WO2022037548A1 - 一种基于深度学习的脊椎mri影像关键点检测方法 - Google Patents

一种基于深度学习的脊椎mri影像关键点检测方法 Download PDF

Info

Publication number
WO2022037548A1
WO2022037548A1 PCT/CN2021/112874 CN2021112874W WO2022037548A1 WO 2022037548 A1 WO2022037548 A1 WO 2022037548A1 CN 2021112874 W CN2021112874 W CN 2021112874W WO 2022037548 A1 WO2022037548 A1 WO 2022037548A1
Authority
WO
WIPO (PCT)
Prior art keywords
vertebra
edge
information
vertebrae
mri
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2021/112874
Other languages
English (en)
French (fr)
Inventor
刘刚
郑友怡
方向前
马成龙
赵兴
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University
Original Assignee
Zhejiang University
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 Zhejiang University filed Critical Zhejiang University
Priority to JP2022578644A priority Critical patent/JP7489732B2/ja
Publication of WO2022037548A1 publication Critical patent/WO2022037548A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G06T2207/10088—Magnetic resonance imaging [MRI]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; 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]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30008—Bone
    • G06T2207/30012—Spine; Backbone

Definitions

  • the invention belongs to the fields of computer vision and artificial intelligence, and particularly relates to a method for detecting key points of spine MRI images based on deep learning.
  • the present invention aims to detect the key points of the spine MRI image, and the previous key point detection work of the spine MRI image mostly relies on the manual annotation of experts. Manual annotation is inefficient and subject to the subjective influence of experts, especially not suitable for large-scale data processing and analysis.
  • most of the methods that try to use artificial intelligence technology to detect are using the underlying features of the image, such as the paper (Ebrahimi S, Angelini E, Gajny L, et al.Lumbar spine posterior corner detection in X-rays using Haar-based features[C]// In 2016 IEEE 13th international symposium on biomedical imaging (ISBI).
  • the Harr feature is used to detect the corners of the vertebrae, but this method only uses the underlying information of the image, and the robustness is poor, only suitable for In some specific scenarios, it is not suitable for complex and changeable medical scenarios.
  • the present invention realizes a more robust and accurate spinal MRI image key point detection method by establishing a high-quality data set and utilizing the excellent learning ability of deep learning.
  • a deep learning-based spine MRI image key point detection method comprising the following steps:
  • Step 1 Input the spine MRI image into the trained target detection network to obtain the position information of each vertebra and whether it is the coarse-grained label of S1;
  • Step 2 Use all the vertebrae obtained in step 1 and the located S1 position, and combine with the physiological structure information of the spine itself to filter the false positive detection results and identify the fine-grained labels to which each vertebra belongs.
  • Step 3 Cut out the vertebrae detected in step 2 and some surrounding areas, and input them into the trained key point detection network to detect the upper and lower boundaries of each vertebra, UA, UM, UP, LA, LM, LP, a total of six key points location information.
  • Step 4 Use the trained segmentation network to segment the vertebrae obtained in step 2 to obtain edge information, and modify the position information of the key points obtained in step 3 according to the edge information obtained in step 4 to obtain the final key point prediction result.
  • the coarse-grained labels of the vertebrae are S1 and NS1 (wherein S1 refers to sacral 1, and NS1 refers to all other vertebrae except sacral 1), and the target detection network is YOLOv3.
  • the third step is realized by the following sub-steps.
  • the aspect ratio threshold is 1.6
  • the upper edge height threshold is 5.
  • step 4 is specifically:
  • the vertebral edge segmentation network consists of a down-sampling part and an up-sampling part.
  • the structure of the down-sampling part is to remove the resnet50 of the fully connected layer.
  • the up-sampling part is composed of the corresponding four-stage up-sampling convolution blocks.
  • the up-sampling convolution block structure is upsampling->conv->bn->relu.
  • (4.2) Training the vertebral edge segmentation network First, use the key point annotation information to establish a coarse-grained segmentation data set to pre-train the vertebral edge segmentation network, and then construct an accurate fine-grained segmentation data set to further train the segmentation network.
  • the segmentation results are further modified by using the characteristics of the CRF and the large gradient of the image at the edge to obtain more accurate edge segmentation information.
  • step (4.3) is specifically:
  • the beneficial effect of the present invention is that the present invention utilizes the deep learning technology to realize the key point detection of the spine MRI image, avoids the tedious manual labeling, and reduces the burden on the doctor.
  • the present invention can avoid the influence of subjective factors of doctors, and can also process large-scale data in batches, providing a data basis for further spinal MRI image analysis, and the method of the present invention can be developed into an interactive visual MRI spinal image.
  • Key point automatic labeling software, the obtained key point detection results can calculate the intervertebral disc height index and lumbar anterior process angle.
