CN112184617A - Spine MRI image key point detection method based on deep learning - Google Patents
Spine MRI image key point detection method based on deep learning Download PDFInfo
- Publication number
- CN112184617A CN112184617A CN202010824727.7A CN202010824727A CN112184617A CN 112184617 A CN112184617 A CN 112184617A CN 202010824727 A CN202010824727 A CN 202010824727A CN 112184617 A CN112184617 A CN 112184617A
- Authority
- CN
- China
- Prior art keywords
- vertebrae
- key point
- vertebra
- mri image
- edge
- 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.)
- Granted
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
- 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; CALCULATING OR 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; CALCULATING OR 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; CALCULATING OR 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; 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
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- 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]
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30008—Bone
- G06T2207/30012—Spine; Backbone
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)
- Apparatus For Radiation Diagnosis (AREA)
- Image Analysis (AREA)
Abstract
The invention is called a spine MRI image key point detection method based on deep learning. The invention discloses a spine MRI image key point detection method based on deep learning, which comprises the steps of firstly detecting and positioning vertebrae in a spine MRI image by utilizing a deep target detection network, identifying S1 (sacrum 1) as positioned vertebrae, and then filtering false positive detection results and judging fine grain labels of all the vertebrae by combining with the structural information of the vertebrae. And then, respectively detecting six key points of the upper and lower boundaries UA, UM, UP, LA, LM and LP of each vertebra by using a key point detection network, determining and correcting the key point positions of each vertebra by combining edge information, and finally developing interactive visual MRI spine image key point automatic labeling software. The method can automatically extract the key points of the spine MRI image, and has great application value in the aspects of medical image analysis, auxiliary medical treatment and the like.
Description
Technical Field
The invention belongs to the field of computer vision and artificial intelligence, and particularly relates to a spine MRI image key point detection method based on deep learning.
Background
Artificial intelligence technology has been used in the medical field in recent years, where computer vision has a great potential for medical image analysis. Aiming at the detection of the key points of the spine MRI image, the detection of the key points of the prior spine MRI image mostly depends on the manual labeling of experts. The manual labeling efficiency is low, the subjective influence of experts is large, and the manual labeling method is particularly not suitable for the condition of large-scale data processing and analysis. Most of the methods for detecting by using artificial intelligence technology at present use image bottom layer features, for example Harr features in papers (Ebrahimi S, Angelini E, Gajny L, et al, lumbar spine spatial coder detection in X-ray using Haar-based features [ C ]//2016IEEE 13th international symposium on biological imaging (ISBI), 2016: 180-. The invention realizes a more robust and accurate spine MRI image key point detection method by establishing a high-quality data set and utilizing the excellent learning ability of deep learning.
Disclosure of Invention
The invention is realized by the following technical scheme: a spine MRI image key point detection method based on deep learning comprises the following steps:
the method comprises the following steps: inputting the spine MRI image into a trained target detection network to obtain the position information of each vertebra and whether the position information is a coarse-grained label of S1;
step two: and D, filtering false positive detection results and identifying fine-grained labels of the vertebrae by using all vertebrae obtained in the step one and the positioned S1 position and combining the physiological structure information of the vertebrae.
Step three: cutting out the vertebrae and the peripheral part area thereof obtained by detection in the step two, and inputting the cut vertebrae and the peripheral part area into a trained key point detection network to detect the position information of six key points which are counted by the upper and lower boundaries UA, UM, UP, LA, LM and LP of each vertebra.
Step four: and C, segmenting the vertebrae obtained in the step two by using the trained segmentation network to obtain edge information, and correcting the position information of the key points obtained in the step three according to the edge information obtained in the step four to obtain a final key point prediction result.
Further, in the first step, the coarse-grained labels of the vertebrae are S1 and NS1 (where S1 refers to sacral 1, and NS1 refers to all other vertebrae except sacral 1), and the target detection network is YOLOv 3.
Further, the third step is realized by the following substeps.
1) Using the detected S1 as the positioned vertebrae, the heights of the centers of the respective vertebrae on the image are calculated and the detected vertebrae are sorted according to the height of the centroid.
2) According to the physiological structure information of the human spine, corresponding fine-grained labels such as S1, L5, L4, L3, L2, L1, T12 and T11 are assigned to each detected vertebra from bottom to top.
3) This false positive target is filtered by calculating the aspect ratio of the vertebrae and the upper lateral margin height and depending on whether the threshold requirement is met.
