Digital pathological scanner capable of automatically identifying and marking lesion area
By combining image preprocessing, feature extraction and classification recognition modules, using convolutional neural networks and deep learning algorithms, automatic identification and labeling of digital pathology scanners are realized, solving the problem of insufficient recognition accuracy and labeling accuracy caused by the diversity of pathological slice types, and improving the efficiency and accuracy of pathological diagnosis.
Patent Information
- Application Number
- CN202510467678.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-15
- Publication Date
- 2025-07-25
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing digital pathology scanners have shortcomings in automatically identifying and labeling lesion areas, making it difficult to adapt to different types of pathological sections, resulting in limited recognition accuracy and labeling accuracy.
The shell, micro-host, touch screen, scanning components, imaging components and intelligent analysis systems are adopted, combined with image preprocessing, feature extraction, classification recognition and labeling modules, and convolutional neural network, deep learning and machine learning algorithms are used to realize automatic identification and labeling of pathological slices.
Through automated scanning, imaging and identification processes, the pathological diagnosis time is shortened, work efficiency and diagnostic accuracy are improved, and high-precision identification and labeling of lesion areas are achieved.
Smart Images
Figure CN120376022A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of digital pathology scanners, and particularly to a digital pathology scanner capable of automatically identifying and marking lesion areas. Background Art
[0002] A digital pathology scanner is an advanced medical device that can quickly convert two-dimensional images of paraffin sections into digital high-definition images and use image processing technology to quantitatively analyze lesions, helping pathologists diagnose diseases more accurately and being widely used in fields such as biopharmaceuticals, clinical diagnosis, environmental detection, and life sciences; Traditional pathological diagnosis relies on pathologists to manually observe and judge tissue sections through a microscope. However, this method has problems such as strong subjectivity, low efficiency, and easy fatigue. With the development of computer technology and artificial intelligence, digital pathology scanners have emerged. Although existing digital pathology scanners have achieved the digitization of tissue sections and image processing, there are still deficiencies in automatically identifying and marking lesion areas. Since the tissue morphologies on pathological sections are diverse, and the characteristics such as the size, shape, and color of lesion areas are also different, it makes it difficult for automatic recognition algorithms to adapt to all types of pathological sections, resulting in limited recognition accuracy and marking precision. Summary of the Invention
[0003] The purpose of the present invention is to provide a digital pathology scanner capable of automatically identifying and marking lesion areas, which has the advantages of automatically identifying and marking lesion areas of human tissues and cells to improve the efficiency and accuracy of pathological diagnosis, and solves the problem that due to the diverse tissue morphologies on pathological sections and the different characteristics such as the size, shape, and color of lesion areas, it makes it difficult for automatic recognition algorithms to adapt to all types of pathological sections, resulting in limited recognition accuracy and marking precision.
[0004] To achieve the above object, the present invention provides the following technical solutions: A digital pathology scanner capable of automatically identifying and annotating lesion areas, comprising a housing, a micro host, a touch screen, a scanning component, an imaging component and an intelligent analysis system. The housing is used to provide a pathology scanning space. The micro host is used to coordinate and control the scanning operation. The touch screen is used to display scanning data. The scanning component is used to scan the pathology section. The imaging component is used to scan and image the pathology edge. The intelligent analysis system is used to identify and annotate the pathology edge. The intelligent analysis system includes an image preprocessing module, a feature extraction module, a classification and recognition module and an annotation module. The image preprocessing module is used to clean, organize and format the image. The feature extraction module is used to extract useful features in the image. The classification and recognition module is used to classify and recognize the extracted information. The annotation module is used to annotate the lesion area. The imaging component is installed on the surface of the scanning component. The scanning component is installed in the inner cavity of the housing. The touch screen is installed on the front of the housing. The micro host is installed on one side of the housing. The output and input ends of the scanning component, the imaging component, the intelligent analysis system and the touch screen are all connected to the micro host.
