based on medical images, and determine a distance map from the trachea to the tracheal centerline in the tracheal segmentation mask. In some embodiments, the preliminary grading module 210 can input the tracheal segmentation mask, the lung lobe segmentation mask, and the distance map from the trachea to the tracheal centerline into a trained tracheal grading model to obtain the output initial tracheal grading mask.

[0039] The centerline correction module 220 can be used to obtain a tracheal centerline with a target grading label. In some embodiments, the centerline correction module 220 can determine a tracheal centerline with an initial grading label based on the tracheal centerline and an initial tracheal grading mask. In some embodiments, the centerline correction module 220 can traverse the tracheal centerlines with initial grading labels and correct the initial grading labels of the tracheal centerlines based on a preset method to obtain a tracheal centerline with a target grading label.

[0040] In some embodiments, the centerline correction module 220 can be used to segment the tracheal centerline according to the bifurcation point to obtain a tree diagram composed of multiple centerline segments. Each centerline segment corresponds to a node in the tree diagram. In some embodiments, the centerline correction module 220 can remove centerline segments in the tree diagram that do not meet preset conditions to obtain a directed acyclic tree diagram.

[0041] In some embodiments, the centerline correction module 220 can be used to unify the initial grading labels contained in the centerline segments represented by each node in the tree diagram to obtain intermediate grading labels for the tracheal centerline. Further, the centerline correction module 220 can correct the intermediate grading labels of each node in the tree diagram based on the intermediate grading labels of neighboring nodes belonging to the same parent node, thereby obtaining the target grading label for the tracheal centerline.

[0042] In some embodiments, the centerline correction module 220 can traverse from parent node to child node based on a tree diagram. For each node in the tree diagram, it counts all the initial grade labels contained in the centerline segment represented by the node, and determines the initial grade label with the highest statistical value as the intermediate grade label corresponding to the current node.

[0043] Furthermore, in some embodiments, the centerline correction module 220 can traverse from parent node to child node based on a tree diagram, and determine whether the intermediate hierarchical label of the current node needs to be corrected according to preset rules. If correction is required, the centerline correction module 220 can count the intermediate hierarchical labels of neighboring nodes that belong to the same parent node as the current node, and determine the intermediate hierarchical label with the highest statistical value as a candidate label. Further, the centerline correction module 220 can determine whether the candidate label meets the preset rules; if so, the candidate label is determined as the target hierarchical label of the current node; otherwise, the target hierarchical label of the current node is determined based on the intermediate hierarchical labels of the parent node of the current node.

[0044] The hierarchical mask reconstruction module 230 can be used to reconstruct the tracheal centerline based on the target hierarchical label to obtain the target tracheal hierarchical mask.

[0045] In some embodiments, the hierarchical mask reconstruction module 230 can divide the tracheal centerline according to the target hierarchical label to obtain the centerline segment corresponding to each target hierarchical label. Further, for each tracheal region in the tracheal segmentation mask, the hierarchical mask reconstruction module 230 can determine the centerline segment closest to the tracheal region and determine the target hierarchical label of that centerline segment as the hierarchical label corresponding to the current tracheal region, thereby obtaining the target tracheal hierarchical mask.

[0046] It should be noted that the above description of the tracheal triage system 200 and its modules is for ease of description only and should not be construed as limiting this specification to the scope of the illustrated embodiments. It is understood that those skilled in the art, after understanding the principles of the system, may arbitrarily combine the various modules or construct subsystems connected to other modules without departing from these principles. However, such modifications and changes are still within the scope of this specification.

[0047] Figure 3 This is a schematic flowchart of an exemplary tracheal triage method according to some embodiments of this specification.

[0048] In some embodiments, method 300 may be executed by imaging device 110 or processing device 120. For example, method 300 may be stored in a storage device (such as storage device 140) in the form of a program or instructions, which may be implemented when imaging device 110 or processing device 120 executes the program or instructions. In some embodiments, method 300 may be executed by tracheal staging system 200.

[0049] Step 310: Based on the trained tracheal grading model and medical images, determine the initial tracheal grading mask. In some embodiments, step 310 may be performed by the processing device 120 or the preliminary grading module 210.

[0050] In some embodiments, medical images may include CT images, MR images, or PET images, etc. In some embodiments, medical images may include 2D images, 3D images, or 4D images, etc. In some embodiments, medical images can be obtained by scanning with an imaging device. For example, a target object within a detection area can be scanned using imaging device 110 to obtain a medical image of the target object. In some embodiments, medical images can be acquired from terminal 130 or storage device 140.

[0051] A tracheal grading model can be used to grade tracheal trees in medical images, starting with the main trachea, based on tracheal diameter and / or branching patterns to obtain an initial tracheal grading mask. In some embodiments, the smaller the tracheal diameter, the lower the corresponding grade. For example, the main trachea generally has the largest diameter, corresponding to grade 1, while the diameter of the bronchi decreases as the grade decreases, for example, grades 2, 3, 4, 5, etc. In some embodiments, each grade may include one or more branch tracheas. For example, there is only one grade 1 main trachea, and the main trachea branches into two or three grade 2 tracheas. Each grade 2 trachea further branches into three or more grade 3 tracheas. In some embodiments, the number of branch tracheas within the same grade can be the same or different. For example, two grade 2 tracheas can further branch into three and four bronchi, respectively.