Further, the aspect ratio threshold is 1.6 and the upper edge height threshold is 5.
Further, the fourth step is specifically:
(4.1) constructing a vertebral margin segmentation network: the vertebral margin segmentation network is comprised of a down-sampling portion and an up-sampling portion. The structure of the down-sampling part is resnet50 with a full connection layer removed, the up-sampling part is composed of up-sampling convolution blocks of corresponding four stages, and the structure of the up-sampling convolution blocks is upsampling- > conv- > bn- > relu.
(4.2) training a vertebral edge segmentation network: firstly, establishing a coarse-grained segmentation data set by using the key point marking information to pre-train the vertebra edge segmentation network, and then establishing an accurate fine-grained segmentation data set to further train the segmentation network.
And (4.3) after the segmentation result is obtained, further correcting the segmentation result by utilizing the characteristics of the CRF and the larger gradient of the edge image so as to obtain more accurate edge segmentation information.
Further, the step (4.3) is specifically:
(4.3.1) making an extension line of the connection line of the two key points, obtaining vertebra edge information by utilizing the vertebra edge segmentation network, and taking the farthest intersection point of the extension line and the vertebra edge as the coordinate of the corrected key point.
(4.3.2) obtaining a final key point prediction result by combining the label obtained in the second step;
the method has the advantages that the method realizes the key point detection of the spine MRI image by utilizing the deep learning technology, avoids complicated manual labeling and reduces the burden of doctors. Compared with the manual labeling of experts, the method can avoid the influence of subjective factors of doctors, can process large-scale data in batch, and provides data basis for further spine MRI image analysis.
Drawings
FIG. 1 is a schematic diagram of key points detected in the present invention.
Fig. 2 is an overall flow chart of the present invention.
Fig. 3 is a schematic diagram of key point correction according to the present invention.
Detailed Description
The present invention is described in detail below with reference to the accompanying drawings.
Referring to fig. 1, the data sets of the training network in the invention are self-built data sets, the key points of the spine are marked by doctors and experts, and the detected key points are six key points in total of the upper and lower boundaries UA, UM, UP, LA, LM and LP of each vertebra. The basic flow of model training is as follows:
1. spinal MRI images are collected and a portion of the images are randomly extracted as an initial data set.
2. And marking or correcting the data set, and training the model by using the marked data set.
3. And predicting the newly acquired spine MRI image by using the trained model, and adding the newly acquired spine MRI image into the data set.
4. And (5) repeating the steps 2 and 3 until the model accuracy meets the use requirement.
The target detection network depicted in fig. 2 is preferably YOLOv3, and the method classifies the vertebrae into two categories of S1 and NS1 (sacral 1 and non-sacral 1) according to whether the vertebrae are S1 or not as coarse-grained label information in the training process. And constructing a data set of the training detection network from the original data set, calculating a bounding box of each vertebra according to the key point labels of the vertebra, and training the YOLOv3 network by using the coarse-grained class labels.
After obtaining the position of the vertebra in the MRI image of the vertebra and whether the coarse-grained category information is S1, the method filters the prediction result of false positive and determines the category to which each vertebra belongs by using the structural information of the vertebra, specifically: using S1 as the positioned vertebrae, the heights of the centers of the respective vertebrae on the image are calculated and the detected vertebrae are sorted according to the height of the centroid. And then, according to the physiological structure information of the human spine, corresponding labels such as S1, L5, L4, L3, L2, L1, T12 and T11 are allocated to each detected vertebra from bottom to top. The method filters the targets by calculating the aspect ratio of the vertebrae and whether the central height meets the threshold requirement, and the aspect ratio of the vertebrae can be calculated according to the target detection result.
The keypoint detection Network depicted in fig. 2 is preferably a U-shaped Network or a Stacked Hourglass Network (Stacked Hourglass Network), the training data is a thermodynamic diagram corresponding to the vertebra image and the surrounding partial images cut out from the original image and the keypoints, and an online hard sample mining method is utilized in the training process of the keypoint detection Network.