[0005] Preferably, the preprocessing steps of the image preprocessing module include image enhancement, noise removal, image smoothing, image sharpening, image segmentation, image matching and image compression. It is based on the convolutional neural network algorithm. The feature extraction module is based on the deep learning algorithm. The classification and recognition module is based on the machine learning algorithm. The annotation module is divided into automatic annotation and manual annotation. The automatic annotation is based on the supervised learning algorithm.
[0006] Preferably, the scanning component includes a base, a bracket, a carrier plate, an X-axis moving platform, a Y-axis moving platform and a Z-axis moving platform. The bracket is fixed on the top of the base. The base is fixedly connected to the inner wall of the housing.
[0007] Preferably, the carrier plate is installed on the surface of the Y-axis moving platform. The carrier plate is used to carry the pathology edge. The Y-axis moving platform is used to adjust the position of the pathology edge.
[0008] Preferably, the X-axis moving platform is fixed on the top of the bracket. The Z-axis moving platform is installed on the surface of the X-axis moving platform. The X-axis moving platform and the Z-axis moving platform are used to adjust the position of the imaging component.
[0009] Preferably, the imaging component includes a support plate, a CCD camera, a lens and a light source. The support plate is installed on the surface of the Z-axis moving platform. The CCD camera is fixed on the surface of the support plate. The lens and the light source are both installed at the bottom of the CCD camera.
[0010] Preferably, the calculation formula of the convolutional neural network includes convolution operation, pooling operation, and calculation of the fully connected layer, as shown below: 1) Convolution operation calculation:
[0011] Where: N represents the size of the output feature map; W represents the size of the input feature map; F represents the size of the convolution kernel; P represents the size of the padding; S represents the size of the stride; 2) Pooling operation: The pooling operation is used to reduce the size of the feature map and extract important features. The pooling operation includes max pooling and average pooling. The pooling operation does not involve complex mathematical formulas, but downsamples the input feature map: Max pooling: Select the maximum value within the pooling window as the output; Average pooling: Calculate the average value within the pooling window as the output; 3) Calculation formula of the fully connected layer: The fully connected layer connects all neurons in the previous layer to all neurons in the current layer, and its calculation formula is usually expressed as: Output vector calculation formula: y = Wx + by = Wx + by = Wx + b Where: y represents the output vector; W represents the weight matrix; x represents the input vector; b represents the bias vector.
[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. Through the cooperation of the housing, the micro host, the touch screen, the scanning component, the imaging component, and the intelligent analysis system, the present invention realizes the automatic recognition and annotation of pathological sections. The automated scanning, imaging, and recognition processes greatly shorten the time of pathological diagnosis, improve the work efficiency of operators, and through image preprocessing and feature extraction technologies, can more accurately identify and extract the features of the lesion area. The classification and recognition module is based on machine learning algorithms and can more accurately judge the type and nature of the lesion area, thereby improving the accuracy of diagnosis.
[0013] 2. According to the classification and recognition results, the present invention automatically labels the lesion area through a labeling module. The labeling module has two modes: automatic labeling and manual labeling. The automatic labeling is based on a supervised learning algorithm and can automatically label the lesion area, greatly improving the labeling efficiency. The intelligent analysis system is based on convolutional neural network algorithms, deep learning algorithms, and machine learning algorithms, realizing the automatic recognition and labeling of pathological sections and enhancing the automation level of the scanner. BRIEF DESCRIPTION OF THE DRAWINGS
[0014] Figure 1 is a front view structural schematic diagram of the present invention; Figure 2 is a sectional view structural schematic diagram of the housing of the present invention; Figure 3 is a front view structural schematic diagram of the scanning component of the present invention; Figure 4 is a system flow structural schematic diagram of the present invention.