[0052] In some embodiments, the tracheal classification model may include any type of convolutional neural network, such as VB-Net, U-Net, fully convolutional networks (FCN), feature pyramid networks (FPN), etc., and this specification does not limit this. In some embodiments, the input of the tracheal classification model may be a three-channel input related to medical images, and the output may be a tracheal classification mask corresponding to the medical images.

[0053] In some embodiments, the tracheal segmentation mask, lung lobe segmentation mask, and tracheal-to-tracheal-centerline distance map corresponding to the medical image can be used as three-channel inputs to a trained tracheal grading model to obtain the initial tracheal grading mask output. The lung lobe segmentation mask reflects the coarse tracheal grading, while the tracheal-to-tracheal-centerline distance map reflects the tracheal diameter.

[0054] In some embodiments, tracheal segmentation masks and lung lobe segmentation masks can be determined based on medical images, such as binary tracheal segmentation masks and binary lung lobe segmentation masks. In some embodiments, tracheal segmentation masks and lung lobe segmentation masks can be determined using trained tracheal segmentation models and lung lobe segmentation models, respectively. For example, medical images can be input into trained tracheal segmentation models and lung lobe segmentation models, respectively, to obtain tracheal segmentation masks and lung lobe segmentation masks.

[0055] In some embodiments, different training samples can be used to train and obtain tracheal segmentation models and lung lobe segmentation models respectively. For example, a large number of chest CT images can be acquired as the first training sample, and the labeled tracheal tree can be used as the gold standard to train the initial neural network model to obtain a trained tracheal segmentation model. Similarly, a large number of chest CT images can be acquired as the second training sample, and the labeled lung lobes can be used as the gold standard to train the initial neural network model to obtain a trained lung lobe segmentation model. In some embodiments, the tracheal segmentation model and / or the lung lobe segmentation model can include any type of neural network model, and this specification does not limit this.

[0056] In some embodiments, the medical images can be preprocessed, and the preprocessed medical images can be input into the trained trachea segmentation model and lung lobe segmentation model respectively to obtain trachea segmentation masks and lung lobe segmentation masks. For example, preprocessing may include, but is not limited to, normalization of CT values ​​by window level and / or window width (e.g., normalization using [-625, 425] as mean and standard deviation), noise reduction, and other operations.

[0057] In some embodiments, the trachea centerline can be determined based on a trachea segmentation mask, thereby determining a distance map (also known as a first distance map) from the trachea to the trachea centerline in the trachea segmentation mask. For example, the trachea segmentation binary mask can be skeletonized to extract the trachea centerline, and then the Euclidean distance from the foreground voxels (i.e., the trachea region) to the trachea centerline in the trachea segmentation binary mask can be calculated as the distance map from the trachea to the trachea centerline. In some embodiments, the trachea centerline can be extracted in any reasonable and feasible manner, such as neighborhood analysis thinning algorithms, etc., and this specification does not limit this.

[0058] In the distance map from the trachea to the tracheal centerline, the closer a voxel is to the tracheal surface, the larger its corresponding distance value. Furthermore, the distance value of a voxel at the surface is equal to the radius of the trachea at that location. Therefore, the distance map can provide information about the radius of the trachea to characterize the variation in the trachea's thickness (e.g., a larger distance value indicates a thicker trachea).

[0059] In some embodiments, a three-channel training sample (i.e., the gold standard for tracheal segmentation, the distance map from the gold standard for tracheal segmentation to the tracheal centerline, and the gold standard for lobar segmentation) and the corresponding gold standard for tracheal grading can be used to iteratively train an initial neural network model (e.g., a VB-Net model) until convergence, thereby obtaining a trained tracheal grading model. In addition to the tracheal segmentation mask, this tracheal grading model utilizes an additional distance map from the trachea to the tracheal centerline and a lobar segmentation mask, enabling it to acquire richer contextual information to guide tracheal grading. This not only provides a stronger ability to capture features of higher-level bronchi (e.g., lobar-level 2 bronchi), but also obtains a relatively reliable and richer initial tracheal grading mask on a global scale (e.g., capable of grading small segmental bronchi up to level 8).

[0060] Step 320: Based on the tracheal centerline and the initial tracheal classification mask, determine the tracheal centerline with the initial classification label. In some embodiments, step 320 may be performed by the processing device 120 or the centerline correction module 220.

[0061] The grading label can reflect the grading level of the tracheal centerline and the corresponding trachea. For example, the tracheal centerline with a grading label of 1 corresponds to a bronchus with a grading level of 1.

[0062] In some embodiments, the intersection of the tracheal centerline and the initial tracheal classification mask can be obtained to obtain the initial classification label of the tracheal centerline. For example, the intersection of the tracheal centerline mask and the initial tracheal classification mask can be performed, and for the intersection portion, the initial classification label of the corresponding tracheal centerline can be determined based on the classification level of the initial tracheal classification mask.

