CN110288589B - Hematoma expansion prediction method and device - Google Patents
Hematoma expansion prediction method and device Download PDFInfo
- Publication number
- CN110288589B CN110288589B CN201910583573.4A CN201910583573A CN110288589B CN 110288589 B CN110288589 B CN 110288589B CN 201910583573 A CN201910583573 A CN 201910583573A CN 110288589 B CN110288589 B CN 110288589B
- Authority
- CN
- China
- Prior art keywords
- hematoma
- matching
- preset
- image
- sample
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/003—Reconstruction from projections, e.g. tomography
-
- 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
- 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
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30016—Brain
-
- 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
- G06T2207/30101—Blood vessel; Artery; Vein; Vascular
- G06T2207/30104—Vascular flow; Blood flow; Perfusion
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Quality & Reliability (AREA)
- Radiology & Medical Imaging (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The embodiment relates to the technical field of medical images, and provides a hematoma expansion prediction method and a hematoma expansion prediction device, wherein the method comprises the following steps: acquiring clinical examination data and a CT image containing a hematoma; inputting the CT image into a preset automatic segmentation model for segmentation to obtain a segmentation result, and calculating the volume of hematoma according to the segmentation result; forming a test sample by the segmentation result, the hematoma volume and the clinical examination data; and calculating the matching degree between each preset supporting sample in the plurality of supporting samples and the test sample, and predicting whether the hematoma is enlarged or not according to the plurality of matching degrees. Compared with the prior art, the hematoma expansion prediction method and device provided by the embodiment can predict whether the hematoma can be expanded in time so as to perform corresponding processing and guarantee the life safety of a patient.
Description
Technical Field
The invention relates to the technical field of medical images, in particular to a hematoma expansion prediction method and device.
Background
Cerebral Hemorrhage (ICH) refers to Intracerebral hemorrhage caused by rupture of blood vessels, and is mainly spontaneous non-traumatic cerebral hemorrhage, i.e., spontaneous cerebral hemorrhage, which is usually caused by hypertension, hyperglycemia, hyperlipidemia, smoking and other factors. The disease is sudden in onset, the disease is fierce, the treatment cost, the recurrence rate, the disability rate and the death rate are high, more than 40 percent of cerebral hemorrhage patients die within one month, 80 percent of surviving patients need to live depending on the nursing of other people, and the burden of ICH disease is reduced, so that the whole burden of China on medical service can be effectively reduced.
In the prior art, whether hematoma is enlarged is determined by comparing a follow-up Computed Tomography (CT for short) with a baseline CT. However, the relatively long time interval between the follow-up CT examination and the baseline CT examination indicates that the patient is likely to be deteriorated or even dead when the hematoma is judged to be enlarged in such a way, and the method for judging whether the hematoma is enlarged not timely can cause part of the patients to be deteriorated or even dead in the period.
Disclosure of Invention
The invention aims to provide a hematoma expansion prediction method and a hematoma expansion prediction device, which are used for solving the problem that in the prior art, the hematoma expansion is not determined timely, so that part of patients are deteriorated and even die in the period.
In order to achieve the above purpose, the embodiment of the present invention adopts the following technical solutions:
in a first aspect, an embodiment of the present invention provides a method for predicting hematoma expansion, where the method includes: acquiring clinical examination data and an electronic Computed Tomography (CT) image containing hematoma; inputting the CT image into a preset automatic segmentation model for segmentation to obtain a segmentation result, and calculating the hematoma volume according to the segmentation result; composing the segmentation result, the hematoma volume and the clinical examination data into a test sample; and calculating the matching degree between each support sample in a plurality of preset support samples and the test sample, and predicting whether the hematoma is enlarged or not according to the matching degrees.
In a second aspect, an embodiment of the present invention provides a hematoma expansion prediction apparatus, including: an acquisition module for acquiring clinical examination data and an electronic Computed Tomography (CT) image containing hematoma; the processing module is used for inputting the CT image into a preset automatic segmentation model for segmentation to obtain a segmentation result, and calculating the hematoma volume according to the segmentation result; composing the segmentation result, the hematoma volume and the clinical examination data into a test sample; and calculating the matching degree between each support sample in a plurality of preset support samples and the test sample, and predicting whether the hematoma is enlarged or not according to the matching degrees.
Compared with the prior art, the hematoma expansion prediction method and device provided by the embodiment of the invention have the advantages that the CT image containing the hematoma is input into the preset automatic segmentation model for segmentation to obtain the segmentation result, the hematoma volume is calculated according to the segmentation result, then the segmentation result, the hematoma volume and the clinical examination data form the test sample, the matching degree of each support sample in the preset plurality of support samples and the test sample is calculated, and whether the hematoma is expanded or not is predicted according to the matching degrees. Whether the hematoma is enlarged or not is predicted according to the existing image data and the clinical examination data, whether the hematoma is enlarged or not can be predicted in advance, and corresponding processing is carried out, so that the condition that part of patients are deteriorated or even die due to untimely hematoma enlargement judgment is avoided.
In order to make the aforementioned and other objects, features and advantages of the present invention comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
Fig. 1 shows a block schematic diagram of an electronic device provided by an embodiment of the present invention.
Fig. 2 illustrates a hematoma expansion prediction method provided by an embodiment of the present invention.
Fig. 3 shows a structural diagram of a hybrid hole segmentation network according to an embodiment of the present invention.
Fig. 4 shows a schematic diagram of the void ratios 1, 2, 3, and 5 provided by the embodiment of the present invention.
Fig. 5 is a flowchart illustrating the sub-steps of step S2 in fig. 2.
Fig. 6 is a diagram illustrating a segmentation result provided by an embodiment of the present invention.
Fig. 7 is a flowchart illustrating a first sub-step of step S4 in fig. 2.
Fig. 8 is a flowchart illustrating sub-steps of step S41 in fig. 7.
FIG. 9 shows a schematic diagram of a three-dimensional twin network provided by an embodiment of the invention.
Fig. 10 is a flowchart illustrating a second sub-step of step S4 in fig. 2.
Fig. 11 is a block diagram illustrating a hematoma expansion prediction apparatus according to an embodiment of the present invention.
Icon: 100-an electronic device; 101-a processor; 102-a memory; 103-a bus; 104-a communication interface; 200-a hematoma expansion prediction device; 201-an acquisition module; 202-processing module.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the invention without making any creative effort, fall within the protection scope of the invention.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures. Meanwhile, in the description of the present invention, the terms "first", "second", and the like are used only for distinguishing the description, and are not to be construed as indicating or implying relative importance.
The hematoma expansion occurring in the early stage after cerebral hemorrhage (generally within 24 hours after the cerebral hemorrhage) is a common change in the cerebral hemorrhage diseases, the hematoma expansion is judged according to whether the volume of the hematoma between follow-up CT and baseline CT is more than 33% or 6 ml, if the hematoma expansion exceeds 33% or 6 ml, the hematoma expansion is judged, the occurrence of the hematoma expansion indicates that the patient is possibly deteriorated and even dead, the time interval between follow-up CT examination and baseline CT examination is relatively long, and the method is used for judging whether the hematoma expansion is not timely enough, so that part of the patients are possibly deteriorated and even dead in the period.
The present invention is directed to solve the above problems, and an improvement of the method for predicting hematoma expansion is that after a cerebral hemorrhage patient is diagnosed, whether hematoma expansion will occur is predicted according to existing image data and clinical examination data, whether hematoma expansion will occur can be predicted in advance, whether hematoma expansion will occur in a patient can be predicted in time, and the method is a key for further diagnosis and treatment of an ICH patient.
The hematoma expansion prediction method provided by the embodiment of the invention is applied to the electronic device 100, and the electronic device 100 may be, but is not limited to, a smart phone, a tablet computer, a personal computer, a vehicle-mounted computer, a Personal Digital Assistant (PDA), and the like. Referring to fig. 1, fig. 1 is a schematic diagram illustrating a partial structure of an electronic device according to an embodiment of the present invention, where the electronic device 100 includes a processor 101, a memory 102, a bus 103, and a communication interface 104. The processor 101, the memory 102 and the communication interface 104 are connected by a bus 103, and the processor 101 is configured to execute an executable module, such as a computer program, stored in the memory 102.
The processor 101 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the hematoma expansion prediction method may be performed by integrated logic circuits of hardware or instructions in the form of software in the processor 101. The Processor 101 may be a general-purpose Processor 101, and includes a Central Processing Unit (CPU), a Network Processor (NP), and the like; the device can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components.
The Memory 102 may comprise a high-speed Random Access Memory (RAM) and may also include a non-volatile Memory (non-volatile Memory), such as at least one disk Memory. The Memory 102 may be, but is not limited to, a Random Access Memory (RAM), a Read Only Memory (ROM), a Programmable Read-Only Memory (PROM), an Erasable Read-Only Memory (EPROM), an electrically Erasable Read-Only Memory (EEPROM), and the like.
The bus 103 may be an ISA (Industry Standard architecture) bus, a PCI (peripheral Component interconnect) bus, an EISA (extended Industry Standard architecture) bus, or the like. Only one bi-directional arrow is shown in fig. 1, but this does not indicate only one bus or one type of bus.
The electronic device 100 implements a communication connection between the electronic device 100 and an external device through at least one communication interface 104 (which may be wired or wireless). The memory 102 is used to store a program, such as the hematoma expansion prediction device 200. The hematoma expansion prediction apparatus 200 includes at least one software function module, which may be stored in the memory 102 in the form of software or firmware (firmware) or fixed in an Operating System (OS) of the electronic device 100. The processor 101, upon receiving the execution instruction, executes the program to implement the hematoma expansion prediction method.
It should be understood that the configuration shown in fig. 1 is merely a schematic application of the configuration of the electronic device 100, and that the electronic device 100 may also include more or fewer components than shown in fig. 1, or have a different configuration than shown in fig. 1. The components shown in fig. 1 may be implemented in hardware, software, or a combination thereof.
Based on the electronic device 100, a possible implementation manner of the hematoma expansion prediction method is given below, an execution subject of the method may be the electronic device 100, please refer to fig. 2, and fig. 2 shows a flowchart of the hematoma expansion prediction method according to an embodiment of the present invention. The hematoma expansion prediction method comprises the following steps:
s1, clinical examination data and CT images containing hematomas are acquired.
In the embodiment of the present invention, the clinical examination data may be, but is not limited to, systolic blood pressure, diastolic blood pressure, blood glucose, renal function, glasgow coma score, anti-platelet therapy status, smoking status, time difference between onset and baseline CT, etc. of the patient, and the CT image may be a CT image including hematoma obtained by photographing the brain of the patient. The CT image comprises a plurality of tomograms, and each tomogram has an image thickness corresponding to the tomogram. It should be noted that the clinical examination data and the CT image belong to the same patient.
And S2, inputting the CT image into a preset automatic segmentation model for segmentation to obtain a segmentation result, and calculating the hematoma volume according to the segmentation result.
In the embodiment of the present invention, the automatic segmentation model may be a hybrid hole segmentation network including a hybrid hole convolution and a U-Net (U-Net for short), where U-Net is an image segmentation network. Referring to fig. 3, the hybrid hole partition network may include 5 pooling layers including 9 convolution blocks, and referring to fig. 4, the convolution kernel of each convolution layer in each convolution block is formed by convolution of four groups of 3 × 3 holes with hole rates of 1, 2, 3, and 5, respectively.
With reference to fig. 3, for each convolution layer in the convolution block 1-1, the convolution kernel is composed of 8 mixed void convolutions of 3 × 3 with void ratios of 1, 2, 3, and 5, and 32 convolution kernels in total, and in order to avoid clipping the feature map after convolution, the feature map size is kept consistent by using the outer liner operation with the same outer liner value and void ratio. After the convolution operation, the feature map is subjected to batch normalization, and then nonlinear activation is performed by using a linear rectification function. In the contraction phase, every time a convolution block operation is performed, 2 × 2 pooling operations are performed on the feature map. In the expansion stage, every time a convolution block operation is performed, the original feature map is subjected to 2 × 2 transposition convolution and is subjected to a folding operation with features of the same order of high resolution, for example, the input feature map of the convolution block 4-2 contains 512 in total, 256 of the feature maps are transposed and convolved by the feature map of low resolution, and the other 256 feature maps output by the convolution block 4-1 are directly copied.
The hybrid hole partition network provided by the embodiment of the invention has the following two advantages: (1) the number of convolution kernels is changed into a half of the original number as a whole, and the two-layer convolution operation of each convolution block is changed into the three-layer convolution operation, so that the model is narrower and deeper, and researches show that the effect of a narrow and deep network is possibly better. (2) The number of the neurons is less, mixed hole convolution is introduced, the receptive field is improved, and the characteristics of multi-scale context information can be captured by combining U-Net, so that the network can adapt to hematomas of different sizes, and meanwhile, the outer lining value corresponding to the void rate is used, and the feature graph is prevented from being cut.
Through adopting mixed hole to cut apart the network, can enlarge the receptive field scope of the neuron of same layer to make this cut apart the network and can adapt to the hematoma of different sizes, and then promote hematoma segmentation performance.
The segmentation result may be obtained by distinguishing a hematoma region from a non-hematoma region in each tomographic image in the CT image. The electronic device 100 stores a hematoma volume calculation formula in advance, and can calculate the hematoma volume by bringing the division result output by the mixed cavity division network into the volume calculation network.
Referring to fig. 5, step S2 may specifically include the following sub-steps:
s21, the multiple tomographic images are input to a mixed-hole segmentation network to perform image segmentation, and segmented images including a hematoma region and a non-hematoma region are obtained for each tomographic image, thereby forming segmentation results.
In the embodiment of the present invention, the segmented image may be an image including a hematoma region and a non-hematoma region other than the hematoma region obtained by image segmentation of the tomographic image, and the segmentation result may be a sum of the segmented images corresponding to all the tomographic images. Referring to fig. 6, the white part is a hematoma area, and the black part is a non-hematoma area.
The step of inputting a plurality of tomographic images into a mixed hole segmentation network to perform image segmentation to obtain segmented images including a hematoma region and a non-hematoma region corresponding to each tomographic image and obtain segmentation results may be understood as inputting each tomographic image into the mixed hole segmentation network to enable the mixed hole segmentation network to segment the tomographic image to obtain segmented images including the hematoma region and the non-hematoma region corresponding to the tomographic image, and performing the same processing on each tomographic image to obtain segmented images corresponding to each tomographic image, wherein all the segmented images corresponding to the tomographic images together form the segmentation results.
And S22, counting the number of hematoma pixel points in the hematoma area corresponding to each sectional image according to the segmentation result.
In the embodiment of the present invention, the number of the hematoma pixel points may be the number of the pixel points in the hematoma area in one divided image. Since each tomographic image is segmented into images to obtain the corresponding segmented image, the step of counting the number of hematoma pixel points in the hematoma region corresponding to each tomographic image according to the segmentation result can be understood as counting the number of pixel points representing the hematoma region (white region in fig. 6) in the segmented image corresponding to each tomographic image to obtain the number of hematoma pixel points corresponding to the image of each tomographic image.
And S23, acquiring the image thickness and hematoma pixel point area of each tomographic image.
In embodiments of the present invention, the image thickness may be the layer thickness of the scan layer. E.g. 5mm, 9mm, 10 mm. The area of the hematoma pixel point can be understood as the area of the hematoma pixel point, namely the product of the length and the width of the hematoma pixel point. Each tomographic image has the corresponding image thickness and hematoma pixel point area, and the image thicknesses of all the tomographic images and the hematoma pixel point areas of the hematoma pixel points obtained by carrying out CT shooting are generally consistent. The step of obtaining the image thickness and the hematoma pixel point area of each tomographic image can be understood as deriving a CT file from the CT device, wherein the CT file comprises the image thickness of each tomographic image and the length and the width of the hematoma pixel point, and the hematoma pixel point area can be obtained according to the length and the width of the hematoma pixel point.
In other embodiments of the present invention, the execution sequence of step S23 and step S22 may be exchanged, or step S23 and step S22 may be executed simultaneously, which is not limited herein.
And S24, substituting the number of all hematoma pixel points, the image thickness of each tomographic image and the area of each hematoma pixel point into a hematoma volume calculation formula for volume calculation to obtain the hematoma volume.
In the embodiment of the present invention, a hematoma volume calculation formula is stored in the electronic device 100 in advance, and the hematoma volume of the patient can be obtained by substituting the number of hematoma pixel points in the hematoma region corresponding to each tomographic image obtained in step S22, and the image thickness and the hematoma pixel point area of each tomographic image obtained in step S23 into the hematoma volume calculation formula. The formula expression for the hematoma volume calculation is as follows:
wherein, pixelsiThe number of hematoma pixel points corresponding to the ith tomographic image, areaiIs the hematoma pixel point area of the ith tomographic image, clicknessiThe thickness of the ith tomographic image, V the hematoma volume, and n the number of tomographic images in the CT image.
As an implementation mode, when the image thickness is 6mm, the area of the hematoma pixel point is 25mm2The CT image comprises 5 tomographic images: the number of hematoma pixel points of the first tomographic image is 234; the number of hematoma pixel points of the second tomographic image is 124; the number of hematoma pixel points of the third tomographic image is 218; the number of hematoma pixel points of the fourth tomographic image is 327; fifth fault mapWhen the number of the image hematoma pixels is 116, the hematoma volume V is 234 × 25 × 6+124 × 25 × 6+218 × 25 × 6+327 × 25 × 6+116 × 25 × 6 ═ 152850mm3。
S3, the segmentation result, the hematoma volume and the clinical examination data are combined into a test sample.
In the embodiment of the present invention, the segmentation result obtained in step S2 and the hematoma volume and the clinical examination data obtained in step S1 are combined to constitute a test sample of the patient.
And S4, calculating the matching degree between each preset supporting sample in the plurality of supporting samples and the test sample, and predicting whether the hematoma is enlarged or not according to the plurality of matching degrees.
In the embodiment of the present invention, the support sample may be a pre-stored segmentation result, hematoma volume, and clinical examination data of a non-hematoma expansion patient, or may be a pre-stored segmentation result, hematoma volume, and clinical examination data of a hematoma expansion patient. It should be noted that, during the process of performing step S4 once, all the support samples should be of the same type, or all the support samples are support samples corresponding to non-hematoma-enlarging patients, or all the support samples are support samples corresponding to hematoma-enlarging patients. The support sample may be obtained in a manner consistent with the manner in which the prediction sample is obtained, with the same processing being performed on the support sample and the test sample.
The electronic device 100 stores a plurality of supporting samples of the same type in advance, calculates a matching degree between each supporting sample of the plurality of supporting samples of the same type and the test sample to obtain a plurality of matching degrees, counts a matching number greater than a preset matching degree among the plurality of matching degrees, and presets whether hematoma will be enlarged or not according to the matching number.
Referring to fig. 7, when the support sample is a non-hematoma-enlarged patient sample, step S4 may include the following sub-steps:
and S41, calculating a first matching degree between each support sample in the preset plurality of support samples and the test sample.
In an embodiment of the present invention, the first matching degree may be a matching degree between the support sample and the test sample corresponding to the support sample being a non-hematoma-enlarging patient. The step of calculating the first matching degree between each of the plurality of preset supporting samples and the test sample may be understood as inputting the test sample and one supporting sample into a preset three-dimensional twin network, so as to calculate the first matching degree between the supporting sample and the test sample, and performing the same processing on each of the plurality of supporting samples in the above manner, so as to obtain the first matching degree between each of the plurality of supporting samples and the test sample.
Referring to fig. 8, step S41 may specifically include the following sub-steps:
s411, feature extraction is carried out on the segmentation result in the test sample, and an image feature vector is obtained.
In the embodiment of the present invention, please refer to fig. 9, the three-dimensional twin network includes a feature extraction network and a matching degree calculation network, the feature extraction network is used for feature extraction, and the matching degree calculation network is used for matching degree calculation. The feature extraction network may include two identical sub-feature extraction networks, which are respectively used for feature extraction of the segmentation result in the test sample and feature extraction of the segmentation result in the support sample.
The parameters of the two sub-feature extraction networks are shared, each sub-feature extraction network consists of four convolution layers and a full connection layer, batch normalization, nonlinear activation and maximum pooling are carried out once per convolution operation in the convolution blocks 1-1, 1-2 and 1-3, the maximum pooling size in the convolution blocks 1-1 and 1-3 is 2 x 1, the maximum pooling size in the convolution block 1-2 is 2 x 2, after the convolution blocks 1-3 are operated, the feature map is flattened and is changed into a one-dimensional feature map, and the features are changed into one-dimensional feature vectors with the length of 64 through the full connection layers 1-1, namely the image feature vectors.
And S412, splicing the image characteristic vector with the hematoma volume and clinical examination data to obtain the hematoma characteristic vector.
In the embodiment of the present invention, the hematoma feature vector may be a one-dimensional feature vector obtained by stitching the image feature vector with the hematoma volume and the clinical examination data. The step of obtaining the hematoma feature vector by splicing the image feature vector with the hematoma volume and the clinical examination data can be understood as that, firstly, the hematoma volume in the test sample is spliced with the clinical examination data to obtain a statistical feature 1, and then, the statistical feature 1 is spliced with the image feature vector to obtain the hematoma feature vector.
S413, obtaining a preset feature vector corresponding to each support sample.
In an embodiment of the present invention, the support sample may be a pre-stored segmentation result, a hematoma volume and clinical examination data of the non-hematoma dilated patient, and the predetermined feature vector may be a hematoma feature vector corresponding to the support sample.
The step of obtaining the preset feature vector corresponding to each support sample can be understood as that, firstly, a sub-feature extraction network in the feature extraction network is used for carrying out feature extraction on a segmentation result in the support sample to obtain a feature vector corresponding to the support sample; then, splicing the hematoma volume in the support sample with clinical examination data to obtain statistical characteristics 2; and finally, splicing the statistical characteristic 2 with the characteristic vector corresponding to the support sample to obtain a preset characteristic vector. The specific process of obtaining the feature vector corresponding to the support sample by using the sub-feature extraction network in the feature extraction network to perform feature extraction on the segmentation result in the support sample is consistent with the process of performing feature extraction on the segmentation result in the test sample, and is not described herein again.
It should be noted that, for each input test sample, the preset feature vector corresponding to the support sample needs to be obtained, so that the obtained preset feature vector corresponding to the support sample may be stored in the memory 102 of the electronic device 100, and when the preset feature vector corresponding to the support sample is obtained again, the preset feature vector can be directly used without performing repeated calculation. The above steps are performed only for the newly added support specimen.
And S414, calculating the matching degree between the hematoma feature vector and each preset feature vector.
In the embodiment of the invention, the haematoma feature vector and the preset feature vector are spliced and then input into a matching degree calculation network, the matching degree calculation network can be composed of three full-connection layer blocks, the spliced haematoma feature vector and the preset feature vector are input into a full-connection block 2-1 to obtain a feature vector with the length of 128, then the feature vector with the length of 128 is input into the full-connection block 2-2 to obtain a feature vector with the length of 64, and finally the feature vector with the length of 64 is input into a full-connection layer with the length of 1 multiplied by 1 to obtain the matching degree between the haematoma feature vector and the preset feature vector, namely the first matching degree.
And performing the same processing on the preset characteristic vector corresponding to each support sample to obtain a first matching degree between the hematoma characteristic vector and each preset characteristic vector.
And S42, comparing each first matching degree in the plurality of first matching degrees with a preset matching value.
And S43, counting the first matching number which is greater than the preset matching value in the plurality of first matching degrees.
In the embodiment of the present invention, the first matching number is a number of first matching degrees greater than a preset matching value among the plurality of first matching degrees. Each of the plurality of first matching degrees obtained in step S41 is compared with a preset matching value (e.g., 60%), the first matching degree greater than the preset matching value is set to 1, the first matching degree smaller than the preset matching value is set to 0, and then the number of the first matching degrees that are 1 is counted, so that the first matching number is obtained.
For example, when the plurality of first matching degrees are respectively 80%, 30%, 25%, 65%, 73%, and the preset matching value is 60%, the first matching degrees 80%, 65%, 73% are all greater than the preset matching value 60%, the first matching degrees 80%, 65%, 73% are all set to 1, the first matching degrees 30%, 25% are all less than the preset matching value 60%, the first matching degrees 30%, 25% are all set to 0, and finally, the number of statistics 1 is 3, and the first matching number is 3.
And S44, predicting that the hematoma will not be enlarged when the first matching number is larger than the preset number.
In an embodiment of the present invention, the first number of matches is compared to a preset number (e.g., 60), and when the first number of matches is greater than the preset number, the hematoma is predicted not to expand.
And S45, predicting the hematoma to be enlarged when the first matching number is less than or equal to the preset number.
In an embodiment of the present invention, the first number of matches is compared to a predetermined number (e.g., 60), and when the first number of matches is less than or equal to the predetermined number, the hematoma is predicted to be enlarged.
It should be noted that the predicted number is related to the number of support samples, and may be 60% of the number of support samples, that is, when the number of support samples is 100, the predicted number may be 60.
Referring to fig. 10, when the supporting sample is a sample of a hematoma-enlarged patient, step S4 may further include the following sub-steps:
and S46, calculating a second matching degree between each support sample in the preset plurality of support samples and the test sample.
In an embodiment of the present invention, the second matching degree may be a matching degree between the support sample and the test sample corresponding to the hematoma expansion patient. The step of calculating the second matching degree between each of the plurality of preset supporting samples and the test sample may be understood as inputting the test sample and one supporting sample into a preset three-dimensional twin network, so as to calculate the second matching degree between the supporting sample and the test sample, and performing the same processing on each of the plurality of supporting samples in the above manner, so as to obtain the second matching degree between each of the plurality of supporting samples and the test sample.
The step of calculating the second matching degree between each of the plurality of support samples and the test sample may specifically refer to substeps S411 to S414, which are not described herein again.
And S47, comparing each second matching degree in the plurality of second matching degrees with a preset matching value.
And S48, counting a second matching number which is greater than a preset matching value in the plurality of second matching degrees.
In the embodiment of the present invention, the second matching number is a number of second matching degrees greater than a preset matching value among the plurality of second matching degrees. Each of the plurality of second matching degrees obtained in step S46 is compared with a preset matching value (e.g., 60%), the second matching degree greater than the preset matching value is set to 1, the second matching degree smaller than the preset matching value is set to 0, and then the number of the second matching degrees that are 1 is counted, so that the second matching number is obtained.
For example, when the plurality of second matching degrees are respectively 80%, 40%, 15%, 65%, and 83%, and the preset matching value is 60%, the second matching degrees 80%, 65%, and 83% are all greater than the preset matching value 60%, the second matching degrees 80%, 65%, and 83% are all set to 1, the second matching degrees 40%, and 15% are all less than the preset matching value 60%, the second matching degrees 40%, and 15% are all set to 0, and finally, the number of statistics 1 is 3, and the second matching number is 3.
And S49, when the second matching number is larger than the preset number, the hematoma is predicted to be enlarged.
In an embodiment of the present invention, the second number of matches is compared to a predetermined number (e.g., 60), and when the second number of matches is greater than the predetermined number, the hematoma is predicted to be enlarged.
And S410, when the second matching number is less than or equal to the preset number, predicting that the hematoma is not enlarged.
In an embodiment of the present invention, the second matching number is compared with a preset number (e.g., 60), and when the second matching number is less than or equal to the preset number, the hematoma is predicted not to be enlarged.
Compared with the prior art, the embodiment of the invention has the following advantages:
firstly, the neural network is used for automatically extracting the image features, so that the extracted image features are not influenced by the subjectivity of people, the film reading efficiency can be effectively improved, and the cost is effectively reduced.
Secondly, whether the hematoma is expanded or not is predicted by comprehensively considering the image characteristics and the statistical characteristics, and the accuracy of hematoma expansion prediction is improved.
Finally, the mixed cavity segmentation network is adopted, the receptive field range of the neurons on the same layer can be expanded, so that the segmentation network can adapt to hematomas with different sizes, and the hematoma segmentation performance is improved.
With reference to the method flows of fig. 2, 5, 7, 8 and 10, a possible implementation manner of the hematoma expansion prediction apparatus 200 is given below, where the hematoma expansion prediction apparatus 200 may be implemented by using the device structure of the electronic device 100 in the above embodiment, or implemented by the processor 101 in the electronic device 100, please refer to fig. 11, and fig. 11 shows a block diagram of the hematoma expansion prediction apparatus provided in the embodiment of the present invention. The hematoma expansion prediction device 200 includes an acquisition module 201 and a processing module 202.
An acquisition module 201 for acquiring clinical examination data and a CT image containing a hematoma;
the processing module 202 is configured to input the CT image into a preset automatic segmentation model for segmentation to obtain a segmentation result, and calculate a hematoma volume according to the segmentation result; forming a test sample by the segmentation result, the hematoma volume and the clinical examination data; and calculating the matching degree between each preset supporting sample in the plurality of supporting samples and the test sample, and predicting whether the hematoma is enlarged or not according to the plurality of matching degrees.
In an embodiment of the present invention, when the supporting sample is a sample of a non-hematoma-enlarging patient, the processing module 202 performs the steps of calculating a matching degree between each supporting sample of a plurality of preset supporting samples and the test sample, and predicting whether the hematoma will be enlarged according to the matching degrees, and specifically: calculating a first matching degree between each preset support sample in a plurality of support samples and the test sample; comparing each first matching degree in the plurality of first matching degrees with a preset matching value; counting a first matching number which is greater than a preset matching value in the plurality of first matching degrees; predicting that the hematoma does not expand when the first matching number is greater than the preset number; when the first number of matches is less than or equal to the predetermined number, the hematoma is predicted to be enlarged.
In an embodiment of the present invention, when the supporting sample is a sample of a patient with enlarged hematoma, the processing module 202 performs the steps of calculating a matching degree between each supporting sample of a plurality of preset supporting samples and the test sample, and predicting whether the hematoma will be enlarged according to the plurality of matching degrees, and specifically: calculating a second matching degree between each preset support sample in the plurality of support samples and the test sample; comparing each second matching degree in the plurality of second matching degrees with a preset matching value; counting a second matching number which is greater than a preset matching value in the plurality of second matching degrees; predicting that the hematoma will expand when the second matching number is greater than the preset number; when the second matching number is less than or equal to the preset number, the hematoma is predicted not to be enlarged.
In an embodiment of the present invention, the automatic segmentation model includes a mixed cavity segmentation network, the CT image includes a plurality of tomographic images, the processing module 202 performs the steps of inputting the CT image into a preset automatic segmentation model for segmentation to obtain a segmentation result, and calculating a hematoma volume according to the segmentation result, and is specifically configured to: inputting a plurality of sectional images into a mixed cavity segmentation network for image segmentation to obtain a segmented image which comprises a hematoma region and a non-hematoma region and corresponds to each sectional image, and forming a segmentation result; counting the number of hematoma pixel points of a hematoma area corresponding to each sectional image according to the segmentation result; acquiring the image thickness and hematoma pixel point area of each tomographic image; and (3) bringing the number of all hematoma pixel points, the image thickness of each tomographic image and the area of each hematoma pixel point into a hematoma volume calculation formula for volume calculation to obtain the hematoma volume.
In the embodiment of the present invention, the formula for calculating the hematoma volume is expressed as follows:
wherein, pixelsiThe number of hematoma pixel points corresponding to the ith tomographic image, areaiIs the hematoma pixel point area of the ith tomographic image, clicknessiThe thickness of the ith tomographic image, n is the number of tomographic images in the CT image, and V is the volume of hematoma.
In the embodiment of the present invention, the hybrid hole segmentation network includes 5 pooling layers, which totally include 9 convolution blocks, and the convolution kernel of each convolution layer in each convolution block is formed by convolution of four groups of 3 × 3 holes with hole rates of 1, 2, 3, and 5, respectively.
In this embodiment of the present invention, the processing module 202 executes a step of calculating a matching degree between each of a plurality of preset supporting samples and the test sample, specifically, for: performing feature extraction on a segmentation result in a test sample to obtain an image feature vector; splicing the image characteristic vector with the hematoma volume and clinical examination data to obtain a hematoma characteristic vector; acquiring a preset feature vector corresponding to each support sample; and calculating the matching degree between the hematoma feature vector and each preset feature vector.
It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working process of the hematoma expansion prediction device 200 described above may refer to the corresponding process in the foregoing method embodiments, and will not be described herein again.
