Intelligent analysis and diagnosis support system for chest radiograph

Abstract

The invention relates to the technical field of image analysis, in particular to a chest radiograph image intelligent analysis and diagnosis support system, which comprises an image acquisition module used for receiving and preprocessing a chest radiograph image; the image enhancement module is used for enhancing the preprocessed chest radiograph image through a multi-scale enhancement algorithm; the focus detection module is used for identifying and positioning an abnormal region in the chest radiograph based on multi-layer feature extraction of a deep learning model; the multi-modal information fusion module is used for fusing the result output by the focus detection module with the historical image data, clinical symptoms and laboratory examination results of the patient to generate multi-dimensional diagnosis suggestions; the diagnosis support module is used for providing preliminary diagnosis suggestions and marking high-risk areas so as to assist doctors in performing final diagnosis; according to the invention, efficient intelligent analysis and accurate diagnosis of the chest radiograph image are realized, the accuracy of focus detection can be improved, and more comprehensive diagnosis support can be provided through multi-modal information fusion.

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and particularly to an intelligent analysis and diagnosis support system for chest X-ray images. Background Art

[0002] With the rapid development of medical imaging technology, chest X-ray has become one of the most common imaging means in medical diagnosis. Chest X-ray can effectively detect various chest diseases such as pneumonia, pulmonary tuberculosis, and lung cancer. However, due to the uneven distribution of medical resources, especially in areas with relatively low medical levels, there is a shortage of professional imaging diagnosticians, resulting in difficulty in ensuring the diagnostic accuracy of chest X-rays.

[0003] In addition, when dealing with chest X-ray images, existing intelligent analysis systems often face problems such as uneven image quality and difficulty in identifying complex lesion areas, which to a certain extent limits their wide application in clinical practice.

[0004] Therefore, how to design a system that can intelligently analyze chest X-ray images and provide diagnostic support has become an urgent technical problem to be solved in the current medical imaging field. Summary of the Invention

[0005] The present invention provides an intelligent analysis and diagnosis support system for chest X-ray images.

[0006] The intelligent analysis and diagnosis support system for chest X-ray images includes: An image acquisition module, configured to receive and preprocess chest X-ray images to eliminate noise and artifacts caused by differences in imaging conditions; An image enhancement module, which enhances the preprocessed chest X-ray images through a multi-scale enhancement algorithm to improve the detail performance in the images and make the lesion areas clearer; A lesion detection module, which identifies and locates abnormal areas in chest X-ray images based on multi-layer feature extraction of a deep learning model, including for pulmonary nodules, pleural effusion, and pulmonary shadow areas; A multi-modal information fusion module, configured to fuse the results output by the lesion detection module with the patient's historical imaging data, clinical symptoms, and laboratory test results to generate multi-dimensional diagnostic suggestions; A diagnosis support module, which provides a preliminary diagnosis opinion by combining the lesion detection results with the output of multi-modal information fusion, and marks high-risk areas to assist doctors in making a final diagnosis.

[0007] Optionally, the image acquisition module includes: An automatic exposure control sub-module: used to dynamically adjust exposure parameters during image acquisition based on real-time analysis of the image brightness and contrast conditions to ensure that the imaging quality of chest X-ray images remains consistent under different lighting conditions; Denoising sub-module: Apply the non-local means algorithm to effectively remove the noise generated by low-dose imaging. Through multi-scale denoising processing, important image details are retained. Gray-level equalization sub-module: Used to perform gray-level equalization processing on the image. By adjusting the gray-level distribution of the image, the overall contrast of the chest X-ray is made more uniform.

[0008] Optionally, the image enhancement module includes: Multi-scale pyramid enhancement sub-module: Based on the image pyramid structure, the chest X-ray image is decomposed into image levels of different scales, and each level of the image is enhanced separately to highlight the lesion details existing in each level. Contrast adaptive adjustment sub-module: Apply adaptive contrast enhancement. According to the local region contrast characteristics of the image, the contrast value is dynamically adjusted, so that the low-contrast regions in the image are enhanced, making it easier to identify hidden lesion regions. Tone mapping sub-module: Used to adjust the tone distribution of the image, including enhancing the visual distinguishability of the target lesions by controlling the tone in different types of lesion regions, enabling doctors to more intuitively identify abnormal regions.