  • FIG. 1 is a schematic diagram of detecting key points in the present invention.
  • Fig. 2 is the overall flow chart of the present invention.
  • FIG. 3 is a schematic diagram of the key point correction of the present invention.
  • the data sets of the training network in the present invention are all self-built data sets, and the key points of the spine are marked by doctors and experts, and the key points of detection are the upper and lower boundaries of each vertebra.
  • UA, UM, UP, LA, LM, LP A total of six key points.
  • the basic process of model training is as follows:
  • the target detection network described in Figure 2 is preferably YOLOv3.
  • the vertebra is divided into two categories: S1 and NS1 (sacral 1 and non-sacral 1) according to whether the vertebra is S1 or not, as coarse-grained label information.
  • the data set for training the detection network is constructed from the original data set, and the YOLOv3 network is trained according to the key point labels of each vertebra to calculate the bounding boxes of the vertebrae and the coarse-grained category labels.
  • the method uses the structural information of the spine to filter the false positive prediction results and determines the category to which each vertebra belongs.
  • the specific method is: use S1 as the To locate the vertebrae, calculate the height of each vertebra center on the image and sort the detected vertebrae according to the centroid height. Then, according to the physiological structure information of the human spine, from bottom to top, corresponding labels such as S1, L5, L4, L3, L2, L1, T12, T11 are assigned to each detected vertebra.
  • the vertebrae may not be completely captured at the top of the picture.
  • This method filters such objects by calculating the aspect ratio of the vertebrae and whether the center height meets the threshold requirements.
  • the aspect ratio of the vertebrae can be calculated according to the target detection results.
  • the key point detection network described in Figure 2 is preferably a U-shaped network or a Stacked Hourglass Network.
  • the training data is the vertebrae image cropped from the original image and the heat map corresponding to the surrounding image and key points.
  • the online hard sample mining method is used in the training process of the point detection network.
  • this method trains a U-shaped deep convolutional neural network as a vertebral edge segmentation network for segmenting vertebral edges.
  • the vertebral edge segmentation network consists of a down-sampling part and an up-sampling part.
  • the structure of the downsampling part is to remove the Resnet50 of the fully connected layer, and the upsampling part is composed of the corresponding four-stage upsampling convolution blocks.
  • the upsampling convolution block structure is upsampling->conv->bn->relu.
  • this method first uses the key point annotation information to establish a coarse-grained segmentation data set to pre-train the vertebral edge segmentation network, and then constructs an accurate fine-grained segmentation data set to further train the segmentation network.
  • the method also uses the conditional random field (CRF) and the characteristics of large image gradient at the edge to further modify the segmentation result to obtain more accurate edge segmentation information.
  • CCF conditional random field
  • this method uses the edge information to correct the detected key points.
  • the edge is an extension line, and the farthest intersection of the extension line and the edge of the vertebra is used as the corrected UA, LA, UM, LM, UP, and LP coordinates.
  • the above method was developed into an interactive visual automatic detection and labeling software.
  • the present invention needs to complete the following steps in the spine MRI image key point detection process:
  • Step 1 Input the spine MRI image into the trained target detection network YOLOv3 (Redmon J, Farhadi A.Yolov3:An incremental improvement[J].arXiv preprint arXiv:1804.02767,2018.), obtain the position information of each vertebra and whether it is Coarse-grained labels for S1;
  • Step 2 Use all the vertebrae obtained in step 1 and the located S1 position, and combine with the physiological structure information of the spine itself to filter the false positive detection results and identify the fine-grained labels to which each vertebra belongs.
  • Step 3 Cut out the vertebrae detected in step 2 and some surrounding areas, and input them into the trained key point detection network to detect the upper and lower boundaries of each vertebra, UA, UM, UP, LA, LM, LP, a total of six key points heatmap.
  • the coordinates of the key points in the heatmap can be extracted from but not limited to papers (Zhang F, Zhu X, Dai H, et al. Distribution-aware coordinate representation for human pose estimation [C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020:7093-7102.) method.
  • Step 4 Use the trained segmentation network to segment the vertebrae obtained in step 2 to obtain edge information, and modify the position information of key points obtained in step 3 according to the edge information obtained in step 4.
  • This method makes an extension line along the line connecting UA-LA, UM-LM and UP-LP to the edge of the vertebra, and takes the farthest intersection of the extension line and the edge of the vertebra as the corrected UA, LA, UM, LM, UP , LP coordinates.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • General Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Medical Informatics (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)
  • Image Analysis (AREA)
  • Apparatus For Radiation Diagnosis (AREA)

