A lesion detection method and system based on image recognition

Abstract

The present invention discloses a lesion detection method and system based on image recognition. The method includes: acquiring a medical image to be detected and inputting the medical image into a pre-trained lesion area localization model to determine a preliminary lesion area containing a lesion in the medical image to be detected; acquiring a preliminary lesion area image of the preliminary lesion area and inputting the preliminary lesion area image into a pre-trained lesion area extraction model to obtain a target lesion area image of the lesion; and inputting the target lesion area image into a pre-trained lesion recognition model to obtain the lesion type in the medical image to be detected. The present invention solves the problem of low accuracy in lesion detection in the prior art.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and particularly relates to a lesion detection method and system based on image recognition. Background Art

[0002] Nowadays, medical image diagnosis mainly relies on traditional means such as doctors' subjective visual analysis and manual annotation. Doctors identify and diagnose lesions by combining imaging features with experience and clinical knowledge. However, manual analysis and annotation of images are not only time-consuming, but also when faced with a large number of images, doctors are prone to fatigue, which may lead to missed diagnoses or misdiagnoses.

[0003] Due to these limitations, medical image recognition has been gradually introduced to improve the accuracy and efficiency of lesion detection by means of deep learning and computer vision technologies. By adopting image recognition technology, the burden on medical staff can be significantly reduced, and the objectivity and consistency of diagnosis can be improved.

[0004] Currently, when performing lesion detection, medical images are directly detected to identify lesion information. However, when the lesion range in the image is small and there are many interference factors, if it is necessary to identify the lesion area and specific type at one time, there may be a problem of inaccurate lesion detection. Summary of the Invention

[0005] In view of this, an object of the present invention is to provide a lesion detection method and system based on image recognition, aiming to solve the problem of low accuracy in lesion detection in the prior art.

[0006] An object of the present invention is to provide a lesion detection method based on image recognition for detecting lesions in medical images, and the method includes:

[0007] Obtain a medical image to be detected, and input the medical image into a pre-trained lesion area localization model to determine a preliminary lesion area containing a lesion in the medical image to be detected;

[0008] Obtain a preliminary lesion area image of the preliminary lesion area, and input the preliminary lesion area image into a pre-trained lesion area extraction model to obtain a target lesion area image of the lesion;

[0009] Input the target lesion area image into a pre-trained lesion recognition model to obtain the type of lesion in the medical image to be detected.

[0010] Further, in the above-mentioned lesion detection method based on image recognition, the training process of the lesion area localization model includes:

[0011] Collect historical medical images containing lesions to obtain the training set and validation set for the lesion area localization model;

[0012] Construct a first detection algorithm based on preset parameters, and train using the first detection algorithm according to the training set and validation set of the lesion area localization model;

[0013] Until the performance of the lesion area localization model meets the preset criteria, so as to train and obtain the lesion area localization model.

[0014] Furthermore, for the above-mentioned lesion detection method based on image recognition, wherein the training process of the lesion area extraction model includes:

[0015] Collect historical preliminary lesion area images to obtain the training set and validation set for the lesion area extraction model;

[0016] Construct a second detection algorithm based on preset parameters, and train using the second detection algorithm according to the training set and validation set of the lesion area extraction model;

[0017] Until the performance of the lesion area extraction model meets the preset criteria, so as to train and obtain the lesion area extraction model.

[0018] Furthermore, for the above-mentioned lesion detection method based on image recognition, wherein the training process of the lesion recognition model includes:

[0019] Construct a preset convolutional neural network, and collect a preset number of historical target lesion area images and corresponding lesion types as training samples;

[0020] Respectively use the target lesion area image containing the lesion and the corresponding lesion type as the input and output of the convolutional neural network, and perform deep learning training on the convolutional neural network until the recognition result output by the convolutional neural network meets the set accuracy, so as to obtain the lesion recognition model.

[0021] Furthermore, for the above-mentioned lesion detection method based on image recognition, before the step of collecting historical medical images containing lesions to obtain the training set and validation set for the lesion area localization model, it further includes:

[0022] Obtain historical medical images, and respectively extract the lesion area images and non-lesion area images from the historical medical images;

[0023] Randomly rotate, non-uniformly scale, and distort the lesion area image to obtain a first lesion area image;

[0024] Determine the target historical medical image according to the first lesion area image and the non-lesion area image.