Certain errors exist in the original data labeling process, and the errors can cause certain errors of the spine edges at the positions of the labeled key points. The method utilizes the edge information of the spine to further correct the detected key point result. Firstly, training a U-shaped deep convolutional neural network as a vertebra edge segmentation network for segmenting vertebra edges, wherein the vertebra edge segmentation network is composed of a down-sampling part and an up-sampling part. The structure of the down-sampling part is Resnet50 with a full connection layer removed, the up-sampling part is composed of up-sampling convolution blocks of corresponding four stages, and the structure of the up-sampling convolution blocks is upsampling- > conv- > bn- > relu. In order to improve the precision of the vertebra edge segmentation network, the method firstly establishes a coarse-grained segmentation data set by using the key point marking information to pre-train the vertebra edge segmentation network, and then establishes an accurate fine-grained segmentation data set to further train the segmentation network. The method further corrects the segmentation result by using the characteristic of the conditional random field CRF and the larger gradient of the edge image to obtain more accurate edge segmentation information. After the vertebra edge information is obtained by using the vertebra edge segmentation network, the method corrects the detected key points by using the edge information, and as shown in fig. 3, the method takes the extension line to the vertebra edge along the connection line of UA-LA, UM-LM and UP-LP, and takes the farthest intersection point of the extension line and the vertebra edge as the corrected coordinates of UA, LA, UM, LM, UP and LP. In order to improve the efficiency of the data set manufacturing process and facilitate the use of medical personnel, the method is developed into interactive visual automatic detection labeling software.
After the network training is completed, the method can be applied to the whole spine MRI image key point detection process, and according to the technical scheme set forth by the invention, the method needs to complete the following steps aiming at the spine MRI image key point detection process:
the method comprises the following steps: inputting the MRI image of the spine into a trained target detection network YOLOv3(Redmon J, Farhadi A. Yolov3: An unknown actual [ J ]. arXiv preprint arXiv:1804.02767,2018.), and obtaining the position information of each vertebra and whether the position information is a coarse-grained label of S1;
step two: and D, filtering false positive detection results and identifying fine-grained labels of the vertebrae by using all vertebrae obtained in the step one and the positioned S1 position and combining the physiological structure information of the vertebrae.
1) Using the detected S1 as the positioned vertebrae, the heights of the centers of the respective vertebrae on the image are calculated and the detected vertebrae are sorted according to the height of the centroid.
2) According to the physiological structure information of the human spine, corresponding fine-grained labels such as S1, L5, L4, L3, L2, L1, T12 and T11 are assigned to each detected vertebra from bottom to top.
3) This false positive target is filtered by calculating the aspect ratio of the vertebrae and the upper lateral margin height and depending on whether the threshold requirement is met. The aspect ratio threshold is 1.6 and the upper edge height threshold is 5.
Step three: cutting out the vertebrae and the peripheral part area thereof obtained by detection in the step two, and inputting the cut vertebrae and the peripheral part area into a trained key point detection network to detect the thermodynamic diagram of six key points of the upper and lower boundaries UA, UM, UP, LA, LM and LP of each vertebra. The method of extracting the coordinates of the key points in the thermodynamic diagram can be adopted by, but not limited to, the methods in the papers (Zhang F, Zhu X, Dai H, et al. distribution-aware correlation representation for human position estimation [ C ]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern registration. 2020:7093 and 7102).
Step four: and C, segmenting the vertebrae obtained in the step two by using the trained segmentation network to obtain edge information, and correcting the position information of the key points obtained in the step three according to the edge information obtained in the step four.
1) Obtaining vertebra edge information by using a vertebra edge segmentation network;
2) the method takes an extension line to the edge of the vertebra along the connection line of UA-LA, UM-LM and UP-LP, and takes the farthest intersection point of the extension line and the edge of the vertebra as the coordinates of the UA, LA, UM, LM, UP and LP after correction.
3) And outputting a final key point prediction result by combining the label obtained in the step two.
And developing interactive visual MRI spine image key point automatic labeling software based on the process, and calculating the intervertebral disc height index and the lumbar vertebra anterior eminence angle based on the key point detection result.
The main contents of the present invention are described above, and all the equivalent structures or equivalent flow transformations made by the contents of the present specification and the attached drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (6)
1. A spine MRI image key point detection method based on deep learning is characterized by comprising the following steps:
the method comprises the following steps: inputting the spine MRI image into a trained target detection network to obtain the position information of each vertebra and whether the position information is a coarse-grained label of S1;
step two: and (4) filtering false positive detection results by using all vertebrae obtained in the step one and the positioned S1 position according to the physiological structure information of the vertebra, and identifying the fine-grained label of each vertebra.
Step three: cutting out the vertebrae and the peripheral part area thereof obtained by detection in the step two, and inputting the cut vertebrae and the peripheral part area into a trained key point detection network to detect the position information of six key points which are counted by the upper and lower boundaries UA, UM, UP, LA, LM and LP of each vertebra.