Abstract

本文提出的发明称为一种基于深度学习的脊椎MRI影像关键点检测方法。本发明公开了一种基于深度学习的脊椎MRI影像关键点检测方法，该方法首先利用深度目标检测网络检测定位脊椎MRI影像中的椎骨并识别出S1(骶1)作为定位脊椎，然后结合脊椎的结构信息过滤假阳性检测结果和判断各块椎骨的细粒度标签。随后利用关键点检测网络分别检测各块椎骨上下边界UA、UM、UP、LA、LM、LP共计六个关键点，再结合边缘信息确定修正各个椎骨的关键点位置，最后可开发成交互式的可视化MRI脊椎影像关键点自动标注软件。本发明可自动提取脊椎MRI影像的关键点，在医学影像分析、辅助医疗等方面有着巨大的应用价值。

Description

一种基于深度学习的脊椎MRI影像关键点检测方法 技术领域
本发明属于计算机视觉，人工智能领域，尤其涉及一种基于深度学习的脊椎MRI影像关键点检测方法。
背景技术
人工智能技术在最近几年在医疗领域中有着极其广泛的应用，其中计算机视觉在医疗影像分析上存在巨大的应用潜力。本发明针对的脊椎MRI影像关键点检测，以往的脊椎MRI影像的关键点检测工作多依赖于专家手动标注。手动标注效率低下，受专家主观影响较大，尤其不适用于大规模数据处理分析的情况。目前尝试利用人工智能技术检测的方法多是利用图像底层特征，例如论文(Ebrahimi S,Angelini E,Gajny L,et al.Lumbar spine posterior corner detection in X-rays using Haar-based features[C]//2016 IEEE 13th international symposium on biomedical imaging(ISBI).IEEE,2016:180-183.)中利用Harr特征检测椎骨的角点，但此类方法只利用到了图像底层信息，鲁棒性较差，只适用于部分特定的场景下，不适用于复杂多变的医疗场景中。本发明则通过建立高质量的数据集和利用深度学习的优秀学习能力实现更加鲁棒准确的脊椎MRI影像关键点检测方法。
发明内容
本发明是通过以下技术方案来实现的：一种基于深度学习的脊椎MRI影像关键点检测方法，包括以下步骤：
步骤一：将脊椎MRI影像输入训练好的目标检测网络，获得各椎骨的位置信息以及是否是S1的粗粒度标签；
步骤二：利用步骤一得到的所有椎骨以及定位的S1位置，并结合脊椎自身的生理结构信息过滤假阳性的检测结果和识别各椎骨所属的细粒度标签。
步骤三：将步骤二中检测得到的椎骨及其周围部分区域裁剪出，输入到训练好的关键点检测网络检测各块椎骨上下边界UA、UM、UP、LA、LM、LP共计六个关键点的位置信息。
步骤四：利用训练的分割网络对步骤二中得到的椎骨进行分割以获得边缘信息，并根据步骤四获得的边缘信息对步骤三中所获得的关键点的位置信息进行修正，获得最终关键点预测结果。
进一步地，所述步骤一中，椎骨的粗粒度标签是S1和NS1(其中S1是指骶1，NS1是除骶1以外的所有其他椎骨)，所述目标检测网络为YOLOv3。
进一步地，所述步骤三通过以下子步骤来实现。
1)采用检测出的S1作为定位的椎骨，计算各个椎骨中心在图像上的高度并根据形心高度对检测出的椎骨进行排序。
2)按照人体脊椎的生理结构信息，自下而上依次为检测出的每块椎骨分配对应的S1，L5，L4，L3，L2，L1，T12，T11等细粒度标签。
3)通过计算椎骨的宽高比以及上侧边缘高度并根据是否满足阈值要求过滤此假阳性目标。
进一步地，宽高比阈值是1.6，上侧边缘高度阈值是5。