[0063] Step 330: Correct the initial grading label of the tracheal centerline based on a preset method to obtain the target grading label of the tracheal centerline. In some embodiments, step 330 may be performed by the processing device 120 or the centerline correction module 220.

[0064] In some embodiments, correcting the initial grading labels of the tracheal centerline based on a preset method may include: unifying the initial grading labels contained in the centerline segments represented by each node of the tracheal centerline to obtain intermediate grading labels of the tracheal centerline; and performing inter-segment correction on the intermediate grading labels of each centerline segment based on the intermediate grading labels of neighboring segments belonging to the same parent node as the current node. In some embodiments, the initial grading labels contained in each centerline segment of the tracheal centerline may be unified based on a tree diagram, and / or inter-segment correction may be performed on the intermediate grading labels of each centerline segment. More related content can be found at [link to relevant documentation]. Figures 4-5 The details and related descriptions will not be repeated here.

[0065] Step 340: Reconstruct the tracheal centerline based on the target tracheal grading label to obtain the target tracheal grading mask. In some embodiments, step 340 may be performed by the processing device 120 or the grading mask reconstruction module 230.

[0066] In some embodiments, for each tracheal region in the tracheal segmentation mask, the centerline segment closest to that tracheal region can be determined, and the target classification label of that centerline segment can be determined as the classification label of the current tracheal region to obtain the target tracheal classification mask.

[0067] In some embodiments, for each tracheal region in the tracheal segmentation mask, the nearest centerline segment can be determined based on the distance map (also referred to as the second distance map) of that tracheal region. In some embodiments, the tracheal centerline can be divided according to the corrected target grading labels to obtain the centerline segment corresponding to each target grading label, and the Euclidean distance between each centerline segment and the tracheal region in the tracheal segmentation mask can be calculated to obtain the distance map from the tracheal region to each centerline segment. For example, after obtaining the centerline segment corresponding to each target grading label, for each tracheal region in the tracheal segmentation binary mask, the shortest distance from all voxels of the tracheal region to all centerline segments contained in each target grading label can be calculated. Based on this, the shortest distance from each voxel of the tracheal region to the centerline segment contained in each target grading label is calculated one by one to obtain the distance map from the tracheal region to the centerline segment of each target grading label. That is, the distance map from the tracheal region to the centerline segment of each target grading label can reflect the distance between the corresponding trachea and each target grading label.

[0068] As an example, if the trachea is divided into 10 levels from the pharynx to the end, then the corrected tracheal centerline will have 10 different target grading labels, such as label 1, 2, 3, ..., 10. The processing device 120 can divide the tracheal centerline with the target grading labels according to their target grading labels, thereby obtaining the centerline segments corresponding to each target grading label, that is, the centerline segments contained in each target grading label. For example... Figure 8 As shown, the upper left corner is an illustration of the centerline with grading labels. The small gray squares represent the tracheal centerline, and the numbers above them represent the grading labels. Then, for each target grading label, the processing device 120 can calculate the shortest distance from each voxel corresponding to the tracheal region to the centerline points on all centerline segments contained in the target grading label 1, obtaining a distance map from the tracheal region to the centerline segments of the target grading label 1. For example... Figure 8As shown in (a), the numbers in the figure represent the distances from the tracheal region to each centerline segment; the shortest distance from each voxel in the tracheal region to the centerline point on all centerline segments included in the target grading label 2 is calculated to obtain the distance map from the tracheal region to the centerline segments of the target grading label 2, for example. Figure 8 As shown in (b), ..., this process is repeated until all 10 target classification labels and tracheal regions have been calculated, thus obtaining a distance map for the 10 channels. Further, the processing device 120 can, based on the aforementioned distance map, compare the distances of each voxel corresponding to the tracheal region to the centerline segments contained in different target classification labels for each voxel, and determine the target classification label of the centerline segment with the smallest distance value as the current tracheal classification, obtaining a target tracheal classification mask, for example... Figure 8 As shown in (c), the closer a voxel in the tracheal region is to the tracheal centerline, the shorter its distance to the tracheal centerline. For example, the voxel of the centerline segment body is 0 units away from the centerline segment.

[0069] It is understood that the above method of determining the centerline segment closest to the tracheal region based on the distance map (second distance map) of the tracheal region, and determining the target classification label of the centerline segment as the classification label of the current tracheal region, is merely an example. In some embodiments, any other reasonable and feasible method can be used to determine the classification label of the tracheal region, and this specification does not limit this. For example, a channel-level argmin operation can be used to determine the centerline segment closest to each foreground voxel in the tracheal segmentation binary mask, and the target classification label of the centerline segment can be determined as the classification of the current voxel. Alternatively, a method of progressively expanding the centerline segments contained in different target classification labels can be used to determine the centerline segment closest to each foreground voxel in the tracheal segmentation binary mask, and the target classification label of the centerline segment can be determined as the classification of the current voxel.

[0070] In some embodiments, the obtained tracheal classification mask may include five or more classifications.

