WO2021169191A1 - 一种基于虚拟立体定位像的快速ct扫描方法及系统 - Google Patents

一种基于虚拟立体定位像的快速ct扫描方法及系统 Download PDF

Info

Publication number
WO2021169191A1
WO2021169191A1 PCT/CN2020/108845 CN2020108845W WO2021169191A1 WO 2021169191 A1 WO2021169191 A1 WO 2021169191A1 CN 2020108845 W CN2020108845 W CN 2020108845W WO 2021169191 A1 WO2021169191 A1 WO 2021169191A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
patient
images
positioning
virtual
Prior art date
Application number
PCT/CN2020/108845
Other languages
English (en)
French (fr)
Inventor
曾凯
傅鹏
何健
Original Assignee
南京安科医疗科技有限公司
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 南京安科医疗科技有限公司 filed Critical 南京安科医疗科技有限公司
Publication of WO2021169191A1 publication Critical patent/WO2021169191A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10081—Computed x-ray tomography [CT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]

Definitions

  • the invention relates to the technical field of medical imaging, in particular to a fast CT scanning method and system based on virtual stereo positioning images.
  • the general CT process includes: patient registration; according to scanning needs, collecting horizontal and/or vertical positioning images; CT scanning. Doctors usually judge the scan area of the patient's positioning image based on experience, and the positioning image obtained by this method is not accurate.
  • the Chinese patent "A Smart Scanning Stereo Monitoring Method and System" with the publication number CN110338835A discloses a method for taking an image of a patient to be scanned on a detection table through the camera in a scene with an auxiliary camera, and then processing the image. A method to quickly and accurately determine the scanning area of the patient's positioning image. According to the different characteristics of different patient categories, it accurately locates the feature points, improves the accuracy of the body position information in the image, and further determines the movement information of each part to obtain a more accurate positioning image, but it does not solve the following technical problems :
  • Positioning image scanning also has a certain radiation dose, especially for low-dose CT screening protocols.
  • the dose of positioning image scanning reaches about 30-50%, which is extremely unfavorable to patients; so if the low dose can be reduced or omitted Scanning can greatly reduce the radiation dose received by the patient.
  • the present invention provides a fast CT scanning method and system based on virtual stereo positioning images, which uses virtual positioning images generated by optical images to replace traditional positioning images, reducing the time required for scanning positioning images.
  • the radiation dose received by the patient is reduced, and the patient's anatomical structure can be clearly displayed, so that the doctor can more accurately locate the scanning area and reduce the doctor's workload.
  • a fast CT scanning method based on virtual stereotactic images which is characterized in that it comprises the following steps:
  • the key points in the patient image and the positioning image data template are used as matching points, and the matching calculation is performed according to the position information of the matching points to obtain a stereo virtual positioning image;
  • the patient image in step S1 includes images taken from a top view and a side view
  • the positioning image data template in step S2 uses CT scan human body data, and averages the human body data along the coronal plane and the sagittal plane.
  • the coronal plane and The sagittal plane direction corresponds to the top view and side view respectively.
  • the transformation function used for matching and processing the CT image is established as follows:
  • x pic and y pic are the abscissa and ordinate of the key points in the patient image
  • x ct and y ct are the abscissa and ordinate of the key points in the positioning image data template.
  • the coordinates are respectively brought into the transformation function to minimize the mean square error, and the coefficients k 11 , k 12 , k 13 , k 21 , k 22 , and k 23 are calculated to determine the transformation function.
  • the patient image is input to the trained neural network for key point detection
  • the method is: input the patient image, extract the features in the patient image through the convolutional network, and then output the mask of the key parts of the human body through upsampling .
  • the patient image is input into the trained neural network for key point detection, and the method is: input the patient image, extract the features in the patient image through the feature extraction network, and output the first key point coordinates through rough regression calculation; Then classify and correct the coordinates of the first key point according to the body part, and output the coordinates of the second key point.
  • the patient image includes an optical image and a depth image.
  • the key points include eyes, nose, corners of mouth, neck, shoulders, crotch, knees, and feet.
  • the invention also discloses a fast CT scanning system based on virtual stereo positioning images, which is characterized in that it comprises:
  • Image acquisition module for patient images taken from different angles including the area to be scanned
  • Image processing module for key point detection on patient images
  • CT scanning module used to scan the target object and obtain the corresponding CT data
  • CT data labeling module used to mark the key points of the positioning image data template
  • the virtual positioning image generation module is used to perform matching operations on the key points in the patient image and the positioning image data template, and generate a virtual positioning image.