[0025] Further, in the above-mentioned lesion detection method based on image recognition, after the steps of acquiring historical medical images and respectively extracting lesion area images and non-lesion area images from the historical medical images, the following steps are further included:

[0026] Obtain the contour information of the current lesion area image, and determine the diffusion law of the lesion area according to the lesion type of the lesion in the lesion area image;

[0027] Perform contour expansion on the lesion area image according to the contour information and the diffusion law to obtain an expanded lesion area image;

[0028] Obtain the center point of the lesion area image, establish a coordinate system with the center point as the coordinate origin, fill the features in the lesion area image into the expanded lesion area image in the same quadrant to obtain a target expanded lesion area image, and then determine the target historical medical image with the non-lesion area image.

[0029] Further, in the above-mentioned lesion detection method based on image recognition, the step of filling the features in the lesion area image into the expanded lesion area image in the same quadrant to obtain a target expanded lesion area image includes:

[0030] Obtain the rectangular area enclosing the expanded lesion area image, and perform grid division on the rectangular area;

[0031] Obtain the central grid of the lesion area image in the same quadrant, and determine the target central grid corresponding to the central grid in the expanded lesion area image;

[0032] And based on the positional relationship between the central grid and other grids, and the target central grid, fill the features of the other grids into the grids in the expanded lesion area image that do not contain features to obtain a target expanded lesion area image.

[0033] Another object of the present invention is to provide a lesion detection system based on image recognition for detecting lesions in medical images. The system includes:

[0034] An acquisition module, configured to acquire a medical image to be detected, and input the medical image into a pre-trained lesion area localization model to determine a preliminary lesion area containing a lesion in the medical image to be detected;

[0035] An extraction module, configured to acquire a preliminary lesion area image of the preliminary lesion area, and input the preliminary lesion area image into a pre-trained lesion area extraction model to obtain a target lesion area image of the lesion;

[0036] A detection module, configured to input the image of the target lesion area into a pre-trained lesion recognition model to obtain the lesion type in the medical image to be detected.

[0037] Another object of the present invention is to provide a readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the method described in any one of the above are implemented.

[0038] Another object of the present invention is to provide an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the above method are implemented.

[0039] The present invention first uses a trained lesion area localization model to preliminarily locate the lesion area and initially identify the area where lesions may exist. When encountering lesions with a small range, the lesion area extraction model is then used to lock the specific location of the lesions with a small range. Finally, the lesion recognition model is used to identify the type of the lesions, realizing the detection of the lesion area and type. Moreover, due to the multi-level recognition method adopted, when identifying the lesion type, interference from other factors is avoided, and the accuracy of lesion recognition is improved. The problem of low accuracy in lesion detection in the prior art is solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a flowchart of a lesion detection method based on image recognition in the first embodiment of the present invention;

[0041] Figure 2 It is a structural block diagram of a lesion detection system based on image recognition in the third embodiment of the present invention.

[0042] The following specific embodiments will further illustrate the present invention in conjunction with the above drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0043] To facilitate the understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0044] It should be noted that when an element is referred to as being "fixed to" another element, it can be directly on the other element or there can also be an intermediate element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be an intermediate element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0045] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this invention belongs. The terms used herein in the specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0046] The following will specifically describe in detail how to accurately detect small lesions in medical images in combination with specific embodiments and drawings.

[0047] Embodiment 1

[0048] Please refer to Figure 1 , which shows a lesion detection method based on image recognition in the first embodiment of the present invention. The method includes steps S10 to S12.

[0049] Step S10, obtain the medical image to be detected, and input the medical image into a pre-trained lesion area localization model to determine a preliminary lesion area containing lesions in the medical image to be detected.

[0050] Among them, the medical image is an image containing lesions collected by a medical device, such as a CT image. The lesion area localization model is used to roughly estimate the lesion area in the medical image and initially identify the area where lesions may exist.