[0015] In the figures: 1, housing; 2, micro host; 3, touch screen; 4, scanning component; 41, base; 42, bracket; 43, carrier plate; 44, X-axis moving platform; 45, Y-axis moving platform; 46, Z-axis moving platform; 5, imaging component; 51, support plate; 52, CCD camera; 53, lens; 54, light source; 6, intelligent analysis system; 61, image preprocessing module; 62, feature extraction module; 63, classification and recognition module; 64, labeling module. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.
[0017] Embodiment 1 As Figures 1-4As shown, this is the first embodiment of the present invention. This embodiment provides a digital pathology scanner capable of automatically identifying and annotating lesion areas, including a housing 1, a micro host 2, a touch screen 3, a scanning component 4, an imaging component 5, and an intelligent analysis system 6. The housing 1 is used to provide a pathology scanning space. The micro host 2 is used to coordinate and control the scanning operation. The touch screen 3 is used to display scanning data. The scanning component 4 is used to scan pathology sections. The imaging component 5 is used to scan and image the pathology edges. The intelligent analysis system 6 is used to identify and annotate the pathology edges. The intelligent analysis system 6 includes an image preprocessing module 61, a feature extraction module 62, a classification and recognition module 63, and an annotation module 64. The image preprocessing module 61 is used to clean, organize, and format the images. The feature extraction module 62 is used to extract useful features in the images. The classification and recognition module 63 is used to classify and recognize the extracted information. The annotation module 64 is used to annotate the lesion areas. The imaging component 5 is installed on the surface of the scanning component 4. The scanning component 4 is installed in the inner cavity of the housing 1. The touch screen 3 is installed on the front of the housing 1. The micro host 2 is installed on one side of the housing 1. The output and input ends of the scanning component 4, the imaging component 5, the intelligent analysis system 6, and the touch screen 3 are all connected to the micro host 2.
[0018] As Figures 1-4As shown, the housing 1 is used to provide a closed and stable pathological scanning space to ensure that the scanning process is not interfered by the outside world. The micro host 2 serves as the control center to coordinate the entire scanning process, including starting the scanning, adjusting parameters, etc. The touch screen 3 is used for operator interaction, displaying information such as scanning progress and parameter settings, and receiving operator instructions. The scanning component 4 is responsible for carrying and moving the pathological section. The imaging component 5 performs high-resolution imaging on the pathological section to ensure clear and accurate image data is obtained. The image preprocessing module 61 performs a series of processes on the acquired image data, including image enhancement, noise removal, image smoothing, image sharpening, etc., to improve the image quality and lay a foundation for subsequent feature extraction and classification recognition. The feature extraction module 62, based on deep learning algorithms, can automatically learn and extract useful features in the image, improving the recognition accuracy of the lesion area. The classification and recognition module 63, based on machine learning algorithms, classifies and recognizes the extracted features to judge the type and nature of the lesion area. The annotation module 64 automatically annotates the lesion area according to the results of classification and recognition, and can select the automatic annotation or manual annotation mode to meet the needs of different scenarios. The micro host 2, as the control center of the entire system, is responsible for coordinating the work of each component to ensure the smooth progress of steps such as scanning, imaging, preprocessing, feature extraction, classification and recognition, and annotation. The automated scanning, imaging, and recognition processes greatly shorten the time of pathological diagnosis and improve the work efficiency of operators. Through image preprocessing and feature extraction technologies, the features of the lesion area can be more accurately identified and extracted. Through the cooperation of the intelligent analysis system 6, the image preprocessing module 61, the feature extraction module 62, the classification and recognition module 63, and the annotation module 64, the automatic recognition and annotation of pathological sections can be realized, thus effectively solving the problem of the existing digital pathological scanners having deficiencies in automatically recognizing and annotating lesion areas.
[0019] Embodiment 2 Referring to Figures 1-4 , this is the second embodiment of the present invention, and this embodiment is based on the previous embodiment.