  • the image acquisition module adopts a binocular vision system to simultaneously acquire two patient images from different angles.
  • the present invention uses virtual positioning images generated by optical images to replace traditional positioning images, reducing the time required for scanning positioning images, reducing scanning time for a single patient, improving scanning efficiency, and reducing the workload of doctors; and can clearly display
  • the patient’s anatomy allows the doctor to more accurately locate the scanning area.
  • binocular vision system can be used to acquire optical image and depth image (rgb-d image) in real time at the same time, saving the time of mechanical movement; generating virtual stereo based on rgb-d image Position the image and quickly select the scan area for the CT scan.
  • This method saves the radiation dose and scanning time of the scanning of the positioning image, greatly speeds up the scanning process, and at the same time provides a three-dimensional positioning image, which enables more precise scanning and positioning.
  • the method of the present invention will be able to display the patient's anatomical structure on the image taken by the camera, so that the doctor can more accurately locate the scanning area, and the virtual stereotaxic image generated by the method of the present invention can avoid the patient's scan positioning image and reduce the number of patients The radiation dose received.
  • the method of the present invention makes full use of the existing CT product hardware, and finally integrates the algorithm in the image chain.
  • Figure 1 is a flow chart of the virtual stereo dual positioning imaging method of the present invention
  • Figure a is a schematic diagram of a patient's image taken from above the hospital bed from a top perspective
  • Figure b is a schematic diagram of a patient's image taken from the side of the hospital bed from a side perspective;
  • Figure 3 Figure c and Figure d are flow charts of two key point detection methods respectively;
  • FIG. 4 is a flowchart of a CT scan using the virtual stereo dual positioning imaging method of the present invention.
  • Fig. 5 is a flowchart of a conventional CT scan of dual positioning images.
  • the present invention provides a fast CT scanning method and system based on virtual stereo positioning images, which uses rgb-d images to generate virtual positioning images, which specifically includes steps 1) to 5).
  • the input patient image is a patient image taken in two directions.
  • a binocular vision system can be used to simultaneously capture optical images and depth images (rgb-d images) in real time, saving time for mechanical movement.
  • the present invention can use two network image key point detection methods as shown in FIG. 3 to process patient images.
  • the input image, the feature is extracted through the convolutional network, and then the mask of the key parts of the human body is output through upsampling.
  • the feature extraction network extracts the features, the rough regression output key point coordinates, and then the key point coordinates are classified and corrected according to the position, and finally the human body key point positions and coordinates are output.
  • the feature extraction network is a network framework such as DenseNet and ResNet.
  • the two network structures use similar training methods.
  • the image size is 640*480, and manually mark the coordinates of the key points of the human body.
  • the key points include but are not limited to the eyes and nose. , Mouth corners, neck, shoulders, hips, knees and feet, etc., train the network through the tensorflow framework.
  • the network input data size is 1*640*480
  • the network output data shape is nnk+1*640*480, where nnk is the number of key points in the human body
  • the network loss function is the cross-entropy function, as follows:
  • y t is the real data mark
  • y p is the network predicted probability
  • the positioning image data template adopts the human body data of the CT scan, and the data is averaged along the coronal plane and the sagittal plane, respectively, corresponding to the image taken from the top view angle and the image taken from the side. Then mark the key points in the positioning image data template, where the key points include but are not limited to the eyes, nose, corners of the mouth, neck, shoulders, crotch, knees and feet.
  • the x-direction transformation function parameter the function is:
  • x pic x ct *(1+k 11 *r 2 +k 12 *r 4 +k 13 *r 6 )
  • y pic y ct *(1+k 21 *r 2 +k 22 *r 4 +k 23 *r 6 )
  • x pic and y pic are the abscissa and ordinate of the key points in the captured image
  • x ct and y ct are the abscissa and ordinate of the key points in the CT image
  • r 2 x 2 +y 2 .
  • the x and y pairs of the matching points are respectively brought into the above two formulas to minimize the mean square error, and the parameters k 11 , k 12 , k 13 , k 21 , k 22 , and k 23 are calculated.
  • the positioning image data template is transformed according to the above two formulas, and the virtual positioning image data matching the captured image is output.
  • the doctor can judge the patient's anatomical structure based on the matched virtual positioning image data, so that the patient's scanning area can be more accurately positioned.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Databases & Information Systems (AREA)
  • Multimedia (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Apparatus For Radiation Diagnosis (AREA)
  • Image Analysis (AREA)