[0009] Optionally, the lesion detection module further includes: Deep convolutional neural network sub-module: Use a pre-trained convolutional neural network model to perform multi-level feature extraction on the chest X-ray image. Through the stacking of multiple convolutional layers and pooling layers, high-dimensional features related to lesions are gradually refined. Active learning sub-module: By introducing an active learning mechanism, it allows for the automatic selection of difficult examples in the training dataset for learning under the guidance of doctors, thereby improving the model's recognition ability for uncommon lesions. Region proposal network sub-module: Used to generate candidate regions that may contain lesions. Based on the candidate regions, precise lesion localization is performed. The region proposal network sub-module generates multi-scale candidate boxes to cover lesion regions of different sizes and realizes the recognition of lesions through the combination with deep features.

[0010] Optionally, the multi-modal information fusion module includes: Temporal data analysis sub-module: Used to analyze the patient's historical imaging data and time series features. By detecting the change trends between imaging data at different time points, it helps doctors identify the development process of lesions. Laboratory test result fusion sub-module: Fuses the patient's laboratory test results with the imaging data for integrated analysis. Through the comprehensive processing of multi-dimensional data, it provides a more comprehensive diagnostic basis.

[0011] Optionally, the diagnostic support module includes a risk assessment sub-module. The risk assessment sub-module conducts a comprehensive risk assessment by comprehensively analyzing the lesion detection results, multi-modal information, and patient medical history data. This sub-module predicts the risk of the patient's disease progression through multivariate logistic regression analysis.

[0012] Optionally, the diagnostic support module further includes an interpretability analysis sub-module. The interpretability analysis sub-module enables doctors to understand the diagnostic basis of the system by visually displaying the decision-making process of the deep learning model.

[0013] Advantages of the present invention: In the present invention, the intelligent analysis and diagnostic support system for chest radiograph images realizes efficient and intelligent analysis and accurate diagnosis of chest radiograph images by introducing a structural design with multi-module collaborative work. It can not only improve the accuracy of lesion detection but also provide more comprehensive diagnostic support through multi-modal information fusion.

[0014] In the present invention, the system has an adaptive learning function and can continuously optimize with the accumulation of doctors' feedback, ensuring its persistence and robustness in clinical applications. Through the intelligent user interface design, doctors can perform diagnostic operations more intuitively and conveniently, thus greatly improving the diagnostic efficiency and the treatment experience of patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are only those of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 It is a schematic diagram of the functional modules of the support system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The present invention will be described in detail below in conjunction with the drawings and specific embodiments. At the same time, it should be noted here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments. For some well-known technologies, those skilled in the art can also adopt other alternative methods for implementation; moreover, the drawing part is only for more specifically describing the embodiments and is not intended to specifically limit the present invention.

[0018] It should be noted that in the specification, references to "an embodiment", "embodiments", "exemplary embodiments", "some embodiments", etc. indicate that the described embodiments may include a particular feature, structure, or characteristic, but not every embodiment necessarily includes that particular feature, structure, or characteristic. Additionally, when combining embodiments to describe a particular feature, structure, or characteristic, implementing such a feature, structure, or characteristic in combination with other embodiments (whether explicitly described or not) should be within the knowledge of those skilled in the relevant art.

[0019] Generally, terms can be understood at least in part from their use in context. For example, at least in part depending on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in a singular sense, or can be used to describe a combination of features, structures, or characteristics in a plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey a set of exclusive factors, but rather, at least in part depending on the context, can allow for the existence of other factors that are not necessarily explicitly described.

[0020] As Figure 1 shown, the intelligent analysis and diagnosis support system for chest radiograph images includes: An image acquisition module, configured to receive and preprocess chest radiograph images to eliminate noise and artifacts caused by differences in imaging conditions; An image enhancement module, which enhances the preprocessed chest radiograph images through a multi-scale enhancement algorithm to improve the detail performance in the images and make the lesion areas clearer; A lesion detection module, which, based on multi-layer feature extraction of a deep learning model, identifies and locates abnormal areas in chest radiographs, including areas for pulmonary nodules, pleural effusion, and pulmonary shadows; A multi-modal information fusion module, configured to fuse the results output by the lesion detection module with the patient's historical imaging data, clinical symptoms, and laboratory test results to generate multi-dimensional diagnostic suggestions; A diagnosis support module, which, by combining the lesion detection results with the output of multi-modal information fusion, provides a preliminary diagnosis opinion and marks high-risk areas to assist doctors in making a final diagnosis.