[0020] In this embodiment, the preprocessing steps of the image preprocessing module 61 include image enhancement, noise removal, image smoothing, image sharpening, image segmentation, image matching, and image compression. It is based on the convolutional neural network algorithm. The feature extraction module 62 is based on deep learning algorithms, the classification and recognition module 63 is based on machine learning algorithms, and the annotation module 64 is divided into automatic annotation and manual annotation. The automatic annotation is based on supervised learning algorithms.
[0021] The calculation formula of the convolutional neural network includes convolutional operation, pooling operation, and calculation of the fully connected layer, as follows: 1) Convolutional operation calculation:
[0022] Among them: N represents the size of the output feature map; W represents the size of the input feature map; F represents the size of the convolutional kernel; P represents the size of the padding; S represents the size of the stride; 2) Pooling operation: The pooling operation is used to reduce the size of the feature map and extract important features. The pooling operation includes max pooling and average pooling. The pooling operation does not involve complex mathematical formulas, but downsamples the input feature map: Max pooling: Select the maximum value within the pooling window as the output; Average pooling: Calculate the average value within the pooling window as the output; 3) Calculation formula of the fully connected layer: The fully connected layer connects all neurons in the previous layer to all neurons in the current layer. Its calculation formula is usually expressed as: Calculation formula of the output vector: y = Wx + b y = Wx + b y = Wx + b Among them: y represents the output vector; W represents the weight matrix; x represents the input vector; b represents the bias vector.
[0023] Such as Figures 1-4As shown, the image preprocessing module 61 performs a series of processes on the acquired image data, including image enhancement, noise removal, image smoothing, image sharpening, etc. The preprocessing steps are based on the convolutional network neural algorithm, which can automatically learn and adapt to different types of pathological images, improving the accuracy and efficiency of processing. The feature extraction module 62 extracts useful features from the preprocessed images, such as the morphology, color, texture, etc. of the lesion area. Based on the deep learning algorithm, it can automatically learn and extract complex image features, providing strong support for subsequent classification and recognition. The classification and recognition module 63 classifies and recognizes the extracted features, determines the type and nature of the lesion area. Based on the machine learning algorithm, it can automatically learn and adapt to different classification and recognition tasks, improving the accuracy and robustness of recognition. The annotation module 64 annotates the identified lesion area so that the operator can intuitively view and analyze the results. It is divided into two types: automatic annotation and manual annotation. The automatic annotation is based on the supervised learning algorithm and can automatically annotate the lesion area. The manual annotation allows the operator to adjust and modify according to needs. Its calculation formula and algorithm design based on the convolutional neural network make the entire system highly automated, intelligent and accurate, realizing the automated and intelligent processing of pathological sections, effectively improving the diagnostic efficiency and diagnostic accuracy.
[0024] Embodiment 3 Refer to Figures 1-3 , which is the third embodiment of the present invention. This embodiment is based on the first two embodiments.
[0025] In this embodiment, the scanning component 4 includes a base 41, a bracket 42, a carrier plate 43, an X-axis moving platform 44, a Y-axis moving platform 45, and a Z-axis moving platform 46. The bracket 42 is fixed to the top of the base 41, and the base 41 is fixedly connected to the inner wall of the housing 1.
[0026] The carrier plate 43 is installed on the surface of the Y-axis moving platform 45. The carrier plate 43 is used to carry the pathological section edge, and the Y-axis moving platform 45 is used to adjust the position of the pathological section edge.
[0027] The X-axis moving platform 44 is fixed to the top of the bracket 42, and the Z-axis moving platform 46 is installed on the surface of the X-axis moving platform 44. The X-axis moving platform 44 and the Z-axis moving platform 46 are used to adjust the position of the imaging component 5.
[0028] The imaging component 5 includes a support plate 51, a CCD camera 52, a lens 53, and a light source 54. The support plate 51 is installed on the surface of the Z-axis moving platform 46. The CCD camera 52 is fixed to the surface of the support plate 51. The lens 53 and the light source 54 are both installed at the bottom of the CCD camera 52.