[0051] Specifically, a relatively lightweight object detection algorithm (such as YOLOv5, SSD, etc.) can be selected to quickly identify the area where lesions may exist. Exemplarily, when training the lesion area localization model, historical medical images containing lesions are collected to obtain the training set and validation set of the lesion area localization model; a first detection algorithm (such as YOLOv5, SSD, etc.) is constructed based on preset parameters, and the first detection algorithm is used for training according to the training set and validation set of the lesion area localization model. Among them, a suitable loss function (such as a combination of cross-entropy loss and bounding box regression loss) can be selected for training, and optimizers such as Adam and SGD are used to adjust hyperparameters such as the learning rate; iterative training is performed for multiple cycles on the training set, and the performance of the model is monitored using the validation set to prevent overfitting until the performance of the lesion area localization model meets the preset standard, so as to train the lesion area localization model.

[0052] Step S11: Obtain the preliminary lesion area image of the preliminary lesion area, and input the preliminary lesion area image into a pre-trained lesion area extraction model to obtain the target lesion area image of the lesion.

[0053] Specifically, after determining the preliminary lesion area, intercept the preliminary lesion area image, and use the lesion area extraction model to obtain a more accurate target lesion area image of the lesion. Specifically, collect historical preliminary lesion area images to obtain the training set and validation set of the lesion area extraction model; construct a second detection algorithm based on preset parameters, and use the second detection algorithm to train according to the training set and validation set of the lesion area extraction model; until the performance of the lesion area extraction model meets the preset standard to train the lesion area extraction model. Among them, the lesion area extraction model aims to accurately identify the specific contour of the lesion. Network models such as U-Net and Mask R-CNN can be used. Collect a large number of training data sets, accurately label the contours of the lesions, usually using polygons or other forms to ensure that the labels accurately reflect the boundaries of the lesions, and finally train to obtain the final lesion area extraction model.

[0054] In addition, in some optional embodiments of the present invention, since the sizes of the extracted preliminary lesion area images are different, in order to ensure that the picture sizes of the data sets used for training are the same and can ensure the original features, the obtained preliminary lesion area images can be normalized, such as unified scaling processing.

[0055] Step S12: Input the target lesion area image into a pre-trained lesion recognition model to obtain the lesion type in the medical image to be detected.

[0056] Among them, the lesion recognition model has mastered the internal logic of lesion type recognition. By inputting the target lesion area image into a pre-trained lesion recognition model, the lesion type in the medical image to be detected can be accurately identified.

[0057] Specifically, construct a preset convolutional neural network, collect a preset number of historical target lesion area images and the corresponding lesion types as training samples; respectively use the target lesion area image containing the lesion and the corresponding lesion type as the input and output of the convolutional neural network, and perform deep learning training on the convolutional neural network until the recognition result output by the convolutional neural network meets the set accuracy to obtain the lesion recognition model. Among them, the more detailed process of the selection and training of the neural network is understandable to those skilled in the art and will not be elaborated here.

[0058] In summary, in the above embodiments of the present invention, a lesion detection method based on image recognition first uses a trained lesion area localization model to preliminarily locate the lesion area, initially identifying the area where lesions may exist. When encountering lesions with a small range, the specific position of the lesion with a small range is then locked using the lesion area extraction model. Finally, the type of the lesion is identified using the lesion recognition model, realizing the detection of the lesion area and type. Moreover, due to the multi-level recognition method adopted, other factors are avoided during the lesion type recognition, improving the accuracy of lesion recognition. This solves the problem of low accuracy in existing technologies during lesion detection.

[0059] Embodiment 2

[0060] This embodiment also proposes a lesion detection method based on image recognition. The difference between the lesion detection method based on image recognition in this embodiment and the lesion detection method based on image recognition in Embodiment 1 lies in:

[0061] Before the step of collecting historical medical images containing lesions to obtain the training set and validation set of the lesion area localization model, it further includes:

[0062] Obtain historical medical images, and respectively extract the lesion area images and non-lesion area images in the historical medical images;

[0063] Randomly rotate, non-uniformly scale, and distort the lesion area image to obtain a first lesion area image;

[0064] Determine the target historical medical image according to the first lesion area image and the non-lesion area image.

[0065] In practice, it is relatively difficult to obtain image samples with lesions. To ensure the accuracy of model training, a sufficient variety of sample data is required. Therefore, to greatly enhance the diversity of the dataset, extract the existing lesion area images in the existing historical medical images, randomly rotate the lesions to simulate different orientations, non-uniformly scale the lesions to adapt to lesions of different sizes, and use affine transformation or perspective transformation to change the shape of the lesions to simulate the variation of biological tissues. The lesion area is changed by means of geometric transformation. Finally, the lesion area image and the extracted non-lesion area image are synthesized to obtain the corresponding expanded target historical medical image.