进一步地，所述步骤四具体为：
(4.1)构建椎骨边缘分割网络：椎骨边缘分割网络由降采样部分和升采样部分构成。降采样部分结构为去掉全连接层的resnet50，升采样部分由对应的四个阶段的升采样卷积块构成，升采样卷积块结构为upsampling->conv->bn->relu。
(4.2)训练椎骨边缘分割网络：先利用关键点标注信息建立粗粒度的分割数据集对椎骨边缘分割网络进行预训练，再构建一个精确的细粒度分割数据集对分割网络进一步进行训练。
(4.3)在获得分割结果后，利用CRF及边缘处图像梯度较大的特点对分割结果进行进一步的修正以获得更精确的边缘分割信息。
进一步地，所述步骤(4.3)具体为：
(4.3.1)作两关键点连线的延长线，利用椎骨边缘分割网络获得椎骨边缘信息，将延长线与椎骨边缘的最远处交点作为修正后的关键点坐标。
(4.3.2)结合步骤二获得标签得到最终关键点预测结果；
本发明的有益效果是，本发明利用深度学习技术实现了脊椎MRI影像的关键点检测，避免了繁琐的人工标注，减轻了医生负担。相比于专家人工标注，本发明能够避免医生主观因素的影响，也能批量化处理大规模的数据，为进一步脊椎MRI影像分析提供数据基础，本发明方法可开发成交互式的可视化MRI脊椎影像关键点自动标注软件，获得的关键点检测结果可计算椎间盘高度指数和腰椎前突角。
附图说明
图1是本发明中检测关键点的示意图。
图2是本发明整体流程图。
图3是本发明关键点矫正示意图。
具体实施方式
下面根据附图详细说明本发明。
参阅图1，本发明中训练网络的数据集均为自建数据集，脊椎关键点均由医生专家标注， 检测的关键点是各块椎骨的上下边界UA、UM、UP、LA、LM、LP共计六个关键点。模型训练基本流程如下：
1.搜集脊椎MRI影像并随机抽取一部分影像作为初始数据集。
2.对数据集进行标注或修正，并利用标注好的数据集训练模型。
3.利用训练好的模型预测新采集到的脊椎MRI影像，并添加到数据集中。
4.重复步骤2，3直至模型准确率满足使用要求。
图2中描述的目标检测网络优选为YOLOv3，本方法在训练过程中将椎骨按是否是S1将椎骨分为S1和NS1(骶1和非骶1)两个类别作为粗粒度标签信息。训练检测网络的数据集则从原始数据集中构建，根据各个椎骨的关键点标注计算椎骨的包围框以及粗粒度类别标签训练YOLOv3网络。
当获得脊椎MRI图像中的椎骨的位置以及是否是S1的粗粒度类别信息后，本方法利用脊椎的结构信息过滤假阳性的预测结果并确定每块椎骨所属的类别，具体做法是：采用S1作为定位的椎骨，计算各个椎骨中心在图像上的高度并根据形心高度对检测出的椎骨进行排序。随后按照人体脊椎的生理结构信息，自下而上依次为检测出的每块椎骨分配对应的S1，L5，L4，L3，L2，L1，T12，T11等标签。其中图片最上方可能出现椎骨未被完整拍摄到的情况，本方法通过计算椎骨的宽高比以及中心高度是否满足阈值要求过滤此类目标，椎骨的宽高比可根据目标检测结果进行计算。
图2中描述的关键点检测网络优选为U型网络或堆叠沙漏网络(Stacked Hourglass Network)，训练数据为从原始图像中裁剪出的椎骨图像及周围部分图像和关键点对应的热力图，在关键点检测网络的训练过程中利用了在线难样本挖掘方法。