[0071] Figure 7 This is a schematic diagram of an exemplary target tracheal grading mask according to some embodiments of this specification. As shown in the figure, the main trachea is a level 1 trachea P1, which divides into two level 2 tracheas P2 at the branching point. Each of the two level 2 tracheas P2 is further divided into two or more level 3 tracheas P3, ..., until the terminal level P8 trachea, which reaches the level 8 bronchial grading within a small lung segment. It is understood that... Figure 7 The label Pn for the bronchial classification is for illustrative purposes only; other bronchi in the diagram also have their corresponding classifications.

[0072] In some embodiments, in clinical settings, different levels of trachea can be displayed in different ways. For example, level 1 trachea is red, level 2 trachea is blue, level 3 trachea is pink, level 4 trachea is ginger yellow, level 5 trachea is green, and level 6 trachea is orange, etc.

[0073] It should be noted that the above description of method 300 is for illustrative purposes only and does not limit the scope of this specification. Those skilled in the art can make various modifications and changes to method 300 under the guidance of this specification. However, these modifications and changes remain within the scope of this specification.

[0074] Figure 4 This is a schematic diagram of an exemplary process for obtaining a tracheal centerline with a grading label, as shown in some embodiments of this specification.

[0075] like Figure 4 As shown, in some embodiments, based on a tree diagram, the initial classification labels contained in each centerline segment node of the tracheal centerline can be uniformly corrected to obtain intermediate classification labels of the tracheal centerline. Furthermore, the intermediate classification label of the current node can be corrected based on the intermediate classification labels of neighboring segments belonging to the same parent node as the current node to obtain the target classification label of the tracheal centerline. In some embodiments, method 400 can be performed by tracheal classification system 100 (e.g., imaging device 110 or processing device 120 in system 100) or tracheal classification system 200 (e.g., centerline correction module 220).

[0076] Step 420: Obtain the tracheal centerline with the initial grading label.

[0077] In some embodiments, a tracheal centerline with an initial classification label can be obtained based on the initial tracheal classification mask 411 and the tracheal centerline 413. More information about the initial tracheal classification mask 411, the tracheal centerline 413, and obtaining the tracheal centerline with the initial classification label can be found in [link to relevant documentation]. Figure 3 The relevant descriptions (e.g., steps 310-320) will not be repeated here.

[0078] Step 430: Divide the tracheal centerline into segments according to the bifurcation points to obtain a tree diagram composed of multiple centerline segments.

[0079] In some embodiments, the tracheal centerline can be segmented according to bifurcation points to obtain a tree diagram composed of multiple centerline segments. Each centerline segment corresponds to a node in the tree diagram. The tree diagram is represented by multiple nodes and multiple edges; each node represents a centerline segment, and the edges between nodes represent the connections between centerline segments. For example, the tracheal centerline can be traced sequentially from the main tracheal inlet (e.g., the pharynx) towards the end. When a bifurcation point is encountered, it is divided into multiple segments according to the number of branches at the bifurcation point. This process continues downwards until the end of the tracheal centerline, thus obtaining a tree diagram composed of multiple centerline segments. Each centerline segment corresponds to a node in the tree diagram, and the centerline segments between centerline segments correspond to the edges between nodes. One or more branch centerline segments within the same centerline segment (e.g., multiple centerline segments branching off from the end of the centerline segment) constitute a node unit.

[0080] It's important to note that the initial classification label for the tracheal central line refers to the classification label for each point on the tracheal central line tree. For example, if the tracheal central line tree contains 100 central line points, then each central line point has an initial classification label, for a total of 100 initial classification labels. By segmenting the tracheal central line according to its bifurcation points, the tracheal central line can be divided into a tree diagram composed of multiple central line segments. For example, if the tracheal central line tree contains 100 central line points, after segmentation according to bifurcation points, the tracheal central line may be divided into 10 segments, each containing the same or different number of central line segments.

[0081] In some embodiments, centerline segments that do not meet preset conditions in the tree diagram can be removed to obtain a directed acyclic tree diagram. The preset conditions can reflect the shape of centerline segments that do not meet the requirements for tracheal classification. In some embodiments, the preset conditions can be reasonably set according to actual conditions, and this specification does not impose any restrictions on this. For example, redundant centerline segments with fewer than 3 centerline points and loop-shaped centerline segments can be removed during the traversal process, thereby making the obtained tree diagram of tracheal centerlines conform to the requirements of a directed acyclic graph. A topological graph consisting of nodes and edges is defined as a directed acyclic graph if the edges are directed (e.g., from node A to node B) and there are no loops in the entire graph, i.e., there is only one path between any two points and no loops. Through the removal operation, invalid centerline segments caused by tracheal adhesions, protrusions, etc., can be removed, thereby avoiding interference from redundant centerline segments or loop-shaped loops.

[0082] Step 440: Based on the tree diagram, unify the initial grading labels contained in each segment of the tracheal centerline to obtain the intermediate grading labels of the tracheal centerline.