Abstract

本发明公开了一种基于虚拟立体定位像的快速CT扫描方法及系统，方法包括步骤：输入不同角度拍摄的包括待扫描区域在内的病人图像，对病人图像进行关键点检测；输入定位像数据模板并标记出关键点；根据病人图像和定位像数据模板中的关键点，进行匹配计算，获得立体虚拟定位像；根据所述立体虚拟定位像确定病人的CT扫描区域，输出最终的CT图像。本发明利用光学图像生成的虚拟定位像替代传统定位像，减少扫描定位像所需要的时间，进而减少了病人所接受的辐射剂量，而且能够清晰地显示病人的解剖结构，使得医生能够更加精确地定位扫描区域，减轻了医生的工作量。

Description

一种基于虚拟立体定位像的快速CT扫描方法及系统 技术领域
本发明涉及医学成像技术领域，具体涉及一种基于虚拟立体定位像的快速CT扫描方法及系统。
背景技术
目前在CT扫描流程中，为了更加准确的选定扫描区域，需要首先扫描定位像，然后在定位像中选择扫描区域。如图5所示，一般CT流程包括：病人登记；根据扫描需要，采集水平和/或垂直方向的定位像；CT扫描。医生通常根据经验判断病人定位像扫描区域，此方法获得的定位像并不准确。公开号为CN110338835A的中国专利“一种智能扫描立体监测方法及系统”，公开了一种在有辅助相机的场景下，通过相机拍摄待扫描病人位于检测台上的图像、然后对图像进行处理，实现快速准确判断病人定位像扫描区域的方法。其根据不同患者类别信息不同的特点，准确地定位特征点，提高了图像中体位信息的准确性，并进一步判断各部位的移动信息，获得较为准确的定位像，但是其并没有解决如下技术问题：
1、定位像扫描也存在一定的辐射剂量，尤其是对于低剂量CT筛查的协议，定位像扫描的剂量占比达到30-50％左右，对患者极为不利；所以如果能够减少或省略低剂量扫描，就能够大幅度减少患者接受的辐射剂量。
2、正常CT扫描时，机架是连续旋转的，准备时间很短。但是扫描定位像需要机架从运动到停止，直到定位在指定角度后再进行扫描，整个机架的加速、减速和定位均需要时间，因此扫描的准备时间比较长，影响了扫描效率，尤其是在应用双定位像的条件下，不能适应体检、筛查等大规模筛查的需求。
3、在常规的CT扫描中，由于扫描效率和剂量等因素的制约，常常只应用一个定位像，而根据一个定位像来精确确定扫描范围，对医生临床经验要求较高，欠缺临床经验的医生选择的定位像区域可能不够准确，如果重复做定位像扫描，病人可能遭受过多的辐射。
发明内容
技术目的：为解决上述技术问题，本发明提供了一种基于虚拟立体定位像的快速CT扫描方法及系统，利用光学图像生成的虚拟定位像替代传统定位像，减少扫描定位像所 需要的时间，进而减少了病人所接受的辐射剂量，而且能够清晰地显示病人的解剖结构，使得医生能够更加精确地定位扫描区域，减轻了医生的工作量。
技术方案：为实现这一技术目的，本发明采用了如下技术方案：
一种基于虚拟立体定位像的快速CT扫描方法，其特征在于，包括步骤：
S1、输入不同角度拍摄的包括待扫描区域在内的病人图像，并对病人图像进行人体部位的关键点检测；
S2、输入人体数据的定位像数据模板，在定位像数据模板中标记出人体部位的关键点；
S3、根据病人图像和定位像数据模板中的关键点，将对应相同部位的关键点作为匹配点，根据匹配点的位置信息进行匹配计算，获得立体虚拟定位像；