[0021] The image acquisition module includes: An automatic exposure control sub-module: used to dynamically adjust exposure parameters during image acquisition, based on real-time analysis of the image brightness and contrast conditions, to ensure that the imaging quality of chest radiograph images remains consistent under different lighting conditions. This sub-module can automatically identify the chest density differences of different patients and perform personalized exposure settings, thereby effectively reducing overexposure or underexposure situations and enhancing the diagnostic value of the images; A denoising processing sub-module: applying the non-local means algorithm to effectively remove the noise generated by low-dose imaging and retain important image details through multi-scale denoising processing; Gray level equalization sub-module: It is used to perform gray level equalization processing on images. By adjusting the gray level distribution of the images, the overall contrast of the chest X-rays is made more uniform, ensuring that the lesion areas can be clearly displayed within different density ranges, facilitating subsequent image analysis and diagnosis.

[0022] The image enhancement module includes: Multi-scale pyramid enhancement sub-module: Based on the image pyramid structure, the chest X-ray image is decomposed into image levels of different scales, and each level of the image is enhanced respectively to highlight the lesion details existing in each level; Adaptive contrast adjustment sub-module: Applying adaptive contrast enhancement, according to the local region contrast characteristics of the image, the contrast value is dynamically adjusted, so that the low-contrast regions in the image are enhanced, making it easier to identify the hidden lesion areas; Tone mapping sub-module: It is used to adjust the tone distribution of the image, including enhancing the visual distinguishability of the target lesions by controlling the tone in different types of lesion areas, enabling doctors to more intuitively identify abnormal areas.

[0023] The lesion detection module further includes: Deep convolutional neural network sub-module: Using a pre-trained convolutional neural network model, multi-level feature extraction is performed on the chest X-ray image. Through the stacking of multiple convolutional layers and pooling layers, high-dimensional features related to the lesions are gradually refined. This sub-module is specifically aimed at complex lesion areas such as lung nodules and pleural effusions, and can improve the detection accuracy of the lesion areas through the adaptive adjustment of the convolutional kernels; Active learning sub-module: By introducing an active learning mechanism, under the guidance of doctors, difficult examples in the training dataset are automatically selected for learning, thereby improving the model's recognition ability for uncommon lesions; Region proposal network sub-module: It is used to generate candidate regions that may contain lesions. Based on the candidate regions, precise localization of the lesions is carried out. The region proposal network sub-module generates multi-scale candidate boxes to cover lesion areas of different sizes, and realizes the recognition of the lesions through the combination with deep features.

[0024] The multi-modal information fusion module includes: Temporal data analysis sub-module: It is used to analyze the patient's historical imaging data and time series features. By detecting the change trends between imaging data at different time points, it helps doctors identify the development process of the lesions; Laboratory test result fusion sub-module: It fuses and analyzes the patient's laboratory test results with the imaging data, and provides a more comprehensive diagnostic basis through the comprehensive processing of multi-dimensional data.

[0025] The described diagnostic support module includes a risk assessment sub-module. The risk assessment sub-module conducts a comprehensive risk assessment by comprehensively analyzing the lesion detection results, multi-modal information, and patient medical history data. Through multivariate logistic regression analysis, this sub-module predicts the risk of the patient's disease progression and generates a detailed risk assessment report to assist doctors in considering potential disease changes during the decision-making process.

[0026] The diagnostic support module also includes an interpretability analysis sub-module. The interpretability analysis sub-module enables doctors to understand the diagnostic basis of the system by visually presenting the decision-making process of the deep learning model.

[0027] This invention covers any substitutions, modifications, equivalent methods, and solutions made within the essence and scope of this invention. To enable the public to have a thorough understanding of this invention, specific details are elaborated in the following preferred embodiments of this invention. However, those skilled in the art can also fully understand this invention without the description of these details. Additionally, to avoid unnecessary confusion to the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0028] The above description is only a preferred embodiment of this invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of this invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of this invention.