Step four: and C, segmenting the vertebrae obtained in the step two by using the trained segmentation network to obtain edge information, and correcting the position information of the key points obtained in the step three according to the edge information obtained in the step four to obtain a final key point prediction result.
2. The deep learning based spine MRI image keypoint detection method according to claim 1, wherein in the first step, the coarse-grained labels of the vertebrae are S1 and NS1 (where S1 refers to sacral 1, and NS1 refers to all other vertebrae except sacral 1), and the target detection network is YOLOv 3.
3. The deep learning-based spine MRI image key point detection method according to claim 1, wherein the second step is realized by the following substeps.
(2.1) using the detected S1 as the positioned vertebrae, calculating the heights of the centers of the respective vertebrae on the image and sorting the detected vertebrae according to the height of the centroid.
And (2.2) according to the physiological structure information of the human spine, allocating corresponding fine-grained labels such as S1, L5, L4, L3, L2, L1, T12 and T11 to each detected vertebra from bottom to top.
(2.3) filtering false positive objects by calculating the aspect ratio and upper lateral margin height of the vertebrae and depending on whether the threshold requirement is met.
4. The deep learning based spine MRI image key point detection method according to claim 3, wherein the aspect ratio threshold is 1.6 and the upper side edge height threshold is 5.
5. The deep learning-based spine MRI image key point detection method according to claim 1, wherein the fourth step is specifically:
(4.1) constructing a vertebral margin segmentation network: the vertebral margin segmentation network is comprised of a down-sampling portion and an up-sampling portion. The structure of the down-sampling part is resnet50 with a full connection layer removed, the up-sampling part is composed of up-sampling convolution blocks of corresponding four stages, and the structure of the up-sampling convolution blocks is upsampling- > conv- > bn- > relu.
(4.2) training a vertebral edge segmentation network: firstly, establishing a coarse-grained segmentation data set by using the key point marking information to pre-train the vertebra edge segmentation network, and then establishing an accurate fine-grained segmentation data set to further train the segmentation network.
And (4.3) after the segmentation result is obtained, further correcting the segmentation result by using the conditional random field and the characteristic of larger image gradient at the edge to obtain more accurate edge segmentation information.
6. The deep learning-based spine MRI image key point detection method according to claim 5, wherein the step (4.3) is specifically:
(4.3.1) making an extension line of the connection line of the two key points, obtaining vertebra edge information by utilizing the vertebra edge segmentation network, and taking the farthest intersection point of the extension line and the vertebra edge as the coordinate of the corrected key point.
And (4.3.2) outputting a final key point prediction result by combining the label obtained in the step two.
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010824727.7A CN112184617B (en) | 2020-08-17 | 2020-08-17 | Spine MRI image key point detection method based on deep learning |
| PCT/CN2021/112874 WO2022037548A1 (en) | 2020-08-17 | 2021-08-16 | Mri spinal image keypoint detection method based on deep learning |
| JP2022578644A JP7489732B2 (en) | 2020-08-17 | 2021-08-16 | Method for detecting key points in spinal MRI images based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010824727.7A CN112184617B (en) | 2020-08-17 | 2020-08-17 | Spine MRI image key point detection method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112184617A true CN112184617A (en) | 2021-01-05 |
| CN112184617B CN112184617B (en) | 2022-09-16 |
Family
ID=73919631
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010824727.7A Active CN112184617B (en) | 2020-08-17 | 2020-08-17 | Spine MRI image key point detection method based on deep learning |
Country Status (3)
| Country | Link |
|---|---|
| JP (1) | JP7489732B2 (en) |
| CN (1) | CN112184617B (en) |