[0083] The hierarchical labels included in a centerline segment are the hierarchical labels of the centerline points corresponding to all branches of that segment node. For example, if the tracheal centerline is divided into 10 segments, the initial hierarchical labels included in the first segment's centerline segment can refer to the initial hierarchical labels of all centerline points included in that segment.

[0084] In some embodiments, a tree diagram can be used as a basis, traversing from parent nodes to child nodes, to unify the initial hierarchical labels contained in the centerline segments represented by each node in the tree diagram within each segment, thereby obtaining intermediate hierarchical labels for the tracheal centerline. Intra-segment unification involves statistically voting on the hierarchical labels of the centerline points contained in each centerline segment node, and using the hierarchical label with the highest number of votes as the hierarchical label for the current centerline segment node.

[0085] In some embodiments, a tree diagram can be used, traversing from parent nodes to child nodes. For each node in the tree diagram, all initial hierarchical labels contained in the centerline segment represented by that node are counted, and the initial hierarchical label with the highest statistical value is determined as the intermediate hierarchical label corresponding to the current node. For example, if the current centerline segment contains 3 centerline points, with point 1 having an initial hierarchical label of 3, point 2 having an initial hierarchical label of 4, and point 3 having an initial hierarchical label of 3, then the number of hierarchical labels 3 is two, and the number of hierarchical labels 4 is one. Therefore, the hierarchical label 3 with the highest statistical value can be used as the unified hierarchical label within the current centerline segment, i.e., the intermediate hierarchical label is 3.

[0086] Step 450: Based on the tree diagram, correct the intermediate hierarchical labels of each centerline segment based on the intermediate hierarchical labels of its neighboring nodes.

[0087] In some embodiments, after unifying the grading labels of the tracheal centerline within a segment, the grading label of each node can be corrected based on the tracheal centerline tree diagram, traversing from parent nodes to child nodes, and the grading label of each node is corrected based on the grading labels of its neighboring segment nodes. Correcting the grading label of the current node based on the grading labels of neighboring segment nodes that share the same parent node as the current node can also be called inter-segment correction.

[0088] Specifically, please see Figure 5 In some embodiments, after traversing from the parent node to the child node, step 453 can be executed first: determine whether the intermediate hierarchical labels of the current node need to be corrected. If so, proceed to steps 454-455 to correct the intermediate hierarchical labels of the current node; otherwise, proceed to step 451: determine whether the traversal has ended. If the traversal has ended, proceed to step 460; otherwise, proceed to step 452: continue traversing until the traversal ends.

[0089] Step 453: Determine whether the intermediate hierarchical labels of the current node need to be corrected.

[0090] In some embodiments, it can be determined whether the intermediate classification label (i.e., the classification label obtained after uniform correction within the segment) of the current node (i.e., the centerline segment) needs to be corrected according to preset rules. In some embodiments, the preset rules may include: the classification label value of the current node is greater than the classification label value of its parent node (e.g., a level 3 bronchus must not be after a level 4 bronchus), and / or, the classification label value of the current node must not exceed the classification label value of its parent node by more than two levels (e.g., a level 6 bronchus must not be adjacent to a level 3 bronchus, and a level 5 bronchus can be adjacent to a level 3 bronchus). In some embodiments, the threshold corresponding to the difference between the classification label value of the current node and the classification label value of its parent node can be set to other reasonable values ​​according to actual conditions, and this specification does not limit this. For example, the preset rule may be that the classification label value of the current node must not exceed the classification label value of its parent node by more than three or four levels.

[0091] Step 454: Determine candidate labels.

[0092] In some embodiments, in response to the need to correct the intermediate hierarchical labels of the current node, the intermediate hierarchical labels of neighboring nodes (i.e., sibling nodes of the current node) belonging to the same parent node can be counted, and the intermediate hierarchical label with the highest statistical value can be determined as a candidate label. For example, if a parent node contains 4 child nodes, when correcting child node 1, the intermediate hierarchical labels corresponding to the other 3 child nodes (i.e., the intermediate hierarchical labels of the center line segments contained in the node) can be counted. If the intermediate hierarchical labels are 4, 5, and 5, then the label with the highest frequency (i.e., the highest statistical value) of 5 can be determined as a candidate label.

[0093] Step 455: Determine if the candidate label meets the preset rules. If yes, determine the candidate label as the target hierarchical label for the current node; otherwise, determine the target hierarchical label for the current node based on the intermediate hierarchical label of its parent node. For example, if the value of the candidate label is greater than the hierarchical label value of the current node's parent node, but does not exceed two levels above the parent node's hierarchical label value, then the candidate label is determined as the target hierarchical label for the current node; otherwise, the hierarchical label value of the current node can be corrected to a value equal to 1 plus the hierarchical label value of its parent node. For example, if the parent node's hierarchical label value is 3, and the candidate label value is 6, the candidate label value is greater than 3, and the difference between the candidate label value and the parent node's hierarchical label value is 3, meaning the candidate label value exceeds three levels above the parent node's hierarchical label value. In this case, the hierarchical label value of the current node can be corrected to a value of 4, which is 3 plus 1 from the parent node's hierarchical label value.

