Remote skin disease image processing system and method

Abstract

The invention relates to the field of medical systems, and discloses a skin disease image remote processing system and method.The skin disease image remote processing system comprises an image collection module, the image collection module collects skin disease images of a patient, and the image collection module is connected with an image processing module; the image processing module receives image information acquired by the image acquisition module and performs clear and illumination processing on the image information, the image processing module is connected with an interactive interface module, and the interactive interface module is responsible for managing and displaying clear image contents. According to the invention, through a non-uniform illumination correction module in the image processing module, an algorithm for removing non-uniform illumination is used, so that the image brightness of a strong light area can be properly reduced, and the image brightness of a low-illumination area can be effectively improved, so that detailed information of the low-illumination area is well presented; meanwhile, the algorithm also has a good color retention characteristic.

Description

Technical Field

[0001] The present invention relates to the technical field of medical systems, and in particular to a remote processing system and method for skin disease images. Background Art

[0002] Skin cancer and various skin diseases pose a serious threat to human health and severely impact people's normal lives. Furthermore, with changes in living environments and lifestyles, the incidence of skin diseases has been increasing year by year. Furthermore, the pathogenesis of skin diseases varies greatly. Small papules, blisters, and pustules on the skin are caused by infection due to a weakened immune system; however, enlarged lesions or development of large erythematous patches or wheals are often caused by viral infections. Many skin diseases are long-lasting and often difficult to fully cure. Dermatoscopy is a commonly used diagnostic tool for various skin diseases. Currently, most medical professionals rely on dermatoscopy to diagnose skin tumors and other skin diseases. Numerous studies have shown that even experienced dermatologists have a clinical diagnostic accuracy rate of only around 75%, and the accuracy rate of general practitioners is even lower. Therefore, patients with skin diseases can only obtain a more accurate diagnosis by visiting a large hospital for a specialized dermatoscope performed by a dermatologist. This is a significant waste of medical resources for patients with mild, self-healing skin conditions whose clinical features strongly resemble those of malignant skin diseases.

[0003] Furthermore, the diagnosis and treatment of skin diseases primarily rely on visual information, a key advantage of image processing. While medical image processing technology is now widely used, it currently focuses solely on dermatoscope images. Research on processing skin disease images captured by smartphones is quite limited. Therefore, designing an image processing algorithm that can be applied to skin disease images captured by mobile phones has significant practical and operational implications for remote diagnosis of skin diseases, particularly pigmented skin diseases. This approach also holds promise for addressing the current challenges of accessing medical care in remote rural areas of my country. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a remote processing system and method for skin disease images, which solves the problem that although current medical image processing technology is widely used in dermatoscope images, there is a lack of research on skin disease images taken by smartphones. The development of such an algorithm can promote remote and accurate diagnosis of pigmented skin diseases and alleviate the shortage of medical resources in remote areas.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a remote processing system for skin disease images, including an image acquisition module, which acquires skin disease images of patients; an image acquisition module is connected to an image processing module, which receives image information acquired from the image acquisition module and performs clarity and lighting processing on the image information; an interactive interface module is connected to the image processing module, which is responsible for managing and displaying the cleared image content.

[0006] Preferably, the image processing module includes a defocus and blur detection module, which preliminarily screens and collects clearer skin disease images based on Fourier transform, removes pictures that are mixed with noise and have lost a lot of original image information, and transmits the processed pictures to the uneven illumination correction module.

[0007] Preferably, the defocus blur detection module is connected to an uneven illumination correction module, which receives the pictures after preliminary screening and implements illumination correction of the skin image based on an uneven illumination adaptive algorithm of a two-dimensional gamma function, thereby presenting the true information of the skin disease image to dermatologists to the greatest extent possible.

[0008] Preferably, the illumination unevenness adaptive algorithm based on the two-dimensional gamma function includes: a first step: extracting illumination components using a multi-scale Gaussian function.