| WO (1) | WO2022037548A1 (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112700448A (en) * | 2021-03-24 | 2021-04-23 | 成都成电金盘健康数据技术有限公司 | Spine image segmentation and identification method |
| WO2022037548A1 (en) * | 2020-08-17 | 2022-02-24 | 浙江大学 | Mri spinal image keypoint detection method based on deep learning |
| CN114494192A (en) * | 2022-01-26 | 2022-05-13 | 西南交通大学 | Deep learning-based thoracolumbar fracture identification, segmentation, detection and positioning method |
| CN114581395A (en) * | 2022-02-28 | 2022-06-03 | 四川大学 | Method for detecting key points of spine medical image based on deep learning |
| CN116797597A (en) * | 2023-08-21 | 2023-09-22 | 邦世科技(南京)有限公司 | Three-stage full-network-based full detection method and system for degenerative spinal diseases |
Families Citing this family (4)
| 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 |
| CN114881930B (en) * | 2022-04-07 | 2023-08-18 | 重庆大学 | 3D target detection method, device, equipment and storage medium based on dimension reduction positioning |
| CN116309591B (en) * | 2023-05-19 | 2023-08-25 | 杭州健培科技有限公司 | Medical image 3D key point detection method, model training method and device |
| CN117474906B (en) * | 2023-12-26 | 2024-03-26 | 合肥吉麦智能装备有限公司 | Intraoperative X-ray machine resetting method based on spine X-ray image matching |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160210740A1 (en) * | 2013-09-17 | 2016-07-21 | Koninklijke Philips N.V. | Method and system for spine position detection |
| CN109523523A (en) * | 2018-11-01 | 2019-03-26 | 郑宇铄 | Vertebra localization based on FCN neural network and confrontation study identifies dividing method |
| CN109919903A (en) * | 2018-12-28 | 2019-06-21 | 上海联影智能医疗科技有限公司 | A kind of vertebra detection positioning and marking method, system and electronic equipment |
| CN110415291A (en) * | 2019-08-07 | 2019-11-05 | 清华大学 | Image processing method and relevant device |
| CN110599508A (en) * | 2019-08-01 | 2019-12-20 | 平安科技（深圳）有限公司 | Spine image processing method based on artificial intelligence and related equipment |
| US20200022609A1 (en) * | 2016-12-08 | 2020-01-23 | Koninklijke Philips N.V. | Simplified navigation of spinal medical imaging data |
| CN110866921A (en) * | 2019-10-17 | 2020-03-06 | 上海交通大学 | Weakly supervised vertebral body segmentation method and system based on self-training and slice propagation |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5620668B2 (en) | 2009-10-26 | 2014-11-05 | 学校法人北里研究所 | Intervertebral disc degeneration evaluation apparatus and program |
| CN106780520B (en) * | 2015-11-18 | 2021-04-13 | 周兴祥 | Automatic extraction method of vertebrae in MRI (magnetic resonance imaging) lumbar vertebra image |
| JP7120560B2 (en) | 2017-07-03 | 2022-08-17 | 株式会社リコー | Diagnosis support system, diagnosis support method and diagnosis support program |
| JP7135473B2 (en) | 2018-01-31 | 2022-09-13 | 株式会社リコー | MEDICAL IMAGE PROCESSING APPARATUS, MEDICAL IMAGE PROCESSING METHOD, PROGRAM AND MEDICAL IMAGE PROCESSING SYSTEM |
| US10902587B2 (en) | 2018-05-31 | 2021-01-26 | GE Precision Healthcare LLC | Methods and systems for labeling whole spine image using deep neural network |
| CN111402269A (en) * | 2020-03-17 | 2020-07-10 | 东北大学 | Vertebral canal segmentation method based on improved FC-DenseNuts |
| CN112184617B (en) * | 2020-08-17 | 2022-09-16 | 浙江大学 | Spine MRI image key point detection method based on deep learning |
-
2020
- 2020-08-17 CN CN202010824727.7A patent/CN112184617B/en active Active
-
2021
- 2021-08-16 JP JP2022578644A patent/JP7489732B2/en active Active
- 2021-08-16 WO PCT/CN2021/112874 patent/WO2022037548A1/en active Application Filing
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160210740A1 (en) * | 2013-09-17 | 2016-07-21 | Koninklijke Philips N.V. | Method and system for spine position detection |
| US20200022609A1 (en) * | 2016-12-08 | 2020-01-23 | Koninklijke Philips N.V. | Simplified navigation of spinal medical imaging data |
| CN109523523A (en) * | 2018-11-01 | 2019-03-26 | 郑宇铄 | Vertebra localization based on FCN neural network and confrontation study identifies dividing method |
| CN109919903A (en) * | 2018-12-28 | 2019-06-21 | 上海联影智能医疗科技有限公司 | A kind of vertebra detection positioning and marking method, system and electronic equipment |
| CN110599508A (en) * | 2019-08-01 | 2019-12-20 | 平安科技（深圳）有限公司 | Spine image processing method based on artificial intelligence and related equipment |