[0094] By correcting the classification label of the current node based on the classification label of the sibling node, it is possible to avoid the branch trachea belonging to the same node being assigned to different levels, thereby improving the accuracy and reliability of tracheal classification.

[0095] In some embodiments, after determining the hierarchical label of the current node, it can be determined whether the traversal has ended. If so, proceed to step 460; otherwise, proceed to step 452: continue traversing to correct the intermediate hierarchical label of the next node, until the traversal and correction of all segment nodes are completed.

[0096] Step 460: Map the corrected centerline segmentation tree diagram onto the tracheal centerline.

[0097] The corrected centerline segmentation is represented by a tree diagram with target grading labels. In some embodiments, the tree diagram with target grading labels can be mapped onto the tracheal centerline to obtain a tracheal centerline with target grading labels. For example, the label information of the centerline segments in the tree diagram can be mapped onto the tracheal centerline 413 according to spatial coordinates to obtain a corrected tracheal centerline 463 with target grading labels (for example, each segment of the tracheal centerline can be labeled with its corresponding grading label using numbers).

[0098] By unifying the intra-segment and correcting the inter-segment labels of segment nodes based on a tree diagram, the accuracy of local grading and the overall orderliness can be improved respectively, thereby obtaining more accurate and orderly tracheal grading results while ensuring the richness of grading.

[0099] It should be noted that the above description of method 400 is for illustrative purposes only and does not limit the scope of this specification. Those skilled in the art can make various modifications and changes to method 400 under the guidance of this specification. However, these modifications and changes remain within the scope of this specification.

[0100] Figure 6 This is a schematic diagram of an exemplary tracheal triage method according to some embodiments of this specification.

[0101] like Figure 6As shown, in some embodiments, after acquiring the medical image 613, preprocessing operations such as window level and window width normalization can be performed on the medical image 613, and the preprocessed medical image is input into the preliminary classification module 210. The preliminary classification module 210 can use the trachea segmentation model and the lung lobe segmentation model respectively to obtain the trachea segmentation mask 615 and the lung lobe segmentation mask 617, and extract the trachea centerline 619 based on the trachea segmentation mask 615 using a neighborhood analysis refinement algorithm. In some embodiments, the preliminary classification module 210 can calculate the distance map from the trachea to the trachea centerline based on the trachea segmentation mask 615 and the trachea centerline 619, and combine the distance map from the trachea to the trachea centerline (first distance map), the trachea segmentation mask 615, and the lung lobe segmentation mask 617 as the three-channel input 621. Further, the preliminary classification module 210 can input the three-channel input 621 into the trachea classification model 623 to obtain a relatively reliable and richly leveled initial trachea classification mask 625 on a global scale.

[0102] In some embodiments, the initial tracheal classification mask 625 and the tracheal centerline 619 can be input to the centerline correction module 220. The centerline correction module 220 can obtain the tracheal centerline with initial classification labels by finding the intersection of the tracheal centerline 619 and the initial tracheal classification mask 625. Further, the centerline correction module 220 can trace and traverse from the centerline point at the main tracheal inlet of the tracheal centerline 619 to the end, and segment the centerline according to the bifurcation point, thereby constructing a tree diagram composed of centerline segments. In some embodiments, the centerline correction module 220 can traverse from parent nodes to child nodes based on this tree diagram to unify the initial classification labels contained in each segment node of the tracheal centerline 619 within the segment, thereby obtaining the intermediate classification labels of the tracheal centerline. Furthermore, based on this tree diagram, the parent node can be traversed to the child node, and the intermediate grade label corresponding to each segment node of the tracheal centerline 619 can be corrected between segments, finally resulting in the tracheal centerline 627 with the target grade label.

[0103] In some embodiments, the hierarchical mask reconstruction module 230 can divide the tracheal centerline 627 with target hierarchical labels according to the target hierarchical labels, and calculate the Euclidean distance between the centerline segment contained in each target hierarchical label and the foreground voxels (i.e., voxels of all tracheal regions) in the tracheal segmentation mask 615, thereby obtaining a distance map (second distance map) from the foreground voxels of all trachea to the centerline segment of each target hierarchical label. Further, for each foreground voxel of the trachea, the hierarchical mask reconstruction module 230 can compare the distances from the foreground voxel to the centerline segments of different target hierarchical labels, and assign the target hierarchical label of the corresponding centerline segment with the smallest distance to the current foreground voxel, thereby obtaining the target tracheal hierarchical mask 629 (e.g., ...). Figure 7 (Target tracheal grading mask shown).

[0104] Understandable. Figure 6 The descriptions and related information provided are for illustrative purposes only and do not limit the scope of this specification. Those skilled in the art can make various modifications and changes based on this specification. However, such modifications and changes remain within the scope of this specification.

[0105] Another aspect of this specification provides a tracheal staging device, which includes a processor and a memory for storing instructions that, when executed by the processor, implement the tracheal staging method as described above.