[0066] In addition, in practice, the above methods for enhancing data diversity have a certain degree of randomness. To specifically enhance the diversity of sample data, after extracting the lesion area, the diffusion law of the lesion area can be determined according to the current lesion type, so as to expand the sample diversity by diffusing the current lesion area. Specifically, obtain the contour information of the current lesion area image, and determine the diffusion law of the lesion area according to the lesion type of the lesion in the lesion area image. Expand the contour of the lesion area image according to the contour information and the diffusion law to obtain an expanded lesion area image. At this time, the contour features of the lesion area are extended. To further and completely expand the features of the lesion area, obtain the center point of the lesion area image, establish a coordinate system with the center point as the coordinate origin, and fill the features in the lesion area image into the expanded lesion area image in the same quadrant to obtain a target expanded lesion area image. Then, determine the target historical medical image with the non-lesion area image. Among them, the center point can be the geometric center of the lesion area image, such as the centroid. Establish a coordinate system according to this center point. Among them, each part of the expanded lesion area image and the lesion area image are respectively distributed in different quadrants of the coordinate system. Randomly fill the features of the lesion area image in the same quadrant into the part of the expanded lesion area image that does not contain features, so as to obtain the final target expanded lesion area image.

[0067] In addition, in some optional embodiments of the present invention, the step of filling the features in the lesion area image into the expanded lesion area image in the same quadrant to obtain a target expanded lesion area image includes:

[0068] Obtain the rectangular area enclosing the expanded lesion area image, and perform grid division on the rectangular area;

[0069] Obtain the central grid of the lesion area image in the same quadrant, and determine the target central grid corresponding to the central grid in the expanded lesion area image;

[0070] And based on the positional relationship between the central grid and other grids, and the target central grid, fill the features of the other grids into the grids in the expanded lesion area image that do not contain features to obtain a target expanded lesion area image.

[0071] Among them, when performing feature migration and filling, a grid division method is adopted to divide the lesion area into multiple small grids, and the migration is performed according to the pixel features in each grid. This method can not only preserve the continuity of local features, but also utilize the spatial position for more flexible feature migration in the extended lesion area. In order to effectively perform feature migration, the key lies in how to correspond the grids according to a specific mapping strategy. In the embodiments of the present invention, a center point mapping strategy is adopted. The center mapping is based on the relationship between the center point of the lesion area and the center point of the extended area to find the corresponding target center grid or the position of the center point in the extended area, and each grid in the lesion area is corresponded to the extended lesion area according to the relative position of the grids.

[0072] Exemplarily, when performing filling, the features of the previous lesion area can be retained, or the features of the lesion area can be not retained, that is, the features can be filled in the entire extended lesion area, or the feature filling can be performed after removing the overlapping part with the lesion area. In some optional embodiments of the present invention, a rectangular area can also be formed around the lesion area, so that it can correspond to the rectangular area of the extended lesion area, facilitating the filling between grid features.

[0073] In summary, a lesion detection method based on image recognition in the above embodiments of the present invention first uses a trained lesion area localization model to preliminarily locate the lesion area, initially identifying the area where lesions may exist. When encountering lesions with a small range, the specific position of the lesions with a small range is then locked by using a lesion area extraction model, and finally the type of the lesions is identified by using a lesion recognition model, realizing the detection of the lesion area and type. Moreover, due to the multi-level recognition method adopted, other factors are avoided from interfering when identifying the lesion type, improving the accuracy of lesion recognition. It solves the problem of low accuracy in lesion detection in the prior art.

[0074] Embodiment 3

[0075] Please refer to Figure 2 , which shows a lesion detection system based on image recognition proposed in the third embodiment of the present invention for detecting lesions in medical images. The system includes:

[0076] An acquisition module 100, configured to acquire a medical image to be detected and input the medical image into a pre-trained lesion area localization model to determine a preliminary lesion area containing lesions in the medical image to be detected;

[0077] An extraction module 200, configured to obtain a preliminary lesion area image of a preliminary lesion area, and input the preliminary lesion area image into a pre-trained lesion area extraction model to obtain a target lesion area image of the lesion;

[0078] A detection module 300, configured to input the target lesion area image into a pre-trained lesion recognition model to obtain the type of lesion in the medical image to be detected.

[0079] Furthermore, for the above-mentioned lesion detection system based on image recognition, wherein the training process of the lesion area localization model includes:

[0080] Collect historical medical images containing lesions to obtain a training set and a validation set for the lesion area localization model;

[0081] Construct a first detection algorithm based on preset parameters, and use the first detection algorithm to train according to the training set and the validation set of the lesion area localization model;

[0082] Until the performance of the lesion area localization model meets the preset standard, so as to train the lesion area localization model.

[0083] Furthermore, for the above-mentioned lesion detection system based on image recognition, wherein the training process of the lesion area extraction model includes:

[0084] Collect historical preliminary lesion area images to obtain a training set and a validation set for the lesion area extraction model;

[0085] Construct a second detection algorithm based on preset parameters, and use the second detection algorithm to train according to the training set and the validation set of the lesion area extraction model;

[0086] Until the performance of the lesion area extraction model meets the preset standard, so as to train the lesion area extraction model.

[0087] Furthermore, for the above-mentioned lesion detection system based on image recognition, wherein the training process of the lesion recognition model includes:

[0088] Construct a preset convolutional neural network, and collect a preset number of historical target lesion area images and the corresponding lesion types as training samples;

[0089] Respectively use the target lesion area image containing the lesion and the corresponding lesion type as the input and output of the convolutional neural network, and perform deep learning training on the convolutional neural network until the recognition result output by the convolutional neural network meets the set accuracy, so as to obtain the lesion recognition model.

[0090] Further, in the above-mentioned lesion detection system based on image recognition, before the step of collecting historical medical images containing lesions to obtain the training set and validation set of the lesion area localization model, the following steps are also included:

[0091] Obtain historical medical images, and separately extract the lesion area images and non-lesion area images from the historical medical images;

[0092] Randomly rotate, non-uniformly scale, and distort the lesion area images to obtain the first lesion area images;

[0093] Determine the target historical medical images based on the first lesion area images and the non-lesion area images.

[0094] Further, in the above-mentioned lesion detection system based on image recognition, after the step of obtaining historical medical images and separately extracting the lesion area images and non-lesion area images from the historical medical images, the following steps are also included:

[0095] Obtain the contour information of the current lesion area images, and determine the diffusion law of the lesion area according to the lesion type of the lesions in the lesion area images;

[0096] Expand the contour of the lesion area images according to the contour information and the diffusion law to obtain the expanded lesion area images;

[0097] Obtain the center points of the lesion area images, establish a coordinate system with the center points as the coordinate origins, fill the features in the lesion area images into the expanded lesion area images in the same quadrant to obtain the target expanded lesion area images, and then determine the target historical medical images with the non-lesion area images.

[0098] Further, in the above-mentioned lesion detection system based on image recognition, the step of filling the features in the lesion area images into the expanded lesion area images in the same quadrant to obtain the target expanded lesion area images includes:

[0099] Obtain the rectangular areas enclosing the expanded lesion area images, and perform grid division on the rectangular areas;

[0100] Obtain the central grids of the lesion area images in the same quadrant, and determine the target central grids corresponding to the central grids in the expanded lesion area images;

[0101] And based on the positional relationship between the central grids and other grids, and the target central grids, fill the features of the other grids into the grids in the expanded lesion area images that do not contain features to obtain the target expanded lesion area images.

[0102] The functions or operation steps implemented when the above modules are executed are substantially the same as those in the above method embodiments, and will not be elaborated here.

[0103] Embodiment 4

[0104] On the other hand, the present invention also provides a readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the method described in any one of the above Embodiment 1 to Embodiment 2 are implemented.

[0105] Embodiment 5

[0106] On the other hand, the present invention also provides an electronic device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the method described in any one of the above Embodiment 1 to Embodiment 2 are implemented.

[0107] The technical features of each of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0108] Those skilled in the art can understand that the logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable storage medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or used in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable storage medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device.

[0109] More specific examples (a non-exhaustive list) of computer-readable storage media include the following: electrical connection parts (electronic devices) having one or more wirings, portable computer disk cartridges (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber devices, and portable compact disc read-only memory (CDROM). Additionally, the computer-readable storage media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other suitable processing when necessary, and then stored in a computer memory.