[0029] As Figures 1-3As shown in the figure, the X-axis moving platform 44, Y-axis moving platform 45, and Z-axis moving platform 46 are responsible for carrying and moving the pathological section, enabling it to be accurately positioned within the scanning area for high-quality scanning. Through the design of the three-axis moving platform, precise movement and positioning of the pathological section are achieved, improving the accuracy and flexibility of scanning. The CCD camera 52, in conjunction with the lens 53 and light source 54, images the pathological section, converting the optical image into a digital signal for subsequent analysis and processing. The CCD camera 52 has high resolution and sensitivity, and the lens 53 and light source 54 can ensure the clarity and uniformity of imaging.
[0030] During use, the housing 1 provides a closed and stable pathological scanning space to ensure that the scanning process is not interfered with by the outside world. The micro host 2 serves as the control center, coordinating the entire scanning process, including starting the scan, adjusting parameters, etc. The touch screen 3 is used for operator interaction, displaying information such as the scanning progress and parameter settings, and receiving operator instructions. The scanning component 4 is responsible for carrying and moving the pathological section. Through the precise control of the X-axis moving platform 44, Y-axis moving platform 45, and Z-axis moving platform 46, full-dimensional and high-precision scanning of the pathological section is achieved. The CCD camera 52 in the imaging component 5, in conjunction with the lens 53 and light source 54, performs high-resolution imaging of the pathological section. During the imaging process, the Z-axis moving platform 46 can adjust the distance between the camera and the section to obtain the clearest image. The image preprocessing module 61 performs a series of preprocessing operations on the acquired image data, including image enhancement, noise removal, image smoothing, image sharpening, etc., to improve the image quality and lay a foundation for subsequent feature extraction and classification recognition. The feature extraction module 62 uses deep learning algorithms to extract useful features from the preprocessed image. These features can reflect the unique properties of the lesion area, such as shape, color, texture, etc. The classification and recognition module 63, based on machine learning algorithms, classifies and recognizes the extracted features. By comparing and learning the features of a large number of known lesion areas, the module can accurately determine whether there is a lesion area in the currently scanned pathological section and identify its type. The annotation module 64 automatically annotates the lesion area according to the results of classification and recognition. The annotation method can be selected as automatic annotation or manual annotation to meet the needs of different scenarios. Automatic annotation is based on supervised learning algorithms and can accurately mark the location and scope of the lesion area. The annotated image and data are displayed to the operator through the touch screen 3. The operator can view the detailed scanning results and annotation information and conduct further diagnosis and analysis. The scanning results and annotation information can be stored in the micro host 2 or a connected external storage device for subsequent reference and analysis. At the same time, these data can also be transmitted to a remote server or medical information system through a network interface or other communication methods to achieve data sharing and remote diagnosis.
[0031] The standard parts used in this application document can all be purchased from the market, and can also be customized according to the descriptions in the specification and drawings. The specific connection methods of each part all adopt conventional means such as bolts, rivets, welding, etc. that are mature in the prior art. The machines, parts and equipment all adopt conventional models in the prior art. The control method is to automatically control through a controller, and the control circuit of the controller can be realized by simple programming by those skilled in the art, which belongs to the common general knowledge in this field. And this application document is mainly used to protect the mechanical device, so the control method and circuit connection will not be explained in detail in this application document.
[0032] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "comprising a..." does not exclude the existence of additional identical elements in the process, method, article or device including the said element.