[0106] The beneficial effects that the embodiments of this specification may bring include, but are not limited to: (1) determining the initial tracheal grading mask based on the trained tracheal grading model. This tracheal grading model utilizes the tube diameter features implied by the Euclidean distance map and the rich contextual information such as the relative relationship between the lung lobes contained in the lung lobe mask, thereby achieving high robustness and grading accuracy for bronchi below grade 4 (e.g., grade 5, grade 6, etc.) and strong generalization ability for tracheal variations; (2) mapping the initial tracheal grading mask onto the tracheal centerline as the grading tracheal centerline, and then traversing all centerline points of the tracheal centerline to correct it, obtaining the corrected tracheal centerline with the target grading label. Under the guidance of the initial tracheal grading mask, the grading label of the tracheal centerline is further corrected by combining anatomical prior information, improving the accuracy of the grading. (3) By mapping the tracheal centerline with the target grading label to the tracheal grading mask, the final tracheal grading result mask can be obtained, which can achieve a more accurate and orderly tracheal grading result; (4) By performing intra-segment unification of the initial grading label of the segment nodes based on the tree diagram, intermediate grading labels can be obtained, and inter-segment correction can be performed on the intermediate grading labels, which can improve the accuracy of local grading and the overall orderliness, thereby obtaining a more accurate and orderly tracheal grading result while ensuring the richness of grading; (5) By calculating the distance map from the foreground voxels to the centerline segments of different grading labels in the tracheal segmentation mask, and assigning the grading label of the nearest centerline segment to each foreground voxel, it is possible to reconstruct a more accurate tracheal grading mask with a smoother interface.

[0107] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects may be any one or a combination of the above, or any other possible beneficial effects.

[0108] The basic concepts have been described above. Obviously, for those skilled in the art, the detailed disclosure above is merely illustrative and does not constitute a limitation of this specification. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this specification. Such modifications, improvements, and corrections are suggested in this specification and therefore remain within the spirit and scope of the exemplary embodiments described herein.

[0109] Furthermore, this specification uses specific terms to describe embodiments thereof. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic associated with at least one embodiment of this specification. Therefore, it should be emphasized and noted that references to "an embodiment," "one embodiment," or "an alternative embodiment" in different locations throughout this specification do not necessarily refer to the same embodiment. Moreover, certain features, structures, or characteristics in one or more embodiments of this specification can be appropriately combined.

[0110] Furthermore, unless expressly stated in the claims, the order of processing elements and sequences, the use of numbers and letters, or other names described in this specification are not intended to limit the order of the processes and methods described herein. Although various examples have been discussed in the foregoing disclosure of some embodiments of the invention that are currently considered useful, it should be understood that such details are for illustrative purposes only, and the appended claims are not limited to the disclosed embodiments; rather, the claims are intended to cover all modifications and equivalent combinations that conform to the spirit and scope of the embodiments described herein. For example, while the system components described above can be implemented using hardware devices, they can also be implemented solely using software solutions, such as installing the described system on existing servers or mobile devices.

[0111] Similarly, it should be noted that, in order to simplify the description disclosed herein and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of embodiments in this specification may sometimes combine multiple features into a single embodiment, drawing, or description thereof. However, this method of disclosure does not imply that the subject matter of this specification requires more features than those mentioned in the claims. In fact, the embodiments contain fewer features than all the features of a single embodiment disclosed above.

[0112] In some embodiments, numbers describing the quantity of components and attributes are used. It should be understood that such numbers used in the description of embodiments are modified in some examples with the terms "approximately," "approximately," or "generally." Unless otherwise stated, "approximately," "approximately," or "generally" indicates that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may be changed depending on the characteristics required by individual embodiments. In some embodiments, numerical parameters should take into account specified significant digits and employ a general method of digit reservation. Although the numerical ranges and parameters used to confirm their breadth of range in some embodiments of this specification are approximate values, in specific embodiments, such values ​​are set as precisely as feasible.

[0113] For each patent, patent application, patent application publication, and other material, such as articles, books, specifications, publications, and documents, referenced in this specification, the entire contents of which are incorporated herein by reference. This excludes historical application documents that are inconsistent with or conflict with the content of this specification, as well as documents that limit the broadest scope of the claims in this specification (currently or subsequently appended to this specification). It should be noted that in the event of any inconsistency or conflict between the descriptions, definitions, and / or terminology used in the supplementary materials to this specification and the content of this specification, the descriptions, definitions, and / or terminology used in this specification shall prevail.