S4、根据所述立体虚拟定位像确定病人的CT扫描区域，输出最终的CT图像。
优选地，所述步骤S1中病人图像包括俯视和侧面视角拍摄的图像，步骤S2中定位像数据模板采用CT扫描的人体数据，分别沿冠状面和矢状面对人体数据取平均，冠状面和矢状面方向分别对应俯视和侧面视角。
优选地，所述步骤S3中，匹配计算时，建立用于匹配处理CT图像的变换函数如下：
Figure PCTCN2020108845-appb-000001
其中，x pic、y pic为病人图像中关键点的横坐标、纵坐标，x ct、y ct为定位像数据模板中关键点的横坐标、纵坐标，将若干个匹配点的横坐标和纵坐标分别带入变换函数中，使得均方误差最小，计算得到系数k 11、k 12、k 13、k 21、k 22、k 23的值，确定变换函数。
优选地，所述步骤S1中将病人图像输入训练好的神经网络进行关键点检测，方法为：输入病人图像、通过卷积网络提取病人图像中的特征，然后通过上采样输出人体关键部位掩膜。
优选地，所述步骤S1中将病人图像输入训练好的神经网络进行关键点检测，方法为：输入病人图像、通过特征提取网络提取病人图像中的特征，粗略回归计算输出第一关键点坐标；然后对第一关键点坐标根据人体部位进行分类和修正，输出第二关键点坐标。
优选地，所述病人图像包括光学图像和深度图像。
优选地，所述关键点包括双眼、鼻子、嘴角、颈部、双肩、胯部、膝盖、双脚。
本发明还公开了一种基于虚拟立体定位像的快速CT扫描系统，其特征在于，包括：
图像获取模块，用于不同角度拍摄的包括待扫描区域在内的病人图像；
图像处理模块，用于对病人图像进行关键点检测；
CT扫描模块，用于扫描目标对像并获得对应的CT数据；
CT数据标注模块，用于对定位像数据模板进行关键点标记；
虚拟定位像生成模块，用于对病人图像和定位像数据模板中的关键点进行匹配运算，并生成虚拟定位像。
优选地，所述图像获取模块采用双目视觉系统，同时采集两个不同角度的病人图像。
技术效果：由于采用了上述技术方案，本发明具有如下技术效果：
(1)、本发明利用光学图像生成的虚拟定位像替代传统定位像，减少扫描定位像所需要的时间，减少单个病人的扫描时间，提升扫描效率，减轻医生的工作量；而且能够清晰地显示病人的解剖结构，使得医生能够更加精确地定位扫描区域。
(2)、利用光学采集图像，可以采用双目视觉的系统来同时实时地采集光学图像和深度图像(rgb-d图像)，节省了机械运动的时间；根据rgb-d图像来生成虚拟的立体定位像，快速的为CT扫描选择扫描区域。此方法节省了定位像的扫描的辐射剂量和扫描时间，大幅度加快了扫描流程，同时提供立体定位像，能够更加精确的进行扫描定位。
(3)、本发明的方法将能够在相机拍摄的图像上展示病人的解剖结构，使得医生能够更加精准定位扫描区域，且本发明方法生成的虚拟立体定位像可以避免病人扫描定位像，减少病人所接受的辐射剂量。
(4)、本发明的方法充分利用现有CT产品硬件，最终将算法集成在影像链中。
附图说明
图1为本发明的虚拟立体双定位像方法的流程图；
图2中，图a为俯视视角从病床上方拍摄病人图像的示意图；
图b为侧视视角从病床侧面拍摄病人图像的示意图；
图3中，图c和图d分别为两种关键点检测方法的流程图；