原始数据标注过程中存在一定的误差，该误差会导致标注的关键点位置脊椎边缘有一定的误差。本方法利用脊椎的边缘信息对检测出的关键点结果作进一步修正。首先本方法训练出一个U型深度卷积神经网络作为椎骨边缘分割网络用于分割椎骨边缘，该椎骨边缘分割网络由降采样部分和升采样部分构成。降采样部分结构为去掉全连接层的Resnet50，升采样部分由对应的四个阶段的升采样卷积块构成，升采样卷积块结构为upsampling->conv->bn->relu。为提高椎骨边缘分割网络的精度，本方法先利用关键点标注信息建立粗粒度的分割数据集对椎骨边缘分割网络进行预训练，再构建一个精确的细粒度分割数据集对分割网络进一步进行训练。本方法还利用条件随机场CRF及边缘处图像梯度较大的特点对分割结果进行进一步的修正以获得更精确的边缘分割信息。在利用椎骨边缘分割网络获得椎骨边缘信息后，本方法利用边缘信息对检测出的关键点进行修正，如图3所示，本方法沿UA-LA、UM-LM、UP-LP连线向椎骨边缘作延长线，将延长线与椎骨边缘的最远处交点作为修正后的UA、LA、UM、 LM、UP、LP坐标。为提高数据集制作流程效率和方便医疗人员使用，将上述方法开发成交互式的可视化自动检测标注软件。
网络训练完成之后即可应用于整个脊椎MRI影像关键点检测流程中，按照本发明阐述的技术方案，本发明需要在针对脊椎MRI影像关键点检测流程完成以下步骤：
步骤一：将脊椎MRI影像输入训练好的目标检测网络YOLOv3(Redmon J,Farhadi A.Yolov3:An incremental improvement[J].arXiv preprint arXiv:1804.02767,2018.)，获得各椎骨的位置信息以及是否是S1的粗粒度标签；
步骤二：利用步骤一得到的所有椎骨以及定位的S1位置，并结合脊椎自身的生理结构信息过滤假阳性的检测结果和识别各椎骨所属的细粒度标签。
1)采用检测出的S1作为定位的椎骨，计算各个椎骨中心在图像上的高度并根据形心高度对检测出的椎骨进行排序。
2)按照人体脊椎的生理结构信息，自下而上依次为检测出的每块椎骨分配对应的S1，L5，L4，L3，L2，L1，T12，T11等细粒度标签。
3)通过计算椎骨的宽高比以及上侧边缘高度并根据是否满足阈值要求过滤此假阳性目标。宽高比阈值是1.6，上侧边缘高度阈值是5。
步骤三：将步骤二中检测得到的椎骨及其周围部分区域裁剪出，输入到训练好的关键点检测网络检测各块椎骨上下边界UA、UM、UP、LA、LM、LP共计六个关键点的热力图。提取热力图中关键点的坐标可以采用但不限于论文(Zhang F,Zhu X,Dai H,et al.Distribution-aware coordinate representation for human pose estimation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2020:7093-7102.)中的方法。
步骤四：利用训练的分割网络对步骤二中得到的椎骨进行分割以获得边缘信息，并根据步骤四获得的边缘信息对步骤三中所获得的关键点的位置信息进行修正。
1)利用椎骨边缘分割网络获得椎骨边缘信息；
2)本方法沿UA-LA、UM-LM、UP-LP连线向椎骨边缘作延长线，将延长线与椎骨边缘的最远处交点作为修正后的UA、LA、UM,、LM，UP、LP坐标。
3)结合步骤二得到的标签输出最终关键点预测结果。
基于上述流程开发交互式的可视化MRI脊椎影像关键点自动标注软件，并基于关键点检测结果计算椎间盘高度指数和腰椎前突角。
以上所述为本发明的主要内容，凡是利用本发明说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本发明的专利保护范围内。