[0033] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A digital pathology scanner capable of automatically identifying and marking lesion areas, comprising a housing (1), a micro host (2), a touch screen (3), a scanning component (4), an imaging component (5) and an intelligent analysis system (6), characterized in that: The housing (1) is used to provide a pathological scanning space, the micro host (2) is used to coordinate and control the scanning operation, the touch screen (3) is used to display scanning data, the scanning component (4) is used to scan pathological sections, the imaging component (5) is used to image the edges of pathological sections, the intelligent analysis system (6) is used to identify and label the edges of pathological sections. The intelligent analysis system (6) includes an image preprocessing module (61), a feature extraction module (62), a classification and recognition module (63) and a labeling module (64). The image preprocessing module (61) is used to clean, organize and format images. The feature extraction module (62) is used to extract useful features from images. The classification and recognition module (63) is used to classify and recognize the extracted information. The labeling module (64) is used to label the lesion area. The imaging component (5) is installed on the surface of the scanning component (4), the scanning component (4) is installed in the inner cavity of the housing (1), the touch screen (3) is installed on the front of the housing (1), the micro host (2) is installed on one side of the housing (1). The output ends and input ends of the scanning component (4), the imaging component (5), the intelligent analysis system (6) and the touch screen (3) are all connected to the micro host (2).
2. The digital pathology scanner capable of automatically identifying and labeling lesion areas according to claim 1, wherein: The preprocessing steps of the image preprocessing module (61) include image enhancement, noise removal, image smoothing, image sharpening, image segmentation, image matching and image compression. It is based on the convolutional neural network algorithm. The feature extraction module (62) is based on the deep learning algorithm. The classification and recognition module (63) is based on the machine learning algorithm. The labeling module (64) is divided into automatic labeling and manual labeling. The automatic labeling is based on the supervised learning algorithm.
3. The digital pathology scanner capable of automatically identifying and marking lesion areas according to claim 1, characterized in that: The scanning component (4) includes a base (41), a bracket (42), a carrier plate (43), an X-axis moving platform (44), a Y-axis moving platform (45) and a Z-axis moving platform (46). The bracket (42) is fixed to the top of the base (41), and the base (41) is fixedly connected to the inner wall of the housing (1).
4. A digital pathology scanner capable of automatically identifying and marking lesion areas according to claim 3, characterized in that: The carrier plate (43) is installed on the surface of the Y-axis moving platform (45). The carrier plate (43) is used to carry the edges of pathological sections, and the Y-axis moving platform (45) is used to adjust the position of the edges of pathological sections.
5. The digital pathology scanner capable of automatically identifying and marking lesion areas according to claim 3, wherein: The X-axis moving platform (44) is fixed to the top of the bracket (42), and the Z-axis moving platform (46) is installed on the surface of the X-axis moving platform (44). The X-axis moving platform (44) and the Z-axis moving platform (46) are used to adjust the position of the imaging component (5).
6. The digital pathology scanner capable of automatically identifying and marking lesion areas according to claim 3, characterized in that: The imaging component (5) includes a support plate (51), a CCD camera (52), a lens (53) and a light source (54). The support plate (51) is installed on the surface of the Z-axis moving platform (46), the CCD camera (52) is fixed to the surface of the support plate (51), and the lens (53) and the light source (54) are both installed at the bottom of the CCD camera (52).
7. The digital pathology scanner capable of automatically identifying and marking lesion areas according to claim 2, wherein: The calculation formula of the convolutional neural network includes convolution operation, pooling operation and the calculation of the fully connected layer, as follows: 1) Convolution operation calculation: Where: N represents the size of the output feature map; W represents the size of the input feature map; F represents the size of the convolution kernel; P represents the size of the padding; S represents the size of the stride; 2) Pooling operation: The pooling operation is used to reduce the size of the feature map and extract important features. The pooling operation includes max pooling and average pooling. The pooling operation does not involve complex mathematical formulas, but downsamples the input feature map: Max pooling: Select the maximum value within the pooling window as the output; Average pooling: Calculate the average value within the pooling window as the output; 3) Fully connected layer calculation formula: The fully connected layer connects all neurons in the previous layer to all neurons in the current layer, and its calculation formula is usually expressed as: Output vector calculation formula: y = Wx + by = Wx + by = Wx + b Where: y represents the output vector; W represents the weight matrix; x represents the input vector; b represents the bias vector.