[0110] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0111] In the description of this specification, the description referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0112] The above-described embodiments merely represent several implementation manners of the present invention, and their descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention should be subject to the appended claims.

Claims

1. A lesion detection method based on image recognition, characterized in that, For detecting lesions in medical images, the method includes: Obtain a medical image to be detected, and input the medical image into a pre-trained lesion area localization model to determine a preliminary lesion area containing a lesion in the medical image to be detected; Obtain a preliminary lesion area image of the preliminary lesion area, and input the preliminary lesion area image into a pre-trained lesion area extraction model to obtain a target lesion area image of the lesion; Input the target lesion area image into a pre-trained lesion recognition model to obtain the lesion type in the medical image to be detected; The training process of the lesion area localization model includes: Collect historical medical images containing lesions to obtain a training set and a validation set for the lesion area localization model; Construct a first detection algorithm based on preset parameters, and use the first detection algorithm to train according to the training set and the validation set of the lesion area localization model; Until the performance of the lesion area localization model meets the preset standard to train the lesion area localization model; Before the step of collecting historical medical images containing lesions to obtain a training set and a validation set for the lesion area localization model, it also includes: Obtain historical medical images, and respectively extract lesion area images and non-lesion area images from the historical medical images; Randomly rotate, non-uniformly scale, and distort the lesion area image to obtain a first lesion area image; Determine a target historical medical image according to the first lesion area image and the non-lesion area image; After the step of obtaining historical medical images and respectively extracting lesion area images and non-lesion area images from the historical medical images, it also includes: Obtain the contour information of the current lesion area image, and determine the diffusion law of the lesion area according to the lesion type of the lesion in the lesion area image; Expand the contour of the lesion area image according to the contour information and the diffusion law to obtain an expanded lesion area image; Obtain the center point of the lesion area image, establish a coordinate system with the center point as the coordinate origin, fill the features in the lesion area image into the expanded lesion area image in the same quadrant to obtain a target expanded lesion area image, and then determine a target historical medical image with the non-lesion area image; The step of filling the features in the lesion area image into the expanded lesion area image in the same quadrant to obtain a target expanded lesion area image includes: Obtain a rectangular area enclosing the expanded lesion area image, and perform grid division on the rectangular area; Obtain the central grid of the lesion area image in the same quadrant, and determine the target central grid corresponding to the central grid in the expanded lesion area image; And based on the positional relationship between the central grid and other grids, and the target central grid, fill the features of the other grids into the grids in the expanded lesion area image that do not contain features to obtain a target expanded lesion area image.

2. The method for lesion detection based on image recognition according to claim 1, wherein The training process of the lesion area extraction model includes: Collect historical initial lesion area images to obtain the training set and validation set of the lesion area extraction model; Construct a second detection algorithm based on preset parameters, and use the second detection algorithm to train according to the training set and validation set of the lesion area extraction model; Until the performance of the lesion area extraction model meets the preset standard, so as to train and obtain the lesion area extraction model.

3. The method for lesion detection based on image recognition according to claim 1, wherein The training process of the lesion recognition model includes: Construct a preset convolutional neural network, and collect a preset number of historical target lesion area images and corresponding lesion types as training samples; Respectively use the target lesion area image containing the lesion and the corresponding lesion type as the input and output of the convolutional neural network, and perform deep learning training on the convolutional neural network until the recognition result output by the convolutional neural network meets the set accuracy, so as to obtain the lesion recognition model.

4. A lesion detection system based on image recognition, characterized in that, For detecting lesions in medical images, and for implementing the lesion detection method based on image recognition according to any one of claims 1 to 3, the system includes: An acquisition module, configured to acquire a medical image to be detected, and input the medical image into a pre-trained lesion area localization model to determine an initial lesion area containing a lesion in the medical image to be detected; An extraction module, configured to acquire an initial lesion area image of the initial lesion area, and input the initial lesion area image into a pre-trained lesion area extraction model to obtain a target lesion area image of the lesion; A detection module, configured to input the target lesion area image into a pre-trained lesion recognition model to obtain the lesion type in the medical image to be detected.

5. A readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 3.

6. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 3.

Images