| CN110415291A (en) * | 2019-08-07 | 2019-11-05 | 清华大学 | Image processing method and relevant device |
| CN110866921A (en) * | 2019-10-17 | 2020-03-06 | 上海交通大学 | Weakly supervised vertebral body segmentation method and system based on self-training and slice propagation |
Non-Patent Citations (3)
| Title |
|---|
| ANA JIMENEZ‑PASTOR 等: "Automated vertebrae localization and identification by decision forests and image‑based refinement on real‑world CT data", 《SPRINGER》 * |
| 叶伟 等: "一种新颖的MR脊柱图像自动分割算法研究", 《中国医疗设备》 * |
| 胡志军等: "椎间盘镜和椎间孔镜技术治疗腰椎间盘突出症的对比研究", 《知网数据库》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022037548A1 (en) * | 2020-08-17 | 2022-02-24 | 浙江大学 | Mri spinal image keypoint detection method based on deep learning |
| CN112700448A (en) * | 2021-03-24 | 2021-04-23 | 成都成电金盘健康数据技术有限公司 | Spine image segmentation and identification method |
| CN112700448B (en) * | 2021-03-24 | 2021-06-08 | 成都成电金盘健康数据技术有限公司 | Spine image segmentation and identification method |
| CN114494192A (en) * | 2022-01-26 | 2022-05-13 | 西南交通大学 | Deep learning-based thoracolumbar fracture identification, segmentation, detection and positioning method |
| CN114581395A (en) * | 2022-02-28 | 2022-06-03 | 四川大学 | Method for detecting key points of spine medical image based on deep learning |
| CN116797597A (en) * | 2023-08-21 | 2023-09-22 | 邦世科技(南京)有限公司 | Three-stage full-network-based full detection method and system for degenerative spinal diseases |
| CN116797597B (en) * | 2023-08-21 | 2023-11-17 | 邦世科技(南京)有限公司 | Three-stage full-network-based full detection method and system for degenerative spinal diseases |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112184617B (en) | 2022-09-16 |
| WO2022037548A1 (en) | 2022-02-24 |
| JP2023530023A (en) | 2023-07-12 |
| JP7489732B2 (en) | 2024-05-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112184617B (en) | Spine MRI image key point detection method based on deep learning | |
| CN111047572B (en) | Automatic spine positioning method in medical image based on Mask RCNN | |
| US10902587B2 (en) | Methods and systems for labeling whole spine image using deep neural network | |
| CN109345527B (en) | Bladder tumor detection method based on MaskRcnn | |
| CN112734757B (en) | Spine X-ray image cobb angle measuring method | |
| CN105760874A (en) | CT image processing system and method for pneumoconiosis | |
| CN102096804A (en) | Method for recognizing image of carcinoma bone metastasis in bone scan | |
| CN110448270B (en) | Artificial intelligence diagnosis and typing system for lumbar disc herniation | |
| Larhmam et al. | Semi-automatic detection of cervical vertebrae in X-ray images using generalized Hough transform | |
| CN115578372A (en) | Bone age assessment method, device and medium based on target detection and convolution transformation | |
| CN112802019A (en) | Leke typing method based on spine AIS image | |
| CN112837264B (en) | Rib positioning and fracture clinical outcome prediction device and automatic diagnosis system | |
| CN104915989A (en) | CT image-based blood vessel three-dimensional segmentation method | |
| CN109886320B (en) | Human femoral X-ray intelligent recognition method and system | |
| KR102570004B1 (en) | spine diagnosis system based on artificial neural network and information providing method therefor | |
| CN115252233B (en) | Automatic planning method for acetabular cup in total hip arthroplasty based on deep learning | |
| CN114862799B (en) | Full-automatic brain volume segmentation method for FLAIR-MRI sequence | |
| CN109697713B (en) | Intervertebral disc positioning and labeling method based on deep learning and spatial relationship reasoning | |
| CN109978861B (en) | Polio detection method, apparatus, device and computer readable storage medium | |
| Hahmann et al. | Epiphyses localization for bone age assessment using the discriminative generalized hough transform | |
| CN113762263A (en) | Semantic segmentation method and system for small-scale similar structure | |
| CN115019045B (en) | Small data thyroid ultrasound image segmentation method based on multi-component neighborhood | |
| Xue et al. | Pathology-based vertebral image retrieval | |
| Daniel et al. | Automatic vertebra detection in X-ray images | |
| CN111369580B (en) | Intelligent segmentation method and terminal for X-ray image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |