CN117952962A - Bone mineral density detection image processing method and system - Google Patents
Bone mineral density detection image processing method and system Download PDFInfo
- Publication number
- CN117952962A CN117952962A CN202410342282.7A CN202410342282A CN117952962A CN 117952962 A CN117952962 A CN 117952962A CN 202410342282 A CN202410342282 A CN 202410342282A CN 117952962 A CN117952962 A CN 117952962A
- Authority
- CN
- China
- Prior art keywords
- image
- bone
- bone tissue
- image quality
- segmented
- 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.)
- Pending
Links
- 210000000988 bone and bone Anatomy 0.000 title claims abstract description 338
- 229910052500 inorganic mineral Inorganic materials 0.000 title claims abstract description 48
- 239000011707 mineral Substances 0.000 title claims abstract description 48
- 238000001514 detection method Methods 0.000 title claims abstract description 43
- 238000003672 processing method Methods 0.000 title claims abstract description 20
- 230000011218 segmentation Effects 0.000 claims abstract description 87
- 238000012545 processing Methods 0.000 claims abstract description 10
- 230000000977 initiatory effect Effects 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims description 57
- 230000037182 bone density Effects 0.000 claims description 46
- 238000000034 method Methods 0.000 claims description 43
- 238000005457 optimization Methods 0.000 claims description 23
- 230000006870 function Effects 0.000 claims description 17
- 238000012360 testing method Methods 0.000 claims description 16
- 230000002708 enhancing effect Effects 0.000 claims description 15
- 230000015654 memory Effects 0.000 claims description 15
- 238000012795 verification Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 10
- 238000011176 pooling Methods 0.000 claims description 9
- 238000003860 storage Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 7
- 238000013480 data collection Methods 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 3
- 238000010200 validation analysis Methods 0.000 claims 1
- 238000003745 diagnosis Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 5
- 238000005520 cutting process Methods 0.000 description 4
- 238000002372 labelling Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 210000003484 anatomy Anatomy 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a bone mineral density detection image processing method and a bone mineral density detection image processing system, which relate to the technical field of bone mineral density image detection, wherein bone mineral image quality parameters of bone mineral images to be segmented are collected, bone mineral image quality coefficients are calculated based on the bone mineral image quality parameters, and image quality grades are obtained based on the bone mineral image quality coefficients; if the image quality grade is low, initiating an image resetting instruction; if the image quality grade is a middle grade, optimizing the quality of the bone tissue image to be segmented; if the image quality grade is high, using a bone tissue segmentation model to segment the bone tissue image to obtain a bone tissue region image; the accurate automatic segmentation of the bone mineral density detection image is realized, the bone tissue characteristic region is extracted, the detection error is reduced, and the accuracy of bone mineral density detection and diagnosis is greatly improved.
Description
Technical Field
The invention relates to the technical field of bone mineral density image detection, in particular to a bone mineral density detection image processing method and system.
Background
The dual-energy X-ray bone densitometer used in the current bone density detection needs to process and identify the detected image so as to measure the bone density related parameters. However, the existing image processing mainly relies on manual or simple automatic segmentation methods to cut the region of interest (bone tissue region), and the methods are easily subjectively affected, have poor segmentation effect, cannot effectively extract the bone tissue region, and cause a larger error in detection results.
At present, an automatic technical scheme for accurately cutting out a bone tissue region from a bone tissue image is not available, and a matched method capable of automatically judging the quality of the bone tissue image and optimizing the bone tissue image is also not available, so that the automatic cutting accuracy is further improved;
The Chinese patent with publication number CN116491969A provides a high-precision detection method and a system for bone mineral density values of human bodies, and relates to the technical field of bone mineral density detection, wherein the method comprises the following steps: acquiring an image of a target user through medical image acquisition equipment to obtain user image information, wherein the user image information comprises a user medical image, acquisition equipment parameters and user basic information; constructing a bone mineral density database; constructing a bone density characteristic mapping library; performing image preprocessing; identifying and dividing the preprocessed image, determining multiple image areas, and carrying out matching analysis by utilizing the multiple image areas and the bone density feature mapping library to obtain a bone density analysis result; and inputting a user evaluation model according to the user basic information and the bone mineral density analysis result to obtain user bone mineral density judgment information. But the method also fails to solve the problem of automatically segmenting the region of interest;
Therefore, the invention provides a bone mineral density detection image processing method and a bone mineral density detection image processing system.
Disclosure of Invention
The present invention aims to solve at least one of the technical problems existing in the prior art. Therefore, the invention provides a bone density detection image processing method and a bone density detection image processing system, which adopt a deep learning model to realize accurate and automatic segmentation of bone density detection images, extract bone tissue characteristic areas, reduce detection errors and greatly improve the accuracy of bone density detection and diagnosis.
In order to achieve the above object, a bone mineral density detection image processing method is provided, comprising the following steps:
step one: collecting a sample image set marked with a bone tissue region label;
Step two: training a bone tissue segmentation model using the set of sample images;
Step three: acquiring a bone tissue image to be segmented, collecting bone image quality parameters of the bone tissue image to be segmented, and calculating a bone image quality coefficient based on the bone image quality parameters;
Step four: obtaining an image quality grade based on the bone image quality coefficient; if the image quality grade is low, initiating an image resetting instruction; if the image quality grade is the middle grade, turning to a step five; if the image quality grade is high, turning to a step six;
step five: optimizing the quality of the bone tissue image to be segmented, and recalculated the bone image quality coefficient of the bone tissue image to be segmented after the quality optimization; turning to the fourth step;
Step six: using a bone tissue segmentation model to segment the bone tissue image to obtain a bone tissue region image;
The collecting a sample image set marked with a bone tissue region tag comprises:
Collecting A pieces of bone density sample images obtained by X-ray or other imaging technologies, wherein A is the number of selected bone density sample images;
manually marking the bone tissue area in each collected bone density sample image by using a boundary frame wire;
Dividing the manually marked bone density sample image into a training set, a verification set and a test set;
The training set, the verification set and the test set form a sample image set;
The method for training the bone tissue segmentation model by using the sample image set comprises the following steps of:
constructing a bone tissue segmentation model;
Training a bone tissue segmentation model by using a training set in a sample image set, inputting a bone density sample image in the training set into the bone tissue segmentation model in each training period, calculating a loss function, and updating parameters of the bone tissue segmentation model by using a gradient descent algorithm according to the loss function;
in the training process, the bone tissue segmentation model monitors the change condition of the loss function value on the verification set; stopping training when the loss function value on the verification set reaches convergence;
Evaluating the trained bone tissue segmentation model by using the test set; inputting the bone mineral density sample image in the test set into a model, calculating the prediction result of the bone tissue segmentation model on the test set, comparing the prediction result with a real bone tissue region label, and evaluating the performance of the model;
The constructing the bone tissue segmentation model comprises the following steps:
input layer: receiving as input an image of a bone density sample;
convolution layer and pooling layer: stacking K1 groups of convolution layers and pooling layers after the input layer for extracting bone density sample image features;
Layer of flat: stacking a layer of flat layer after the convolution layer and the pooling layer, and flattening the output of the convolution layer into a one-dimensional vector;
full tie layer: k2 full-connection layers are added behind the flat layer so as to learn the relation between the image characteristics and the region of interest; k1 and K2 are parameters set according to practical model training experience;
Output layer: stacking an output layer behind the full-connection layer, wherein the output layer outputs a prediction area sample label of each pixel point in the bone density sample image; the prediction area sample label is one of 0 or 1, when the pixel point is considered to belong to bone tissue by the bone tissue segmentation model, the prediction area sample label of the pixel point is 1, and when the pixel point is considered not to belong to bone tissue by the bone tissue segmentation model, the prediction area sample label of the pixel point is 0;
the loss function of the bone tissue segmentation model is as follows:
;
wherein N is the number of bone density sample images input into the bone tissue segmentation model, and M is the number of pixels of each bone density sample image;
i is the number of the bone mineral density sample image, j is the number of the pixel point;
yij is the real label of the j-th pixel point in the i-th bone mineral density sample image; the real label is one of 0 or 1, and when the pixel point is in the range of the line frame of the boundary frame, the real label is 1; when the pixel point is out of the range of the boundary frame line frame, the real label is 0;
pij is a prediction area sample label of a j-th pixel point in an i-th bone density sample image output by the bone tissue segmentation model;
the collecting bone image quality parameters of the bone tissue image to be segmented comprises:
Extracting definition scores, contrast scores, noise scores and structural information scores of bone tissue images to be segmented, and forming the definition scores, the contrast scores, the noise scores and the structural information scores into bone image quality parameters;
The method for calculating the bone image quality coefficient based on the bone image quality parameter comprises the following steps:
Marking definition scores, contrast scores, noise scores and structural information scores in bone image quality parameters as w1, w2, w3 and w4 respectively;
Marking the bone image quality coefficient as W;
The calculation formula of the bone image quality coefficient W is: wherein a1, a2, a3 and a4 are all preset proportionality coefficients.
The method for obtaining the image quality grade based on the bone image quality coefficient comprises the following steps:
presetting a low quality coefficient threshold value and a medium quality coefficient threshold value;
if the bone image quality coefficient is less than the low quality coefficient threshold value, setting the image quality level to be low;
If the quality coefficient is low Setting the image quality grade as a middle grade if the bone image quality coefficient is less than the middle quality coefficient threshold;
If the quality coefficient of bone image Setting the image quality grade to be high-grade if the medium quality coefficient is threshold;
the method for optimizing the quality of the bone tissue image to be segmented comprises the following steps:
Sequentially carrying out operations of enhancing definition, enhancing contrast, denoising and enhancing edges on the bone tissue image to be segmented, calculating index scores corresponding to each operation after each operation is completed, and returning the bone tissue image to be segmented to the front of the operation and skipping the operation if the index scores after the operation are smaller than the corresponding index scores before the operation; the index score corresponding to each operation comprises the following steps: enhancing the definition corresponding to the definition score, enhancing the contrast corresponding to the contrast score, denoising corresponding to the noise score, and enhancing the structural information score corresponding to the edge operation;
The method for segmenting the bone tissue region image comprises the following steps:
Inputting the bone tissue image to be segmented into a bone tissue segmentation model, obtaining a prediction area sample label of each pixel point in the bone tissue image to be segmented output by the bone tissue segmentation model, and segmenting an area consisting of pixel points with the prediction area sample label of 1 to obtain a bone tissue area image.
The bone mineral density detection image processing system comprises a training data collection module, a model training module, an image quality grade calculation module, a quality optimization module and a bone tissue segmentation module; wherein, each module is electrically connected;
the training data collection module is used for collecting a sample image set marked with a bone tissue region label and sending the sample image set to the model training module;
the model training module is used for training the bone tissue segmentation model by using the sample image set and sending the bone tissue segmentation model to the bone tissue segmentation module;
The image quality grade calculating module is used for acquiring the bone tissue image to be segmented, collecting bone image quality parameters of the bone tissue image to be segmented, calculating a bone image quality coefficient based on the bone image quality parameters, obtaining an image quality grade based on the bone image quality coefficient, initiating an image resetting instruction if the image quality grade is low, sending the bone tissue image to be segmented to the quality optimizing module if the image quality grade is medium, and sending the bone tissue image to be segmented to the bone tissue segmentation module if the image quality grade is high;
the quality optimization module is used for carrying out quality optimization on the bone tissue image to be segmented, recalculates the bone image quality coefficient of the bone tissue image to be segmented after the quality optimization, and sends the bone image quality coefficient and the bone tissue image to be segmented after the quality optimization to the image quality grade calculation module;
And the bone tissue segmentation module is used for segmenting the bone tissue region image by using a bone tissue segmentation model to the bone tissue image to be segmented.
An electronic device is proposed, comprising: a processor and a memory, wherein the memory stores a computer program for the processor to call;
The processor executes the above-described bone mineral density detection image processing method by calling a computer program stored in the memory.
A computer-readable storage medium is proposed, on which a computer program is stored that is erasable;
the computer program, when run on a computer device, causes the computer device to perform the bone mineral density detection image processing method described above.
Compared with the prior art, the invention has the beneficial effects that:
According to the invention, a bone tissue segmentation model is trained by collecting a sample image set marked with a bone tissue region label, a bone tissue image to be segmented is obtained, bone image quality parameters of the bone tissue image to be segmented are collected, a bone image quality coefficient is calculated based on the bone image quality parameters, and an image quality grade is obtained based on the bone image quality coefficient; if the image quality grade is low, initiating an image resetting instruction; if the image quality grade is a medium grade, carrying out quality optimization on the bone tissue image to be segmented, and recalculating the bone image quality coefficient of the bone tissue image to be segmented after quality optimization; if the image quality grade is high, using a bone tissue segmentation model to segment the bone tissue image to obtain a bone tissue region image; through training the bone tissue segmentation model, the automatic cutting of the region of interest from the bone tissue image is realized, and then an image quality optimization scheme is further provided, so that the accuracy of automatically cutting the region of interest is further improved, the accurate automatic segmentation of the bone density detection image is realized by adopting the deep learning model, and the bone tissue characteristic region is extracted, so that the detection error is reduced, and the accuracy of bone density detection and diagnosis is greatly improved.
Drawings
FIG. 1 is a flowchart of a bone mineral density detection image processing method according to embodiment 1 of the present invention;
FIG. 2 is a block diagram showing a bone mineral density detection image processing system according to embodiment 2 of the present invention;
fig. 3 is a schematic structural diagram of an electronic device in embodiment 3 of the present invention;
Fig. 4 is a schematic diagram of a computer-readable storage medium according to embodiment 4 of the present invention.
Detailed Description
The technical solutions of the present invention will be clearly and completely described in connection with the embodiments, and it is obvious that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
As shown in fig. 1, a bone mineral density detection image processing method includes the following steps:
step one: collecting a sample image set marked with a bone tissue region label;
Step two: training a bone tissue segmentation model using the set of sample images;
Step three: acquiring a bone tissue image to be segmented, collecting bone image quality parameters of the bone tissue image to be segmented, and calculating a bone image quality coefficient based on the bone image quality parameters;
Step four: obtaining an image quality grade based on the bone image quality coefficient; if the image quality grade is low, initiating an image resetting instruction; if the image quality grade is the middle grade, turning to a step five; if the image quality grade is high, turning to a step six;
step five: optimizing the quality of the bone tissue image to be segmented, and recalculated the bone image quality coefficient of the bone tissue image to be segmented after the quality optimization; turning to the fourth step;
Step six: using a bone tissue segmentation model to segment the bone tissue image to obtain a bone tissue region image;
Wherein the collecting the sample image set marked with the bone tissue region tag comprises:
Collecting A pieces of bone density sample images obtained by X-ray or other imaging technologies, wherein A is the number of selected bone density sample images; it should be noted that, the bone mineral density sample image covers various anatomical structures, lesion types and shooting conditions as much as possible, so as to ensure that the model has good generalization capability;
Manually marking the bone tissue area in each collected bone density sample image by using a boundary frame wire; the labeling needs to use a professional medical image labeling tool, such as LabelImg, ITK-SNAP and the like, by a medical image expert or a specially trained labeling person; boundary frame lines in the labeling process should outline the boundary of bone tissues as accurately as possible so as to improve the accuracy of model training;
Dividing the manually marked bone density sample image into a training set, a verification set and a test set; it can be understood that the training set, the verification set and the test set are generally divided by adopting the proportion of 70% -15% -15%, so that the model can be fully learned and generalized in the training process;
The training set, the verification set and the test set form a sample image set;
further, the method for training the bone tissue segmentation model by using the sample image set is as follows:
constructing a bone tissue segmentation model;
Training a bone tissue segmentation model by using a training set in a sample image set, inputting a bone density sample image in the training set into the bone tissue segmentation model in each training period, calculating a loss function, and updating parameters of the bone tissue segmentation model by using a gradient descent algorithm according to the loss function;
in the training process, the bone tissue segmentation model monitors the change condition of the loss function value on the verification set; stopping training when the loss function value on the verification set reaches convergence so as to prevent over fitting;
evaluating the trained bone tissue segmentation model by using the test set; specifically, inputting an image of a bone density sample in a test set into a model, calculating a prediction result of a bone tissue segmentation model on the test set, comparing the prediction result with a real bone tissue region label, and evaluating the performance of the model;
specifically, the constructing the bone tissue segmentation model includes:
input layer: receiving as input an image of a bone density sample;
convolution layer and pooling layer: stacking K1 groups of convolution layers and pooling layers after the input layer for extracting bone density sample image features;
Layer of flat: stacking a layer of flat layer after the convolution layer and the pooling layer, and flattening the output of the convolution layer into a one-dimensional vector;
full tie layer: k2 full-connection layers are added behind the flat layer so as to learn the relation between the image characteristics and the region of interest; k1 and K2 are parameters set according to practical model training experience;
Output layer: stacking an output layer behind the full-connection layer, wherein the output layer outputs a prediction area sample label of each pixel point in the bone density sample image; the prediction area sample label is one of 0 or 1, when the pixel point is considered to belong to bone tissue by the bone tissue segmentation model, the prediction area sample label of the pixel point is 1, and when the pixel point is considered not to belong to bone tissue by the bone tissue segmentation model, the prediction area sample label of the pixel point is 0;
the loss function of the bone tissue segmentation model is as follows:
;
wherein N is the number of bone density sample images input into the bone tissue segmentation model, and M is the number of pixels of each bone density sample image;
i is the number of the bone mineral density sample image, j is the number of the pixel point;
yij is the real label of the j-th pixel point in the i-th bone mineral density sample image; the real label is one of 0 or 1, and when the pixel point is in the range of the line frame of the boundary frame, the real label is 1; when the pixel point is out of the range of the boundary frame line frame, the real label is 0;
pij is a prediction area sample label of a j-th pixel point in an i-th bone density sample image output by the bone tissue segmentation model;
it can be appreciated that such a loss function can better measure the prediction accuracy of the model at each pixel point, thereby more accurately guiding the training and optimization of the model;
Further, the acquiring of the bone tissue image to be segmented refers to acquiring a bone tissue image of a real patient collected by an X-ray or other imaging technology, so as to further segment a bone tissue region of the bone tissue image of the real patient;
Further, the collecting bone image quality parameters of the bone tissue image to be segmented includes:
Extracting definition scores, contrast scores, noise scores and structural information scores of bone tissue images to be segmented, and forming the definition scores, the contrast scores, the noise scores and the structural information scores into bone image quality parameters;
Specifically, the sharpness score may be represented by using any one of a Sobel gradient, an amplitude spectrum, and a blur degree of the image;
The contrast score can be represented by any one of indexes such as variance, dynamic range and the like of the image after the contrast enhancement;
The noise score can be represented by any one of indexes such as mean square error or signal to noise ratio of the image;
the structural information score may be represented using a structural similarity index;
Specifically, the method for calculating the structural similarity index comprises the following steps:
Extracting structural information from the bone tissue image to be segmented using an image feature extraction method such as a local binary pattern (Local Binary Patterns, LBP), a Gray-Level Co-occurrence Matrix, GLCM, a directional gradient histogram (Histogram of Oriented Gradients, HOG), etc.;
comparing the extracted structural information of the bone tissue image to be segmented with the structural information of the reference image, and calculating the similarity between the structural information, wherein the specific similarity calculation mode is any one of calculating a structural similarity index (Structural Similarity Index, SSIM), a correlation coefficient or Euclidean distance; the reference image is a bone tissue image with a complete bone tissue structure which is manually selected in advance;
The similarity between the structural information is a structural similarity index;
Further, the method for calculating the bone image quality coefficient based on the bone image quality parameter is as follows:
Marking definition scores, contrast scores, noise scores and structural information scores in bone image quality parameters as w1, w2, w3 and w4 respectively;
Marking the bone image quality coefficient as W;
The calculation formula of the bone image quality coefficient W is: Wherein a1, a2, a3 and a4 are all preset proportionality coefficients;
Further, the method for obtaining the image quality grade based on the bone image quality coefficient is as follows:
presetting a low quality coefficient threshold value and a medium quality coefficient threshold value;
if the bone image quality coefficient is less than the low quality coefficient threshold value, setting the image quality level to be low;
If the quality coefficient is low Setting the image quality grade as a middle grade if the bone image quality coefficient is less than the middle quality coefficient threshold;
If the quality coefficient of bone image Setting the image quality grade to be high-grade if the medium quality coefficient is threshold;
Further, the image resetting instruction is an instruction for suggesting the real patient to re-shoot the bone tissue image;
the method for optimizing the quality of the bone tissue image to be segmented comprises the following steps:
sequentially carrying out operations of enhancing definition, enhancing contrast, denoising and enhancing edges on the bone tissue image to be segmented, calculating index scores corresponding to each operation after each operation is completed, and if the index scores after the operation are smaller than the corresponding index scores before the operation, backing the bone tissue image to be segmented to the front of the operation, and skipping the operation so as to ensure that the quality of the bone tissue image to be segmented is continuously optimized and cannot be backed; the index score corresponding to each operation comprises the following steps: enhancing the definition corresponding to the definition score, enhancing the contrast corresponding to the contrast score, denoising corresponding to the noise score, and enhancing the structural information score corresponding to the edge operation;
in particular, the enhanced sharpness may be achieved by using deblurring models including, but not limited to, using MSSNet, MIMO-UNet, and the like;
the enhanced contrast may increase the contrast of the image by redistributing the gray level distribution of the pixels of the image;
The denoising can be carried out by using a Gaussian filter to smooth the image and remove high-frequency noise, or by using median filtering to replace a central pixel value with a median value of pixel values in a sliding window and carrying out impulse noise removing treatment on the image;
The edge enhancement operation can be performed by utilizing a Sobel operator, the edge information of the image is enhanced by calculating the gradient of the pixel points in the image, and the edge in the image can be detected by using a Canny algorithm and the brightness of the edge pixels is improved, so that more accurate edge detection and extraction are realized;
further, the method for segmenting the bone tissue region image by using the bone tissue segmentation model comprises the following steps:
Inputting the bone tissue image to be segmented into a bone tissue segmentation model, obtaining a prediction area sample label of each pixel point in the bone tissue image to be segmented output by the bone tissue segmentation model, and segmenting an area consisting of pixel points with the prediction area sample label of 1 to obtain a bone tissue area image.
Example 2
As shown in fig. 2, a bone mineral density detection image processing system includes a training data collection module, a model training module, an image quality grade calculation module, a quality optimization module, and a bone tissue segmentation module; wherein, each module is electrically connected;
The training data collection module is mainly used for collecting a sample image set marked with a bone tissue region label and sending the sample image set to the model training module;
The model training module is mainly used for training a bone tissue segmentation model by using a sample image set and sending the bone tissue segmentation model to the bone tissue segmentation module;
The image quality grade calculating module is mainly used for acquiring bone tissue images to be segmented, collecting bone image quality parameters of the bone tissue images to be segmented, calculating bone image quality coefficients based on the bone image quality parameters, acquiring image quality grades based on the bone image quality coefficients, initiating an image resetting instruction if the image quality grades are low-grade, sending the bone tissue images to be segmented to the quality optimizing module if the image quality grades are medium-grade, and sending the bone tissue images to be segmented to the bone tissue segmentation module if the image quality grades are high-grade;
The quality optimization module is mainly used for performing quality optimization on the bone tissue image to be segmented, recalculates the bone image quality coefficient of the bone tissue image to be segmented after quality optimization, and sends the bone image quality coefficient and the bone tissue image to be segmented after quality optimization to the image quality grade calculation module;
The bone tissue segmentation module is mainly used for segmenting the bone tissue region image by using a bone tissue segmentation model to the bone tissue image to be segmented.
Example 3
Fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present application. As shown in fig. 3, there is also provided an electronic device 100 according to yet another aspect of the present application. The electronic device 100 may include one or more processors and one or more memories. Wherein the memory has stored therein computer readable code which, when executed by the one or more processors, may perform the bone mineral density detection image processing method implementation as described above.
The method or apparatus according to embodiments of the present application may also be implemented by means of the architecture of the electronic device shown in fig. 3. As shown in fig. 3, the electronic device 100 may include a bus 101, one or more CPUs 102, a ROM103, a RAM104, a communication port 105 connected to a network, an input/output component 106, a hard disk 107, and the like. A storage device in the electronic device 100, such as the ROM103 or the hard disk 107, may store the bone mineral density detection image processing method implementation provided by the present application. The bone mineral density detection image processing method implementation may for example comprise the steps of: step one: collecting a sample image set marked with a bone tissue region label; step two: training a bone tissue segmentation model using the set of sample images; step three: acquiring a bone tissue image to be segmented, collecting bone image quality parameters of the bone tissue image to be segmented, and calculating a bone image quality coefficient based on the bone image quality parameters; step four: obtaining an image quality grade based on the bone image quality coefficient; if the image quality grade is low, initiating an image resetting instruction; if the image quality grade is the middle grade, turning to a step five; if the image quality grade is high, turning to a step six; step five: optimizing the quality of the bone tissue image to be segmented, and recalculated the bone image quality coefficient of the bone tissue image to be segmented after the quality optimization; turning to the fourth step; step six: using a bone tissue segmentation model to segment the bone tissue image to obtain a bone tissue region image;
Further, the electronic device 100 may also include a user interface 108. Of course, the architecture shown in fig. 3 is merely exemplary, and one or more components of the electronic device shown in fig. 3 may be omitted as may be practical in implementing different devices.
Example 4
FIG. 4 is a schematic diagram of a computer-readable storage medium according to one embodiment of the present application. As shown in fig. 4, is a computer-readable storage medium 200 according to one embodiment of the application. The computer-readable storage medium 200 has stored thereon computer-readable instructions. When the computer readable instructions are executed by the processor, the bone mineral density detection image processing method according to the embodiment of the present application described with reference to the above drawings may be performed. Computer-readable storage medium 200 includes, but is not limited to, volatile memory and/or nonvolatile memory. Volatile memory can include Random Access Memory (RAM), cache memory (cache), and the like. The non-volatile memory may include Read Only Memory (ROM), hard disk, flash memory, and the like.
In addition, according to embodiments of the present application, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, the present application provides a non-transitory machine-readable storage medium storing machine-readable instructions executable by a processor to perform instructions corresponding to the method steps provided by the present application, which when executed by a Central Processing Unit (CPU), perform the functions defined above in the method of the present application.
The methods and apparatus, devices of the present application may be implemented in numerous ways. For example, the methods and apparatus, devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, firmware. The above-described sequence of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the sequence specifically described above unless specifically stated otherwise. Furthermore, in some embodiments, the present application may also be embodied as programs recorded in a recording medium, the programs including machine-readable instructions for implementing the methods according to the present application. Thus, the present application also covers a recording medium storing a program for executing the method according to the present application.
In addition, in the foregoing technical solutions provided in the embodiments of the present application, parts consistent with implementation principles of corresponding technical solutions in the prior art are not described in detail, so that redundant descriptions are avoided.
The purpose, technical scheme and beneficial effects of the invention are further described in detail in the detailed description. It is to be understood that the above description is only of specific embodiments of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
The above preset parameters or preset thresholds are set by those skilled in the art according to actual conditions or are obtained by mass data simulation.
The above embodiments are only for illustrating the technical method of the present invention, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical method of the present invention may be modified or substituted with equivalents thereof without departing from the spirit and scope of the technical method of the present invention.
Claims (12)
1. A bone mineral density detection image processing method, characterized by comprising the steps of:
step one: collecting a sample image set marked with a bone tissue region label;
Step two: training a bone tissue segmentation model using the set of sample images;
Step three: acquiring a bone tissue image to be segmented, collecting bone image quality parameters of the bone tissue image to be segmented, and calculating a bone image quality coefficient based on the bone image quality parameters;
Step four: obtaining an image quality grade based on the bone image quality coefficient; if the image quality grade is low, initiating an image resetting instruction; if the image quality grade is the middle grade, turning to a step five; if the image quality grade is high, turning to a step six;
step five: optimizing the quality of the bone tissue image to be segmented, and recalculated the bone image quality coefficient of the bone tissue image to be segmented after the quality optimization; turning to the fourth step;
step six: and using a bone tissue segmentation model to segment the bone tissue image to obtain a bone tissue region image.
2. The method of claim 1, wherein collecting the sample image set labeled with the bone tissue region label comprises:
Collecting A bone density sample images, wherein A is the number of selected bone density sample images;
manually marking the bone tissue area in each collected bone density sample image by using a boundary frame wire;
Dividing the manually marked bone density sample image into a training set, a verification set and a test set;
the training set, the validation set and the test set form a sample image set.
3. The method of claim 2, wherein the training the bone tissue segmentation model using the sample image set is performed by:
constructing a bone tissue segmentation model;
Training a bone tissue segmentation model by using a training set in a sample image set, inputting a bone density sample image in the training set into the bone tissue segmentation model in each training period, calculating a loss function, and updating parameters of the bone tissue segmentation model by using a gradient descent algorithm according to the loss function;
in the training process, the bone tissue segmentation model monitors the change condition of the loss function value on the verification set; stopping training when the loss function value on the verification set reaches convergence;
Evaluating the trained bone tissue segmentation model by using the test set; inputting the bone mineral density sample image in the test set into the model, calculating the prediction result of the bone tissue segmentation model on the test set, comparing with the real bone tissue region label, and evaluating the performance of the model.
4. A bone mineral density detection image processing method as in claim 3 wherein constructing a bone tissue segmentation model comprises:
input layer: receiving as input an image of a bone density sample;
convolution layer and pooling layer: stacking K1 groups of convolution layers and pooling layers after the input layer for extracting bone density sample image features;
Layer of flat: stacking a layer of flat layer after the convolution layer and the pooling layer, and flattening the output of the convolution layer into a one-dimensional vector;
full tie layer: k2 full-connection layers are added behind the flat layer so as to learn the relation between the image characteristics and the region of interest; k1 and K2 are parameters set according to practical model training experience;
Output layer: stacking an output layer behind the full-connection layer, wherein the output layer outputs a prediction area sample label of each pixel point in the bone density sample image; the prediction area sample label is one of 0 or 1, when the pixel point is considered to belong to bone tissue by the bone tissue segmentation model, the prediction area sample label of the pixel point is 1, and when the pixel point is considered not to belong to bone tissue by the bone tissue segmentation model, the prediction area sample label of the pixel point is 0;
the loss function of the bone tissue segmentation model is as follows:
;
wherein N is the number of bone density sample images input into the bone tissue segmentation model, and M is the number of pixels of each bone density sample image;
i is the number of the bone mineral density sample image, j is the number of the pixel point;
yij is the real label of the j-th pixel point in the i-th bone mineral density sample image; the real label is one of 0 or 1, and when the pixel point is in the range of the line frame of the boundary frame, the real label is 1; when the pixel point is out of the range of the boundary frame line frame, the real label is 0;
pij is a prediction area sample label of the jth pixel point in the ith bone density sample image output by the bone tissue segmentation model.
5. The method according to claim 4, wherein the collecting bone image quality parameters of the bone tissue image to be segmented comprises:
And extracting definition scores, contrast scores, noise scores and structural information scores of the bone tissue images to be segmented, and forming the definition scores, the contrast scores, the noise scores and the structural information scores into bone image quality parameters.
6. The method according to claim 5, wherein the calculating the bone image quality coefficient based on the bone image quality parameter is:
Marking definition scores, contrast scores, noise scores and structural information scores in bone image quality parameters as w1, w2, w3 and w4 respectively;
Marking the bone image quality coefficient as W;
The calculation formula of the bone image quality coefficient W is: wherein a1, a2, a3 and a4 are all preset proportionality coefficients.
7. The method according to claim 6, wherein the image quality level is obtained based on the bone image quality coefficient by:
presetting a low quality coefficient threshold value and a medium quality coefficient threshold value;
if the bone image quality coefficient is less than the low quality coefficient threshold value, setting the image quality level to be low;
If the quality coefficient is low Setting the image quality grade as a middle grade if the bone image quality coefficient is less than the middle quality coefficient threshold;
If the quality coefficient of bone image And setting the image quality level to be high-level if the quality coefficient is threshold.
8. The method for processing bone mineral density detection image according to claim 7, wherein the method for optimizing the quality of the bone tissue image to be segmented is as follows:
Sequentially carrying out operations of enhancing definition, enhancing contrast, denoising and enhancing edges on the bone tissue image to be segmented, calculating index scores corresponding to each operation after each operation is completed, and returning the bone tissue image to be segmented to the front of the operation and skipping the operation if the index scores after the operation are smaller than the corresponding index scores before the operation; the index score corresponding to each operation comprises the following steps: the sharpness is enhanced, the contrast is enhanced, the noise is removed, and the structural information is enhanced.
9. The method of claim 8, wherein the dividing the image of the bone tissue region is:
Inputting the bone tissue image to be segmented into a bone tissue segmentation model, obtaining a prediction area sample label of each pixel point in the bone tissue image to be segmented output by the bone tissue segmentation model, and segmenting an area consisting of pixel points with the prediction area sample label of 1 to obtain a bone tissue area image.
10. A bone mineral density detection image processing system for implementing the bone mineral density detection image processing method according to any one of claims 1 to 9, characterized by comprising a training data collection module, a model training module, an image quality level calculation module, a quality optimization module and a bone tissue segmentation module; wherein, each module is electrically connected;
the training data collection module is used for collecting a sample image set marked with a bone tissue region label and sending the sample image set to the model training module;
the model training module is used for training the bone tissue segmentation model by using the sample image set and sending the bone tissue segmentation model to the bone tissue segmentation module;
The image quality grade calculating module is used for acquiring the bone tissue image to be segmented, collecting bone image quality parameters of the bone tissue image to be segmented, calculating a bone image quality coefficient based on the bone image quality parameters, obtaining an image quality grade based on the bone image quality coefficient, initiating an image resetting instruction if the image quality grade is low, sending the bone tissue image to be segmented to the quality optimizing module if the image quality grade is medium, and sending the bone tissue image to be segmented to the bone tissue segmentation module if the image quality grade is high;
the quality optimization module is used for carrying out quality optimization on the bone tissue image to be segmented, recalculates the bone image quality coefficient of the bone tissue image to be segmented after the quality optimization, and sends the bone image quality coefficient and the bone tissue image to be segmented after the quality optimization to the image quality grade calculation module;
And the bone tissue segmentation module is used for segmenting the bone tissue region image by using a bone tissue segmentation model to the bone tissue image to be segmented.
11. An electronic device, comprising: a processor and a memory, wherein,
The memory stores a computer program which can be called by the processor;
the processor executes the bone mineral density detection image processing method according to any one of claims 1 to 9 in the background by calling a computer program stored in the memory.
12. A computer readable storage medium having stored thereon a computer program that is erasable;
The computer program, when run on a computer device, causes the computer device to perform the bone density detection image processing method of any one of claims 1-9 in the background.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410342282.7A CN117952962A (en) | 2024-03-25 | 2024-03-25 | Bone mineral density detection image processing method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410342282.7A CN117952962A (en) | 2024-03-25 | 2024-03-25 | Bone mineral density detection image processing method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117952962A true CN117952962A (en) | 2024-04-30 |
Family
ID=90792314
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410342282.7A Pending CN117952962A (en) | 2024-03-25 | 2024-03-25 | Bone mineral density detection image processing method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117952962A (en) |
Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2690596A1 (en) * | 2012-07-24 | 2014-01-29 | Agfa Healthcare | Method, apparatus and system for automated spine labeling |
| CN107358608A (en) * | 2017-08-23 | 2017-11-17 | 西安邮电大学 | Bone tissue geometric state parameter auto-testing device and method based on image processing techniques |
| CN110428410A (en) * | 2019-07-31 | 2019-11-08 | 腾讯医疗健康(深圳)有限公司 | Eyeground medical image processing method, device, equipment and storage medium |
| CN112529863A (en) * | 2020-12-04 | 2021-03-19 | 推想医疗科技股份有限公司 | Method and device for measuring bone density |
| CN113469961A (en) * | 2021-06-24 | 2021-10-01 | 浙江大学医学院附属第一医院 | Neural network-based carpal tunnel image segmentation method and system |
| CN113538329A (en) * | 2020-04-20 | 2021-10-22 | Fei 公司 | Method implemented by a data processing device and charged particle beam apparatus for inspecting a sample using such a method |
| CN114066857A (en) * | 2021-11-18 | 2022-02-18 | 烟台艾睿光电科技有限公司 | Infrared image quality evaluation method and device, electronic equipment and readable storage medium |
| CN114140487A (en) * | 2020-08-12 | 2022-03-04 | 通用电气精准医疗有限责任公司 | Brain image segmentation method and device |
| CN114429468A (en) * | 2022-01-27 | 2022-05-03 | 华中科技大学同济医学院附属同济医院 | Bone age measuring method, bone age measuring system, electronic device and computer-readable storage medium |
| WO2022261550A1 (en) * | 2021-06-11 | 2022-12-15 | Trustees Of Tufts College | Method and apparatus for image processing |
| KR102531759B1 (en) * | 2021-11-11 | 2023-05-12 | 한국기초과학지원연구원 | Method and Apparatus for Screening Osteoporosis Drug Based on Image Analysis of Bone Tissue Mimetics |
| CN116109570A (en) * | 2022-12-22 | 2023-05-12 | 北京医准智能科技有限公司 | Bone mineral density measuring method, device, equipment and storage medium |
| CN116681627A (en) * | 2023-08-03 | 2023-09-01 | 佛山科学技术学院 | Cross-scale fusion self-adaptive underwater image generation countermeasure enhancement method |
| CN117115600A (en) * | 2023-09-05 | 2023-11-24 | 佳讯飞鸿（北京）智能科技研究院有限公司 | No-reference image quality evaluation method and device and electronic equipment |
| CN117314872A (en) * | 2023-10-12 | 2023-12-29 | 武汉大学人民医院（湖北省人民医院） | Intelligent segmentation method and device for retina image |
| CN117475121A (en) * | 2023-12-04 | 2024-01-30 | 河南科技大学 | Image processing system and image processing method |
| US20240070835A1 (en) * | 2022-08-23 | 2024-02-29 | Samsung Electronics Co., Ltd. | System and platform for automatic optimization of image quality of image sensor and operating method thereof |
-
2024
- 2024-03-25 CN CN202410342282.7A patent/CN117952962A/en active Pending
Patent Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2690596A1 (en) * | 2012-07-24 | 2014-01-29 | Agfa Healthcare | Method, apparatus and system for automated spine labeling |
| CN107358608A (en) * | 2017-08-23 | 2017-11-17 | 西安邮电大学 | Bone tissue geometric state parameter auto-testing device and method based on image processing techniques |
| CN110428410A (en) * | 2019-07-31 | 2019-11-08 | 腾讯医疗健康(深圳)有限公司 | Eyeground medical image processing method, device, equipment and storage medium |
| CN113538329A (en) * | 2020-04-20 | 2021-10-22 | Fei 公司 | Method implemented by a data processing device and charged particle beam apparatus for inspecting a sample using such a method |
| CN114140487A (en) * | 2020-08-12 | 2022-03-04 | 通用电气精准医疗有限责任公司 | Brain image segmentation method and device |
| CN112529863A (en) * | 2020-12-04 | 2021-03-19 | 推想医疗科技股份有限公司 | Method and device for measuring bone density |
| WO2022261550A1 (en) * | 2021-06-11 | 2022-12-15 | Trustees Of Tufts College | Method and apparatus for image processing |
| CN113469961A (en) * | 2021-06-24 | 2021-10-01 | 浙江大学医学院附属第一医院 | Neural network-based carpal tunnel image segmentation method and system |
| KR102531759B1 (en) * | 2021-11-11 | 2023-05-12 | 한국기초과학지원연구원 | Method and Apparatus for Screening Osteoporosis Drug Based on Image Analysis of Bone Tissue Mimetics |
| CN114066857A (en) * | 2021-11-18 | 2022-02-18 | 烟台艾睿光电科技有限公司 | Infrared image quality evaluation method and device, electronic equipment and readable storage medium |
| CN114429468A (en) * | 2022-01-27 | 2022-05-03 | 华中科技大学同济医学院附属同济医院 | Bone age measuring method, bone age measuring system, electronic device and computer-readable storage medium |
| US20240070835A1 (en) * | 2022-08-23 | 2024-02-29 | Samsung Electronics Co., Ltd. | System and platform for automatic optimization of image quality of image sensor and operating method thereof |
| CN116109570A (en) * | 2022-12-22 | 2023-05-12 | 北京医准智能科技有限公司 | Bone mineral density measuring method, device, equipment and storage medium |
| CN116681627A (en) * | 2023-08-03 | 2023-09-01 | 佛山科学技术学院 | Cross-scale fusion self-adaptive underwater image generation countermeasure enhancement method |
| CN117115600A (en) * | 2023-09-05 | 2023-11-24 | 佳讯飞鸿（北京）智能科技研究院有限公司 | No-reference image quality evaluation method and device and electronic equipment |
| CN117314872A (en) * | 2023-10-12 | 2023-12-29 | 武汉大学人民医院（湖北省人民医院） | Intelligent segmentation method and device for retina image |
| CN117475121A (en) * | 2023-12-04 | 2024-01-30 | 河南科技大学 | Image processing system and image processing method |
Non-Patent Citations (2)
| Title |
|---|
| LAUREN N. HECKELMAN 等: "Design and validation of a semi-automatic bone segmentation algorithm from MRI to improve research efficiency", 《SCIENTIFIC REPORTS》, 12 May 2022 (2022-05-12), pages 1 - 10 * |
| 杜威、陈胜: "基于Hessian自适应增强的口腔结构光图像处理", 《控制工程》, vol. 30, no. 3, 31 March 2023 (2023-03-31), pages 543 - 551 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108665456B (en) | Method and system for real-time marking of breast ultrasound lesion region based on artificial intelligence | |
| CN109685060B (en) | Image processing method and device | |
| Shen et al. | An automated lung segmentation approach using bidirectional chain codes to improve nodule detection accuracy | |
| CN108133476B (en) | Method and system for automatically detecting pulmonary nodules | |
| CN105894517B (en) | CT image liver segmentation method and system based on feature learning | |
| CN111862044B (en) | Ultrasonic image processing method, ultrasonic image processing device, computer equipment and storage medium | |
| CN108062749B (en) | Identification method and device for levator ani fissure hole and electronic equipment | |
| CN111507965A (en) | Novel coronavirus pneumonia focus detection method, system, device and storage medium | |
| CN111598875A (en) | Method, system and device for building thyroid nodule automatic detection model | |
| JP2005296605A (en) | Method of segmenting a radiographic image into diagnostically relevant and diagnostically irrelevant regions | |
| CN110415208A (en) | A kind of adaptive targets detection method and its device, equipment, storage medium | |
| CN109241867B (en) | Method and device for recognizing digital rock core image by adopting artificial intelligence algorithm | |
| CN106780520B (en) | Automatic extraction method of vertebrae in MRI (magnetic resonance imaging) lumbar vertebra image | |
| CN114240978B (en) | Cell edge segmentation method and device based on adaptive morphology | |
| CN108305247B (en) | Method for detecting tissue hardness based on CT image gray value | |
| CN113393454A (en) | Method and device for segmenting pathological target examples in biopsy tissues | |
| CN114092450A (en) | Real-time image segmentation method, system and device based on gastroscopy video | |
| CN116758336A (en) | Medical image intelligent analysis system based on artificial intelligence | |
| CN114581709A (en) | Model training, method, apparatus, and medium for recognizing target in medical image | |
| CN111612749B (en) | Focus detection method and device based on lung image | |
| CN111401102A (en) | Deep learning model training method and device, electronic equipment and storage medium | |
| CN101847260A (en) | Image processing equipment, image processing method and program | |
| CN114092470B (en) | Deep learning-based automatic detection method and device for pulmonary fissure | |
| CN117952962A (en) | Bone mineral density detection image processing method and system | |
| CN110008881A (en) | The recognition methods of the milk cow behavior of multiple mobile object and device |
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 |