Claims (4)

  1. 一种基于深度学习的脊椎MRI影像关键点检测方法，其特征在于，包括以下步骤：
    步骤一：将脊椎MRI影像输入训练好的目标检测网络，获得各椎骨的位置信息以及是否是S1的粗粒度标签；
    步骤二：利用步骤一得到的所有椎骨以及定位的S1位置，结合脊椎自身的生理结构信息过滤假阳性的检测结果并识别各椎骨所属的细粒度标签；
    步骤三：将步骤二中检测得到的椎骨及其周围部分区域裁剪出，输入到训练好的关键点检测网络检测各块椎骨上下边界UA、UM、UP、LA、LM、LP共计六个关键点的位置信息；
    步骤四：利用训练的分割网络对步骤二中得到的椎骨进行分割以获得边缘信息，并根据步骤四获得的边缘信息对步骤三中所获得的关键点的位置信息进行修正，获得最终关键点预测结果，具体为：
    (4.1)构建椎骨边缘分割网络：椎骨边缘分割网络由降采样部分和升采样部分构成；降采样部分结构为去掉全连接层的resnet50，升采样部分由对应的四个阶段的升采样卷积块构成，升采样卷积块结构为upsampling->conv->bn->relu；
    (4.2)训练椎骨边缘分割网络：先利用关键点标注信息建立粗粒度的分割数据集对椎骨边缘分割网络进行预训练，再构建一个精确的细粒度分割数据集对分割网络进一步进行训练；
    (4.3)在获得分割结果后，利用条件随机场及边缘处图像梯度较大的特点对分割结果进行进一步的修正以获得更精确的边缘分割信息，具体为：
    (4.3.1)作两关键点连线的延长线，利用椎骨边缘分割网络获得椎骨边缘信息，将延长线与椎骨边缘的最远处交点作为修正后的关键点坐标；
    (4.3.2)结合步骤二得到的标签输出最终关键点预测结果。
  2. 根据权利要求1所述基于深度学习的脊椎MRI影像关键点检测方法，其特征在于，所述步骤一中，椎骨的粗粒度标签是S1和NS1，其中S1是指骶1，NS1是除骶1以外的所有其他椎骨，所述目标检测网络为YOLOv3。
  3. 根据权利要求1所述基于深度学习的脊椎MRI影像关键点检测方法，其特征在于，所述步骤二通过以下子步骤来实现；
    (2.1)采用检测出的S1作为定位的椎骨，计算各个椎骨中心在图像上的高度并根据形心高度对检测出的椎骨进行排序；
    (2.2)按照人体脊椎的生理结构信息，自下而上依次为检测出的每块椎骨分配对应的S1，L5，L4，L3，L2，L1，T12，T11细粒度标签；
    (2.3)通过计算椎骨的宽高比以及上侧边缘高度并根据是否满足阈值要求过滤假阳性目标。
  4. 根据权利要求3所述基于深度学习的脊椎MRI影像关键点检测方法，其特征在于，宽高比阈值是1.6，上侧边缘高度阈值是5。
PCT/CN2021/112874 2020-08-17 2021-08-16 一种基于深度学习的脊椎mri影像关键点检测方法 Ceased WO2022037548A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2022578644A JP7489732B2 (ja) 2020-08-17 2021-08-16 深層学習に基づく脊椎ｍｒｉ映像キーポイント検出方法

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010824727.7A CN112184617B (zh) 2020-08-17 2020-08-17 一种基于深度学习的脊椎mri影像关键点检测方法
CN202010824727.7 2020-08-17

Publications (1)

Publication Number Publication Date
WO2022037548A1 true WO2022037548A1 (zh) 2022-02-24

Family

ID=73919631

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/112874 Ceased WO2022037548A1 (zh) 2020-08-17 2021-08-16 一种基于深度学习的脊椎mri影像关键点检测方法

Country Status (3)

Country Link
JP (1) JP7489732B2 (zh)
CN (1) CN112184617B (zh)
WO (1) WO2022037548A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114881930A (zh) * 2022-04-07 2022-08-09 重庆大学 基于降维定位的3d目标检测方法、装置、设备和存储介质
US20230169644A1 (en) * 2021-11-30 2023-06-01 Pong Yuen Holdings Limited Computer vision system and method for assessing orthopedic spine condition
CN116309591A (zh) * 2023-05-19 2023-06-23 杭州健培科技有限公司 一种医学影像3d关键点检测方法、模型训练方法及装置
CN116721266A (zh) * 2023-06-20 2023-09-08 天津农学院 基于HRNet网络的单分蘖水稻植株表型参数提取方法
CN117011522A (zh) * 2023-04-17 2023-11-07 安徽大学 基于深度对比学习的胎儿脑mri分割方法
CN117474906A (zh) * 2023-12-26 2024-01-30 合肥吉麦智能装备有限公司 脊柱x光图像匹配方法及术中x光机复位方法
CN120163825A (zh) * 2025-05-20 2025-06-17 青岛大学 一种颈椎后纵韧带骨化症自动定位及分型方法