In summary, an embodiment of the present invention provides a method and an apparatus for predicting hematoma expansion, where the method includes: acquiring clinical examination data and a CT image containing a hematoma; inputting the CT image into a preset automatic segmentation model for segmentation to obtain a segmentation result, and calculating the volume of hematoma according to the segmentation result; forming a test sample by the segmentation result, the hematoma volume and the clinical examination data; and calculating the matching degree between each preset supporting sample in the plurality of supporting samples and the test sample, and predicting whether the hematoma is enlarged or not according to the plurality of matching degrees. Compared with the prior art, the invention has the following advantages: whether the hematoma is enlarged or not is predicted according to the existing image data and the clinical examination data, whether the hematoma is enlarged or not can be predicted in advance, and corresponding processing is carried out, so that the condition that part of patients are deteriorated or even die due to untimely hematoma enlargement judgment is avoided.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are merely illustrative, and for example, the flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
In addition, the functional modules in the embodiments of the present invention may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.
The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention. It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
Claims (8)
1. A method for predicting hematoma expansion, the method comprising:
acquiring clinical examination data and an electronic Computed Tomography (CT) image containing hematoma;
inputting the CT image into a preset automatic segmentation model for segmentation to obtain a segmentation result, and calculating the hematoma volume according to the segmentation result;
composing the segmentation result, the hematoma volume and the clinical examination data into a test sample;
calculating the matching degree between each support sample in a plurality of preset support samples and the test sample, and predicting whether the hematoma is expanded or not according to the matching degrees;
wherein, when the supporting sample is a sample of a patient with non-hematoma expansion, the step of calculating a matching degree between each supporting sample in a plurality of preset supporting samples and the test sample, and predicting whether the hematoma will expand according to the plurality of matching degrees comprises:
calculating a first matching degree between each support sample in a plurality of preset support samples and the test sample;
comparing each first matching degree in the plurality of first matching degrees with a preset matching value;
counting a first matching number which is greater than the preset matching value in the plurality of first matching degrees;
predicting that the hematoma does not expand when the first matching number is greater than a preset number;
when the first number of matches is less than or equal to a predetermined number, the hematoma is predicted to be enlarged.
2. The method according to claim 1, wherein the step of calculating a matching degree between each of the predetermined plurality of support samples and the test sample when the support sample is a sample of a patient with enlarged hematoma, and predicting whether the hematoma will be enlarged according to the plurality of matching degrees comprises:
calculating a second matching degree between each support sample in a plurality of preset support samples and the test sample;
comparing each second matching degree in the plurality of second matching degrees with a preset matching value;
counting a second matching number which is greater than the preset matching value in the plurality of second matching degrees;
predicting that the hematoma will expand when the second matching number is greater than a preset number;
predicting that the hematoma does not expand when the second number of matches is less than or equal to a preset number.
3. The method according to claim 1, wherein the automatic segmentation model comprises a mixed cavity segmentation network, the CT image comprises a plurality of tomographic images, and the step of inputting the CT image into a preset automatic segmentation model for segmentation to obtain a segmentation result and calculating the hematoma volume according to the segmentation result comprises:
inputting the multiple tomographic images into the mixed cavity segmentation network for image segmentation to obtain segmentation images which comprise hematoma regions and non-hematoma regions and correspond to each tomographic image, and forming segmentation results;
counting the number of hematoma pixel points of a hematoma area corresponding to each sectional image according to the segmentation result;
acquiring the image thickness and hematoma pixel point area of each tomographic image;
and (3) bringing the number of all hematoma pixel points, the image thickness of each tomographic image and the area of each hematoma pixel point into a hematoma volume calculation formula for volume calculation to obtain the hematoma volume.
4. The method of claim 3, wherein the hematoma volume calculation formula is expressed as follows:
wherein, pixelsiThe number of hematoma pixel points corresponding to the ith tomographic image, areaiIs the hematoma pixel point area of the ith tomographic image, clicknessiThe thickness of the ith tomographic image, n is the number of tomographic images in the CT image, and V is the volume of hematoma.
5. The method of claim 4, wherein the hybrid hole segmentation network comprises 5 pooling layers including 9 convolution blocks, and wherein the convolution kernel of each convolution layer in each convolution block is composed of four groups of 3 x 3 hole convolutions with a hole rate of 1, 2, 3, and 5, respectively.
6. The method according to any one of claims 1 to 3, wherein the step of calculating the degree of matching between each of the predetermined plurality of support samples and the test sample comprises:
performing feature extraction on the segmentation result in the test sample to obtain an image feature vector;
splicing the image characteristic vector with the hematoma volume and clinical examination data to obtain a hematoma characteristic vector;
acquiring a preset feature vector corresponding to each support sample;
and calculating the matching degree between the hematoma feature vector and each preset feature vector.
7. A hematoma expansion prediction device, the device comprising:
an acquisition module for acquiring clinical examination data and an electronic Computed Tomography (CT) image containing hematoma;
the processing module is used for inputting the CT image into a preset automatic segmentation model for segmentation to obtain a segmentation result, and calculating the hematoma volume according to the segmentation result; composing the segmentation result, the hematoma volume and the clinical examination data into a test sample; calculating the matching degree between each support sample in a plurality of preset support samples and the test sample, and predicting whether the hematoma is expanded or not according to the matching degrees;
wherein, when the support sample is a sample of a patient with a hematoma that is not enlarged, the processing module is specifically configured to:
calculating a first matching degree between each support sample in a plurality of preset support samples and the test sample;
comparing each first matching degree in the plurality of first matching degrees with a preset matching value;
counting a first matching number which is greater than the preset matching value in the plurality of first matching degrees;
predicting that the hematoma does not expand when the first matching number is greater than a preset number;
when the first number of matches is less than or equal to a predetermined number, the hematoma is predicted to be enlarged.