[0009] Preferably, the illumination unevenness adaptive algorithm based on the two-dimensional gamma function further includes: a second step: adaptive brightness correction based on the two-dimensional gamma function.

[0010] Preferably, the interactive interface module includes a file management module and a help module. The interactive interface module includes a visualization window, which facilitates users to delete or modify controls or modify codes through the visualization window. The file management module includes functions of opening pictures, taking screenshots, saving and exiting the system. The file management module is connected to the image processing module.

[0011] A method for using a remote skin disease image processing system comprises the following steps:

[0012] S1. Image acquisition: The user takes a photo of the patient's skin disease using a camera;

[0013] S2. Open the system: run the system and enter the interactive interface;

[0014] S3. Interactive interface module operation: choose to operate directly or view the help guide according to the user's proficiency;

[0015] S4. Open the image and perform preliminary screening: The system uses the image processing module to initially screen the image and remove blurry images;

[0016] S5. Determine whether the image is clear: Based on the image clarity, determine whether to enter the lighting correction;

[0017] S6. Eliminate uneven illumination: Extract illumination components using multi-scale Gaussian filtering and dynamically adjust the gamma value to achieve brightness balance.

[0018] S7. Output results: Finally, the processed image will be displayed to the staff.

[0019] Preferably, the step S3 includes:

[0020] S301. Understand that system operators use files directly.

[0021] Preferably, the step S3 includes:

[0022] S302. Operators who are not familiar with the system need to check the help module.

[0023] Preferably, the step S5 includes:

[0024] S501. If the image is clear, select to display the spectrum:

[0025] S502. If the image is judged to be blurry, the preliminary screening step is repeated for the image.

[0026] The present invention provides a system and method for remote processing of skin disease images. It has the following beneficial effects:

[0027] 1. The present invention sets an image processing module to perform preliminary screening on the image through the defocus blur detection module, taking into account the screening of blurred images and avoiding the impact of blurred images on the doctor's work efficiency.

[0028] 2. The present invention uses an uneven illumination correction module in the image processing module and an algorithm for removing uneven illumination to appropriately reduce the image brightness in the bright light area and effectively improve the image brightness in the low illumination area, so that the detailed information in the low illumination area is well presented; at the same time, the algorithm also has excellent color preservation characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is the overall framework diagram of the skin image processing system of the present invention;

[0030] Figure 2 This is an operation flow chart of the skin image processing system of the present invention;

[0031] Figure 3A schematic diagram of a physical model of digital image formation according to the present invention;

[0032] Figure 4 It is the algorithm flow chart of the present invention;

[0033] Figure 5 This is a diagram showing the main interface of the skin image processing system of the present invention;

[0034] Figure 6 This is a diagram showing the functions of the file submodule of the present invention;

[0035] Figure 7 This is a functional diagram of the image processing algorithm module of the present invention;

[0036] Figure 8 This is a schematic diagram showing the effect of the image blur preprocessing algorithm of the present invention;

[0037] Figure 9 Schematic diagram showing the effect of the adaptive illumination unevenness adaptive algorithm based on the two-dimensional gamma function of the present invention. DETAILED DESCRIPTION

[0038] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0039] Example:

[0040] Please see the attached Figure 1 - Attachment Figure 9 An embodiment of the present invention provides a remote processing system for skin disease images, including an image acquisition module, which acquires skin disease images of patients and transmits the acquired skin disease images to an image processing module. The image acquisition module is connected to an image processing module, which receives image information acquired from the image acquisition module, performs clarity and lighting processing on the image, and transmits the processed image to an interactive interface module. The image processing module is connected to an interactive interface module, which is responsible for managing the cleared image content and displaying it for convenient viewing by medical staff.