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112184617B (zh) * 2020-08-17 2022-09-16 浙江大学 一种基于深度学习的脊椎mri影像关键点检测方法
CN112700448B (zh) * 2021-03-24 2021-06-08 成都成电金盘健康数据技术有限公司 一种脊椎图像分割与识别方法
CN113392872A (zh) * 2021-04-30 2021-09-14 上海市第六人民医院 一种基于人工智能辅助的椎体骨折阅片方法及系统
CN113963251A (zh) * 2021-11-26 2022-01-21 山东省计算中心（国家超级计算济南中心） 一种海洋生物检测方法、系统和设备
CN114549396B (zh) * 2021-12-28 2024-11-12 河海大学 一种基于图神经网络的脊椎交互式与自动分割细化方法
CN114463298B (zh) * 2022-01-25 2025-03-28 浙江大学 一种x光脊椎图像关键点检测与识别方法
CN114494192B (zh) * 2022-01-26 2023-04-25 西南交通大学 一种基于深度学习的胸腰椎骨折识别分割与检测定位方法
CN114581395A (zh) * 2022-02-28 2022-06-03 四川大学 一种基于深度学习的脊椎医学图像的关键点检测方法
CN114913257A (zh) * 2022-04-29 2022-08-16 深圳市纳诺艾医疗科技有限公司 医用脊椎图像的生成方法、装置、计算机设备及存储介质
CN116309455B (zh) * 2023-03-16 2026-03-17 山东大学 腰椎异常图像识别方法及系统
CN116797597B (zh) * 2023-08-21 2023-11-17 邦世科技(南京)有限公司 一种基于三阶段全网络的脊柱退行性疾病全检测方法及系统
CN117137439A (zh) * 2023-08-22 2023-12-01 联影智能医疗科技(北京)有限公司 一种容积骨密度的测量方法、装置及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106780520A (zh) * 2015-11-18 2017-05-31 周兴祥 一种mri腰椎图像中椎骨的自动提取方法
CN110415291A (zh) * 2019-08-07 2019-11-05 清华大学 图像处理方法及相关设备
US20190370957A1 (en) * 2018-05-31 2019-12-05 General Electric Company Methods and systems for labeling whole spine image using deep neural network
CN111402269A (zh) * 2020-03-17 2020-07-10 东北大学 一种基于改进的FC-DenseNets的椎管分割方法
CN112184617A (zh) * 2020-08-17 2021-01-05 浙江大学 一种基于深度学习的脊椎mri影像关键点检测方法

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP5620668B2 (ja) 2009-10-26 2014-11-05 学校法人北里研究所 椎間板変性の評価装置及びプログラム
CN103479380B (zh) * 2013-09-17 2015-05-13 飞利浦(中国)投资有限公司 椎骨和椎间盘形状位置参数的识别方法
CN110072436B (zh) * 2016-12-08 2022-12-13 皇家飞利浦有限公司 脊柱医学成像数据的简化导航
JP7120560B2 (ja) 2017-07-03 2022-08-17 株式会社リコー 診断支援システム、診断支援方法及び診断支援プログラム
JP7135473B2 (ja) 2018-01-31 2022-09-13 株式会社リコー 医用画像処理装置、医用画像処理方法、プログラム及び医用画像処理システム
CN109523523B (zh) * 2018-11-01 2020-05-05 郑宇铄 基于fcn神经网络和对抗学习的椎体定位识别分割方法
CN109919903B (zh) * 2018-12-28 2020-08-07 上海联影智能医疗科技有限公司 一种脊椎检测定位标记方法、系统及电子设备
CN110599508B (zh) * 2019-08-01 2023-10-27 平安科技（深圳）有限公司 基于人工智能的脊柱影像处理方法及相关设备
CN110866921A (zh) * 2019-10-17 2020-03-06 上海交通大学 基于自训练和切片传播的弱监督脊椎椎体分割方法和系统

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106780520A (zh) * 2015-11-18 2017-05-31 周兴祥 一种mri腰椎图像中椎骨的自动提取方法
US20190370957A1 (en) * 2018-05-31 2019-12-05 General Electric Company Methods and systems for labeling whole spine image using deep neural network
CN110415291A (zh) * 2019-08-07 2019-11-05 清华大学 图像处理方法及相关设备
CN111402269A (zh) * 2020-03-17 2020-07-10 东北大学 一种基于改进的FC-DenseNets的椎管分割方法
CN112184617A (zh) * 2020-08-17 2021-01-05 浙江大学 一种基于深度学习的脊椎mri影像关键点检测方法