图4是采用本发明的虚拟立体双定位像方法的CT扫描的流程图；
图5为常规的双定位像的CT扫描的流程图。
具体实施方式
如图1至图4所示，本发明提供一种基于虚拟立体定位像的快速CT扫描方法及系统，其利用rgb-d图像生成虚拟定位像，具体包括步骤1)～5)。
1)、病人图像输入：
如图2所示，输入病人图像为两个方向拍摄的病人图像。利用光学采集图像，可以采用双目视觉的系统来同时实时的采集光学图像和深度图像(rgb-d图像)，节省了机械运动的时间。
2)、关键点检测：
本发明可采用如图3所示的两种网络图像关键点检测方法对病人图像进行处理。图a所示，输入图像、通过卷积网络提取特征，然后通过上采样输出人体关键部位掩膜。图b所示，输入图像、通过特征提取网络提取特征，粗略回归输出关键点坐标，然后将关键点坐标根据部位进行分类和修正，最终输出人体关键点部位和坐标。特征提取网络为DenseNet、ResNet等网络框架。两种网络结构采用类似的训练方法。
以图a所示方法中的网络为例，对其进行网络训练：从临床数据中选取10000多图像，图像大小为640*480，手动标记人体关键点坐标，其中关键点包括不限于双眼、鼻子、嘴角、颈部、双肩、胯部、膝盖和双脚等，通过tensorflow框架训练网络。网络输入数据大小为1*640*480，(a)网络输出数据形状为nkp+1*640*480，其中nkp为人体关键点数量，网络损失函数为交叉熵函数，如下：
Loss＝y t*logy p+(1-y t)*log(1-y p)
其中，y t为真实数据标记，y p为网络预测概率。
3)、定位像数据模板以及关键点标记
定位像数据模板采用CT扫描的人体数据，分别沿冠状面和矢状面对数据取平均，分别对应俯视视角拍摄的图像和侧面拍摄的图像。然后在定位像数据模板中标记出关键点，其中关键点部位包括不限于双眼、鼻子、嘴角、颈部、双肩、胯部、膝盖和双脚等。
4)、图像和定位像数据模板匹配
根据图像关键点和定位像数据模板的关键点将图像和定位像数据模板匹配。通过图像和定位像模板数据中相同部位的关键点计算图像匹配的结果。
具体计算匹配关键点的方法如下：
x方向变换函数参数，其函数为：
x pic＝x ct*(1+k 11*r 2+k 12*r 4+k 13*r 6)
y方向变换函数参数，其函数为：
y pic＝y ct*(1+k 21*r 2+k 22*r 4+k 23*r 6)
其中，x pic、y pic为拍摄图像中关键点的横坐标、纵坐标，x ct、y ct为CT图像中关键点横坐标、纵坐标，r 2＝x 2+y 2。
将匹配点的x，y对分别带入上面两个公式中，使得均方误差最小，计算参数k 11、k 12、k 13、k 21、k 22、k 23。
5)、输出匹配的虚拟定位像图像
将定位像数据模板根据上述两个公式进行变换，输出与拍摄图像匹配的虚拟定位像数据。医生可以根据匹配后的虚拟定位像数据判断病人的解剖结构，从而能够更加精确定位病人的扫描区域。
以上所述仅是本发明的优选实施方式，应当指出：对于本技术领域的普通技术人员来说，在不脱离本发明原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也应视为本发明的保护范围。