[0114] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. A method for tracheal triage, characterized in that, The method includes: Based on the trained tracheal grading model and medical images, the initial tracheal grading mask is determined. The tracheal centerline is determined based on the medical images, and the tracheal centerline with the initial tracheal classification label is determined based on the tracheal centerline and the initial tracheal classification mask. The initial grading label is corrected based on a preset method to obtain the target grading label for the tracheal centerline; Reconstruction is performed based on the tracheal centerline with the target classification label to obtain the target tracheal classification mask. The determination of the initial tracheal classification mask based on the trained tracheal classification model and medical images includes: Based on the medical images, determine the tracheal segmentation mask and the lung lobe segmentation mask; Determine the distance from the trachea to the centerline of the trachea in the trachea segmentation mask; The tracheal segmentation mask, the lung lobe segmentation mask, and the distance map from the trachea to the tracheal centerline are input into the trained tracheal grading model to obtain the output initial tracheal grading mask. The lung lobe segmentation mask reflects the tracheal classification, and the distance map from the trachea to the tracheal centerline reflects the tracheal diameter. The step of correcting the initial grading label based on a preset method to obtain the target grading label for the tracheal centerline includes: The initial grading labels contained in the centerline segments represented by each node in the tree diagram are unified to obtain the intermediate grading labels of the tracheal centerline; and For each node in the tree diagram, based on the intermediate hierarchical labels of the neighboring nodes that belong to the same parent node, the intermediate hierarchical label corresponding to the node is corrected to obtain the target hierarchical label of the tracheal centerline; the tree diagram is obtained by segmenting the tracheal centerline according to the bifurcation point.

2. The method according to claim 1, characterized in that, Before correcting the initial grading label based on a preset method, the method further includes: The tracheal centerline is segmented according to the bifurcation point to obtain a tree diagram composed of multiple centerline segments, where each node of the tree diagram represents one of the centerline segments.

3. The method according to claim 2, characterized in that, The process of unifying the initial grading labels contained in the centerline segments represented by each node of the tree diagram to obtain intermediate grading labels for the tracheal centerline includes: Based on the tree diagram, traverse from the parent node to the child node; For each node in the tree diagram, count all the initial hierarchical labels contained in the centerline segment represented by the node, and determine the initial hierarchical label with the highest statistical value as the intermediate hierarchical label corresponding to the current node.

4. The method according to claim 2, characterized in that, For each node in the tree diagram, based on the intermediate hierarchical labels of neighboring nodes belonging to the same parent node, the intermediate hierarchical label corresponding to that node is corrected to obtain the target hierarchical label of the tracheal centerline, including: Based on the tree diagram, the process traverses from the parent node to the child node and determines whether the intermediate hierarchical labels of the current node need to be corrected according to preset rules. In response to the need for correction, Count the intermediate hierarchical labels of neighboring nodes that share the same parent node as the current node, and determine the intermediate hierarchical label with the highest statistical value as the candidate label. Determine whether the candidate label meets the preset rule. If yes, determine the candidate label as the target hierarchical label of the current node. Otherwise, determine the target hierarchical label of the current node based on the intermediate hierarchical label of the parent node of the current node. The preset rules include: the hierarchical label value of the current node should be greater than the hierarchical label value of its parent node, and the difference between the hierarchical label value of the current node and the hierarchical label value of its parent node should not exceed a threshold.

5. The method according to claim 2, characterized in that, The method further includes: Remove the center line segments in the tree diagram that do not meet the preset conditions to obtain a directed acyclic tree diagram.

6. The method according to claim 1, characterized in that, The process of reconstructing the trachea centerline based on the target tracheal classification label to obtain the target tracheal classification mask includes: The tracheal centerline is divided according to the target classification label to obtain the centerline segment corresponding to each target classification label; For each tracheal region in the tracheal segmentation mask, determine the centerline segment closest to the tracheal region, and set the target classification label of the centerline segment as the classification label of the current tracheal region to obtain the target tracheal classification mask.

7. A tracheal triage system, characterized in that, The system includes: The preliminary grading module is used to determine the initial tracheal grading mask based on the trained tracheal grading model and medical images. A centerline correction module is used to determine the tracheal centerline based on the medical image, and to determine a tracheal centerline with an initial grading label based on the tracheal centerline and the initial tracheal grading mask; and to correct the initial grading label based on a preset method to obtain a target grading label for the tracheal centerline; the correction of the initial grading label based on the preset method to obtain the target grading label for the tracheal centerline includes: unifying the initial grading labels contained in the centerline segments represented by each node in the tree diagram to obtain intermediate grading labels for the tracheal centerline; and for each node in the tree diagram, correcting the intermediate grading label corresponding to that node based on the intermediate grading labels of neighboring nodes belonging to the same parent node to obtain the target grading label for the tracheal centerline; the tree diagram is obtained by segmenting the tracheal centerline according to the branching points. The hierarchical mask reconstruction module is used to reconstruct the trachea centerline based on the target hierarchical label to obtain the target trachea hierarchical mask. The determination of the initial tracheal classification mask based on the trained tracheal classification model and medical images includes: Based on the medical images, determine the tracheal segmentation mask and the lung lobe segmentation mask; Determine the distance from the trachea to the centerline of the trachea in the trachea segmentation mask; The tracheal segmentation mask, the lung lobe segmentation mask, and the distance map from the trachea to the tracheal centerline are input into the trained tracheal grading model to obtain the output initial tracheal grading mask. The lung lobe segmentation mask reflects the tracheal classification, and the distance map from the trachea to the tracheal centerline reflects the tracheal diameter.

8. A computer-readable storage medium storing computer instructions that, when read by a computer, execute the method as described in any one of claims 1-6.

Images