8. The device according to claim 7, wherein, when the support sample is a sample of a haematoma volume-enlarging patient, the processing module is particularly adapted to:
calculating a second matching degree between each support sample in a plurality of preset support samples and the test sample;
comparing each second matching degree in the plurality of second matching degrees with a preset matching value;
counting a second matching number which is greater than the preset matching value in the plurality of second matching degrees;
predicting that the hematoma will expand when the second matching number is greater than a preset number;
predicting that the hematoma does not expand when the second number of matches is less than or equal to a preset number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910583573.4A CN110288589B (en) | 2019-06-28 | 2019-06-28 | Hematoma expansion prediction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910583573.4A CN110288589B (en) | 2019-06-28 | 2019-06-28 | Hematoma expansion prediction method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110288589A CN110288589A (en) | 2019-09-27 |
| CN110288589B true CN110288589B (en) | 2021-07-02 |
Family
ID=68020860
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910583573.4A Active CN110288589B (en) | 2019-06-28 | 2019-06-28 | Hematoma expansion prediction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110288589B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111079671B (en) * | 2019-12-20 | 2020-11-03 | 深圳集智数字科技有限公司 | Method and device for detecting abnormal articles in scene |
| CN111723817B (en) * | 2020-06-30 | 2023-09-29 | 重庆大学 | Auxiliary detection method for pulmonary nodules |
| CN113349810B (en) * | 2021-05-27 | 2022-03-01 | 北京安德医智科技有限公司 | Cerebral hemorrhage focus identification and hematoma expansion prediction system and device |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107506579A (en) * | 2017-08-14 | 2017-12-22 | 西南大学 | Cerebral hemorrhage forecast model method for building up and system based on integrated study |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10657642B2 (en) * | 2015-06-19 | 2020-05-19 | New York University | System, method and computer-accessible medium for the determination of accelerated brain atrophy and an optimal drainage site for a subdural hematoma using computed tomography |
| CN109190690B (en) * | 2018-08-17 | 2021-10-19 | 东北大学 | Method for detecting and identifying cerebral microhemorrhage points based on SWI image of machine learning |
-
2019
- 2019-06-28 CN CN201910583573.4A patent/CN110288589B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107506579A (en) * | 2017-08-14 | 2017-12-22 | 西南大学 | Cerebral hemorrhage forecast model method for building up and system based on integrated study |
Non-Patent Citations (3)
| Title |
|---|
| "A New Method for Predicting Intracerebral Hematoma Expansion:Volume Difference Measured by ABC/2 and Slicer Software";Wu Gang etc.;《第十三届中国医师协会神经外科医师年会摘要集》;20181231;论文Introduction,Data acquisition and analysis * |
| "基于深度学习的图像分割研究与应用";张华博;《中国优秀硕士学位论文全文数据库 信息科技辑》;20180815(第08期);参见论文第3章 * |
| Wu Gang etc.."A New Method for Predicting Intracerebral Hematoma Expansion:Volume Difference Measured by ABC/2 and Slicer Software".《第十三届中国医师协会神经外科医师年会摘要集》.2018, * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110288589A (en) | 2019-09-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110288589B (en) | Hematoma expansion prediction method and device | |
| CN107665491B (en) | Pathological image identification method and system | |
| Casanova et al. | Application of random forests methods to diabetic retinopathy classification analyses | |
| Mantas | Classification of intracranial hemorrhage subtypes using deep learning on CT scans | |
| RU2533500C2 (en) | System and method for combining clinical signs and image signs for computer-aided diagnostics | |
| CN111710420B (en) | Complication onset risk prediction method, system, terminal and storage medium based on electronic medical record big data | |
| AU2019205013B2 (en) | Method and system for identification of cerebrovascular abnormalities | |
| CN111862187B (en) | Cup-to-tray ratio determining method, device, equipment and storage medium based on neural network | |
| CN111325725A (en) | Retina image recognition method and device, electronic equipment and storage medium | |
| CN113506310A (en) | Medical image processing method and device, electronic equipment and storage medium | |
| CN111325266A (en) | Method and device for detecting micro-calcified clusters in breast molybdenum target image and electronic equipment | |
| CN112634231A (en) | Image classification method and device, terminal equipment and storage medium | |
| WO2021011775A1 (en) | Systems and methods for generating classifying and quantitative analysis reports of aneurysms from medical image data | |
| CA3213194A1 (en) | Electrocardiogram analysis assistance device, program, electrocardiogram analysis assistance method, and electrocardiogram analysis assistance system | |
| CN116703880A (en) | Immune state prediction model construction method, prediction method and construction device | |
| US20220114726A1 (en) | Identifying vessel occlusions using spatial patterns | |
| CN115829980A (en) | Image recognition method, device, equipment and storage medium for fundus picture | |
| CN115393314A (en) | Deep learning-based oral medical image identification method and system | |
| CN115565666A (en) | Cerebral infarction assessment method and device, electronic equipment and storage medium | |
| CN115274094A (en) | Liver cancer imaging combined AFP diagnosis model construction method, system and device | |
| CN112990339A (en) | Method and device for classifying stomach pathological section images and storage medium | |
| CN113724262A (en) | CNV segmentation method in retina OCT image | |
| CN113130050A (en) | Medical information display method and system | |
| Teomete et al. | Automated computer-aided diagnosis of splenic lesions due to abdominal trauma | |
| Nugroho et al. | Quad Convolutional Layers (QCL) CNN Approach for Classification of Brain Stroke in Diffusion Weighted (DW)-Magnetic Resonance Images (MRI). |
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 |