Claims (9)

  1. 一种基于虚拟立体定位像的快速CT扫描方法，其特征在于，包括步骤：
    S1、输入不同角度拍摄的包括待扫描区域在内的病人图像，并对病人图像进行人体部位的关键点检测；
    S2、输入人体数据的定位像数据模板，在定位像数据模板中标记出人体部位的关键点；
    S3、根据病人图像和定位像数据模板中的关键点，将对应相同部位的关键点作为匹配点，根据匹配点的位置信息进行匹配计算，获得立体虚拟定位像；
    S4、根据所述立体虚拟定位像确定病人的CT扫描区域，输出最终的CT图像。
  2. 根据权利要求1所述的一种基于虚拟立体定位像的快速CT扫描方法，其特征在于：所述步骤S1中病人图像包括俯视和侧面视角拍摄的图像，步骤S2中定位像数据模板采用CT扫描的人体数据，分别沿冠状面和矢状面对人体数据取平均，冠状面和矢状面方向分别对应俯视和侧面视角。
  3. 根据权利要求1所述的一种基于虚拟立体定位像的快速CT扫描方法，其特征在于：所述步骤S3中，匹配计算时，建立用于匹配处理CT图像的变换函数如下：
    Figure PCTCN2020108845-appb-100001
    其中，x pic、y pic为病人图像中关键点的横坐标、纵坐标，x ct、y ct为定位像数据模板中关键点的横坐标、纵坐标，将若干个匹配点的横坐标和纵坐标分别带入变换函数中，使得均方误差最小，计算得到系数k 11、k 12、k 13、k 21、k 22、k 23的值，确定变换函数。
  4. 根据权利要求1所述的一种基于虚拟立体定位像的快速CT扫描方法，其特征在于，所述步骤S1中将病人图像输入训练好的神经网络进行关键点检测，方法为：输入病人图像、通过卷积网络提取病人图像中的特征，然后通过上采样输出人体关键部位掩膜。
  5. 根据权利要求1所述的一种基于虚拟立体定位像的快速CT扫描方法，其特征在于，所述步骤S1中将病人图像输入训练好的神经网络进行关键点检测，方法为：输入病人图像、通过特征提取网络提取病人图像中的特征，粗略回归计算输出第一关键点坐标；然后对第一关键点坐标根据人体部位进行分类和修正，输出第二关键点坐标。
  6. 根据权利要求1所述的一种基于虚拟立体定位像的快速CT扫描方法，其特征在于：所述病人图像包括光学图像和深度图像。
  7. 根据权利要求1所述的一种基于虚拟立体定位像的快速CT扫描方法，其特征在于：所述关键点包括双眼、鼻子、嘴角、颈部、双肩、胯部、膝盖、双脚。
  8. 用于实现权利要求1至7任一所述方法的一种基于虚拟立体定位像的快速CT扫描系统，其特征在于，包括：
    图像获取模块，用于不同角度拍摄的包括待扫描区域在内的病人图像；
    图像处理模块，用于对病人图像进行关键点检测；
    CT扫描模块，用于扫描目标对像并获得对应的CT数据；
    CT数据标注模块，用于对定位像数据模板进行关键点标记；
    虚拟定位像生成模块，用于对病人图像和定位像数据模板中的关键点进行匹配运算，并生成虚拟定位像。
  9. 根据权利要求8所述的一种基于虚拟立体定位像的快速CT扫描系统，其特征在于：所述图像获取模块采用双目视觉系统，同时采集两个不同角度的病人图像。
PCT/CN2020/108845 2020-02-26 2020-08-13 一种基于虚拟立体定位像的快速ct扫描方法及系统 WO2021169191A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010118433.2A CN110956633B (zh) 2020-02-26 2020-02-26 一种基于虚拟立体定位像的快速ct扫描方法及系统
CN202010118433.2 2020-02-26

Publications (1)

Publication Number Publication Date
WO2021169191A1 true WO2021169191A1 (zh) 2021-09-02

Family

ID=69985759

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/108845 WO2021169191A1 (zh) 2020-02-26 2020-08-13 一种基于虚拟立体定位像的快速ct扫描方法及系统

Country Status (2)

Country Link
CN (1) CN110956633B (zh)
WO (1) WO2021169191A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102459888B1 (ko) * 2021-12-15 2022-10-31 성균관대학교산학협력단 최적 스캔 방법 및 이를 적용한 스캔 시스템
CN116643505A (zh) * 2023-07-21 2023-08-25 深圳视腾达科技有限公司 一种智能家居用交互控制方法及系统
CN117672472A (zh) * 2023-12-19 2024-03-08 天河超级计算淮海分中心 一种基于腹部ct影像的对称标注方法

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110956633B (zh) * 2020-02-26 2020-11-24 南京安科医疗科技有限公司 一种基于虚拟立体定位像的快速ct扫描方法及系统
CN112700844B (zh) * 2020-12-23 2023-05-23 中国人民解放军总医院 一种通过图像识别ct关键环节的方法及装置
CN112741643A (zh) * 2020-12-31 2021-05-04 苏州波影医疗技术有限公司 一种可自动定位和扫描的ct系统及其定位和扫描方法
CN112862869B (zh) * 2020-12-31 2024-05-28 上海联影智能医疗科技有限公司 图像扫描处理方法、成像扫描设备、电子设备及可读介质
CN112932522A (zh) * 2021-02-03 2021-06-11 上海联影医疗科技股份有限公司 医学图像获取方法、装置和计算机设备
CN113645307A (zh) * 2021-08-18 2021-11-12 中国人民解放军东部战区总医院 核磁共振系统网络共享定位数据库系统及建立使用方法
CN113520443A (zh) * 2021-08-19 2021-10-22 上海联影医疗科技股份有限公司 Pet系统参数推荐方法以及pet系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140205168A1 (en) * 2013-01-23 2014-07-24 Samsung Electronics Co., Ltd. Image processing apparatus, ultrasonic imaging apparatus, and imaging processing method for the same
US20140233820A1 (en) * 2012-11-01 2014-08-21 Virginia Commonweath University Segmentation and Fracture Detection in CT Images
CN104274196A (zh) * 2013-07-09 2015-01-14 上海西门子医疗器械有限公司 计算机断层扫描设备
CN108670286A (zh) * 2018-06-13 2018-10-19 上海联影医疗科技有限公司 一种ct系统和ct扫描方法
CN110544285A (zh) * 2019-10-30 2019-12-06 南京安科医疗科技有限公司 一种矫正头部ct图像中头部位置的方法及系统
CN110956633A (zh) * 2020-02-26 2020-04-03 南京安科医疗科技有限公司 一种基于虚拟立体定位像的快速ct扫描方法及系统

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017205351A1 (en) * 2016-05-23 2017-11-30 Mako Surgical Corp. Systems and methods for identifying and tracking physical objects during a robotic surgical procedure
WO2019084331A1 (en) * 2017-10-25 2019-05-02 Verma Nikhil N GUIDE TO SPECIFIC SURGICAL INSTRUMENT OF PATIENT
CN108831235A (zh) * 2018-06-12 2018-11-16 苏州敏行医学信息技术有限公司 虚拟医学影像检查训练方法及医学影像检查训练系统

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140233820A1 (en) * 2012-11-01 2014-08-21 Virginia Commonweath University Segmentation and Fracture Detection in CT Images
US20140205168A1 (en) * 2013-01-23 2014-07-24 Samsung Electronics Co., Ltd. Image processing apparatus, ultrasonic imaging apparatus, and imaging processing method for the same
CN104274196A (zh) * 2013-07-09 2015-01-14 上海西门子医疗器械有限公司 计算机断层扫描设备
CN108670286A (zh) * 2018-06-13 2018-10-19 上海联影医疗科技有限公司 一种ct系统和ct扫描方法
CN110544285A (zh) * 2019-10-30 2019-12-06 南京安科医疗科技有限公司 一种矫正头部ct图像中头部位置的方法及系统
CN110956633A (zh) * 2020-02-26 2020-04-03 南京安科医疗科技有限公司 一种基于虚拟立体定位像的快速ct扫描方法及系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
GAO QI, ZENG DONG ,BIAN ZHAOYING, HUANG JING, MA JIANHUA: "CT scanning protocol optimization based on scout image noise analysis", THE 16TH CHINA ACADEMIC CONFERENCE ON STEREOLOGY AND IMAGE ANALYSIS-INTERSECTION, INTEGRATION, INNOVATION; HAIKOU, HAINAN, CHINA, CHINESE SOCIETY OF STEREOLOGY, CN, 1 October 2019 (2019-10-01) - 17 October 2019 (2019-10-17), CN, pages 226 - 230, XP055841605 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102459888B1 (ko) * 2021-12-15 2022-10-31 성균관대학교산학협력단 최적 스캔 방법 및 이를 적용한 스캔 시스템
CN116643505A (zh) * 2023-07-21 2023-08-25 深圳视腾达科技有限公司 一种智能家居用交互控制方法及系统
CN116643505B (zh) * 2023-07-21 2023-11-03 深圳视腾达科技有限公司 一种智能家居用交互控制方法及系统
CN117672472A (zh) * 2023-12-19 2024-03-08 天河超级计算淮海分中心 一种基于腹部ct影像的对称标注方法
CN117672472B (zh) * 2023-12-19 2024-07-12 天河超级计算淮海分中心 一种基于腹部ct影像的对称标注方法

Also Published As

Publication number Publication date
CN110956633B (zh) 2020-11-24
CN110956633A (zh) 2020-04-03

Similar Documents

Publication Publication Date Title
WO2021169191A1 (zh) 一种基于虚拟立体定位像的快速ct扫描方法及系统
US10318839B2 (en) Method for automatic detection of anatomical landmarks in volumetric data
US10881353B2 (en) Machine-guided imaging techniques
EP3378534B1 (en) Treatment assistance system and operation method therefor, and storage medium for storing treatment assistance program
Bulatova et al. Assessment of automatic cephalometric landmark identification using artificial intelligence
US20120235993A1 (en) Display Orientation Adjustment Device And Adjustment Program For Medical Three-Dimensional Image
JP6345468B2 (ja) 医用画像診断装置
JP2018068400A (ja) 動態画像処理装置
KR102461343B1 (ko) 금속 인공 음영이 포함된 의료영상에서 치아 랜드마크 자동 검출 방법 및 시스템
CN109528305A (zh) 计算机辅助的股骨粗隆间骨折对位对线半自动复位方法
CN105894508A (zh) 一种医学图像自动定位质量的评估方法
CN113274129A (zh) 一种基于虚拟现实的心胸外科手术辅助控制系统
CN107240128A (zh) 一种基于轮廓特征的x线片和彩色照片配准方法
CN116309302A (zh) 一种头颅侧位片关键点提取方法
Singh et al. Estimating a patient surface model for optimizing the medical scanning workflow
CN116958443A (zh) 基于smplx重建数字人可量化检测模型、方法及应用
JP6249972B2 (ja) 粒子線治療システム
CN115006737A (zh) 基于深度相机的放射治疗体位监测系统
CN108143501A (zh) 一种基于体表静脉特征的解剖投影方法
CN115880469B (zh) 一种表面点云数据与三维影像的配准方法
JP2018038908A (ja) 医用画像診断装置
KR102534981B1 (ko) 표면 영상유도 기반의 환자 위치 정렬 및 모니터링 시스템
EP4277529B1 (en) Chest x-ray system and method
Cao et al. An improved multi-resolution 2D/3D registration method
Desai et al. Arytenoid cartilage feature point detection using laryngeal 3D CT images in Parkinson's disease

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: 20921098

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: 20921098

Country of ref document: EP

Kind code of ref document: A1