Claims

1. An intelligent analysis and diagnosis support system for chest radiograph images, characterized in that, Including: An image acquisition module, which is used to receive and preprocess chest X-ray images to eliminate noise and artifacts caused by differences in imaging conditions; An image enhancement module, which enhances the preprocessed chest X-ray images through a multi-scale enhancement algorithm to improve the detail performance in the images and make the lesion areas clearer; A lesion detection module, which identifies and locates abnormal areas in chest X-rays based on multi-layer feature extraction of a deep learning model, including for pulmonary nodules, pleural effusion, and pulmonary shadow areas; A multi-modal information fusion module, which is used to fuse the results output by the lesion detection module with the patient's historical imaging data, clinical symptoms, and laboratory test results to generate multi-dimensional diagnostic suggestions; A diagnostic support module, which provides preliminary diagnostic opinions by combining the lesion detection results with the output of multi-modal information fusion and marks high-risk areas to assist doctors in making a final diagnosis.

2. The intelligent analysis and diagnosis support system for chest radiograph images according to claim 1, wherein The described image acquisition module includes: An automatic exposure control sub-module: which is used to dynamically adjust exposure parameters during image acquisition based on real-time analysis of the image brightness and contrast conditions to ensure that the imaging quality of chest X-ray images remains consistent under different lighting conditions; A denoising processing sub-module: which applies the non-local means algorithm to effectively remove the noise generated by low-dose imaging and retains important image details through multi-scale denoising processing; A gray-level equalization sub-module: which is used to perform gray-level equalization processing on the image and makes the overall contrast of the chest X-ray more uniform by adjusting the gray-level distribution of the image.

3. The intelligent analysis and diagnosis support system for chest radiograph images according to claim 1, characterized in that, The described image enhancement module includes: A multi-scale pyramid enhancement sub-module: based on the image pyramid structure, decomposes the chest X-ray image into image levels of different scales and enhances each level of the image separately to highlight the lesion details existing in each level; A contrast adaptive adjustment sub-module: which applies adaptive contrast enhancement, dynamically adjusts the contrast value according to the local area contrast characteristics of the image, so that the low-contrast areas in the image are enhanced, making it easier to identify hidden lesion areas; A tone mapping sub-module: which is used to adjust the tone distribution of the image, including enhancing the visual distinguishability of the target lesion by controlling the tone in different types of lesion areas.

4. The intelligent analysis and diagnosis support system for chest radiograph images according to claim 1, characterized in that, The described lesion detection module further includes: A deep convolutional neural network sub-module: which uses a pre-trained convolutional neural network model to perform multi-level feature extraction on chest X-ray images, and gradually extracts high-dimensional features related to lesions through the stacking of multiple convolutional layers and pooling layers; An active learning sub-module: which allows, under the guidance of a doctor, to automatically select difficult examples in the training dataset for learning by introducing an active learning mechanism, thereby improving the model's recognition ability for uncommon lesions; A region proposal network sub-module: which is used to generate candidate regions that may contain lesions, and based on the candidate regions, accurately locate the lesions. The region proposal network sub-module generates multi-scale candidate boxes to cover lesion areas of different sizes and realizes the recognition of lesions by combining with deep features.

5. The intelligent analysis and diagnosis support system for chest radiograph images according to claim 1, wherein The described multi-modal information fusion module includes: Time series data analysis sub-module: It is used to analyze the patient's historical imaging data and time series features. By detecting the change trend between imaging data at different time points, it helps doctors identify the development process of lesions. Laboratory test result fusion sub-module: It fuses and analyzes the patient's laboratory test results and imaging data. Through the comprehensive processing of multi-dimensional data, it provides a more comprehensive diagnostic basis.

6. The intelligent analysis and diagnosis support system for chest radiograph images according to claim 1, wherein The diagnosis support module includes a risk assessment sub-module. The risk assessment sub-module conducts a comprehensive risk assessment by comprehensively analyzing the lesion detection results, multi-modal information, and patient medical history data. This sub-module predicts the risk of the patient's disease development through multivariate logistic regression analysis.

7. The intelligent analysis and diagnosis support system for chest radiograph images according to claim 6, wherein The diagnosis support module also includes an interpretability analysis sub-module. The interpretability analysis sub-module visually displays the decision-making process of the deep learning model, enabling doctors to understand the diagnostic basis of the system.

Images