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20230169644A1 (en) * 2021-11-30 2023-06-01 Pong Yuen Holdings Limited Computer vision system and method for assessing orthopedic spine condition
CN114881930A (zh) * 2022-04-07 2022-08-09 重庆大学 基于降维定位的3d目标检测方法、装置、设备和存储介质
CN114881930B (zh) * 2022-04-07 2023-08-18 重庆大学 基于降维定位的3d目标检测方法、装置、设备和存储介质
CN117011522A (zh) * 2023-04-17 2023-11-07 安徽大学 基于深度对比学习的胎儿脑mri分割方法
CN116309591A (zh) * 2023-05-19 2023-06-23 杭州健培科技有限公司 一种医学影像3d关键点检测方法、模型训练方法及装置
CN116309591B (zh) * 2023-05-19 2023-08-25 杭州健培科技有限公司 一种医学影像3d关键点检测方法、模型训练方法及装置
CN116721266A (zh) * 2023-06-20 2023-09-08 天津农学院 基于HRNet网络的单分蘖水稻植株表型参数提取方法
CN117474906A (zh) * 2023-12-26 2024-01-30 合肥吉麦智能装备有限公司 脊柱x光图像匹配方法及术中x光机复位方法
CN117474906B (zh) * 2023-12-26 2024-03-26 合肥吉麦智能装备有限公司 基于脊柱x光图像匹配的术中x光机复位方法
CN120163825A (zh) * 2025-05-20 2025-06-17 青岛大学 一种颈椎后纵韧带骨化症自动定位及分型方法

Also Published As

Publication number Publication date
JP7489732B2 (ja) 2024-05-24
CN112184617B (zh) 2022-09-16
CN112184617A (zh) 2021-01-05
JP2023530023A (ja) 2023-07-12

Similar Documents

Publication Publication Date Title
JP7489732B2 (ja) 深層学習に基づく脊椎ｍｒｉ映像キーポイント検出方法
CN112734757B (zh) 一种脊柱X光图像cobb角测量方法
CN114494192B (zh) 一种基于深度学习的胸腰椎骨折识别分割与检测定位方法
US20220198214A1 (en) Image recognition method and device based on deep convolutional neural network
CN115115567B (zh) 一种图像处理方法、装置、计算机设备及介质
CN113420826B (zh) 一种肝脏病灶图像处理系统及图像处理方法
Hogeweg et al. Clavicle segmentation in chest radiographs
CN111047572A (zh) 一种基于Mask RCNN的医学图像中脊柱自动定位方法
CN101403743B (zh) 一种x型交叠、粘连染色体自动分割方法
CN112802019B (zh) 一种基于脊柱AIS影像的lenke分型方法
CN113469942B (zh) 一种ct图像病变检测方法
CN113362282B (zh) 基于多任务学习的髋关节关键点位置检测方法及系统
CN112241961B (zh) 基于深度卷积神经网络的胸部x光片辅助诊断方法及系统
Zhang et al. Automatic Lenke classification of adolescent idiopathic scoliosis with deep learning
Tran et al. MBNet: A multi-task deep neural network for semantic segmentation and lumbar vertebra inspection on X-ray images
CN119941731B (zh) 基于大模型的肺结节分析方法、系统、设备及介质
CN119693362B (zh) 脊柱Cobb角的测量方法、装置以及电子设备
CN105512688A (zh) 基于ct定位片的高性能椎骨检测与分割方法
WO2020215485A1 (zh) 胎儿生长参数测量方法、系统及超声设备
CN121839045A (zh) 一种婴幼儿发育性髋关节发育不良的超声图像处理方法及系统
CN120374990B (zh) 基于类别嵌入分离对比学习网络的乳腺超声图像分割方法
CN116128754A (zh) 颈椎分割矫正、曲度检测方法、系统、电子设备及介质
Zhu et al. Automatic measurement of scoliosis based on an improved segmentation model
CN120782706A (zh) 脊椎图像识别方法及相关设备
CN118982699A (zh) 一种基于图像特征提取的脊柱侧凸患者识别分型系统

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: 21857642

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022578644

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21857642

Country of ref document: EP

Kind code of ref document: A1