[0041] The image processing module includes a defocus and blur detection module, which preliminarily screens and collects relatively clear skin disease images based on Fourier transform, removes images mixed with noise and that have lost a lot of original image information, and transmits the processed images to an uneven illumination correction module. The defocus and blur detection module is connected to an uneven illumination correction module, which receives the preliminarily screened images and implements illumination correction of the skin images based on an uneven illumination adaptive algorithm based on a two-dimensional gamma function, thereby presenting the true information of the skin disease images to dermatologists to the greatest extent possible;

[0042] The illumination unevenness adaptive algorithm based on the two-dimensional gamma function includes:

[0043] Step 1: Use multi-scale Gaussian function to extract illumination components.

[0044] According to the principle of image formation, the image formed in the visible light range is formed by the light emitted from the surface of the object in the scene reaching the imaging unit. Usually, the digital image can be regarded as a two-dimensional function f(x, y). The value of the function is the brightness value of the image at the coordinate point (x, y). f(x, y) is composed of the product of the illumination component i(x, y) incident on the scene and the reflection component r(x, y) on the object surface. The expression is as follows:

[0045] f(x,y)=i(x,y)r(x,y)

[0046] This model is called the illumination-reflection imaging model, and its physical model is as follows: Figure 3 shown.

[0047] To correct images with uneven lighting, it is necessary to accurately extract the lighting components in the scene. In order to better use the adaptive brightness correction algorithm, this paper makes the following assumptions based on the Retinex theory: the lighting components in the scene are mainly concentrated in the low-frequency area of ​​the image, and the overall changes are relatively stable; the reflective components are mainly distributed in high-frequency areas such as edges and textures, and their changes are large. Therefore, the extracted lighting components only contain information about light changes and do not involve detailed information about the image. Since the multi-scale Gaussian function method can effectively compress the dynamic range and accurately estimate the lighting components in the scene. Therefore, this paper chooses the multi-scale Gaussian function method to extract the lighting components of images with uneven lighting. The Gaussian function is as follows:

[0048]

[0049] In the formula, c is the scale factor; λ is the normalization constant, which ensures that the Gaussian function G(x,y) satisfies the normalization condition;

[0050] By convolving the Gaussian function with the original image, we can get an estimate of the illumination component. The result is as follows:

[0051] I(x,y)=F(x,y)G(x,y)

[0052] Where F(x,y) is the input image; I(x,y) is the estimated illumination component of the image.

[0053] This paper adopts a method based on multi-scale Gaussian function to extract the illumination components in the scene and weight the illumination components. Finally, an estimated value of the illumination component is obtained, and the estimation formula is as follows:

[0054]

[0055] I(x,y) in the formula is the illumination component value extracted and weighted by multiple Gaussian functions of different scales at the point (x,y); w i is the weight coefficient of the illumination component extracted by the i-th scale Gaussian function; i = 1, 2, 3…N

[0056] is the number of scales used;

[0057] Step 2: Adaptive brightness correction based on two-dimensional gamma function

[0058] After extracting the illumination components of the natural scene in the image, we can construct an uneven illumination correction function based on the distribution characteristics of the illumination components. This function can be used to correct uneven illumination in skin disease images, reducing the brightness of areas with excessive illumination and increasing the brightness of areas with insufficient illumination. To this end, we use an adaptive brightness correction algorithm based on a two-dimensional gamma function. This algorithm fully considers the distribution characteristics of the illumination components and adaptively adjusts the parameters of the gamma function to adjust the brightness of each area, thereby improving the overall image quality. The function expression is as follows:

[0059]

[0060] Where O(x,y) is the brightness value of the output image after correction; λ is the brightness enhancement index value, including the characteristics of the image's illumination components; m is the average brightness of the illumination components;

[0061] The interactive interface module includes a file management module and a help module. The interactive interface module includes a visualization window, which facilitates users to delete or modify controls or modify codes through the visualization window. The file management module includes functions such as opening pictures, taking screenshots, saving and exiting the system. The file management module is connected to the image processing module.

[0062] A method for using a remote skin disease image processing system comprises the following steps:

[0063] S1. Image acquisition: The user takes a photo of the patient's skin disease using a camera;

[0064] S2. Open the system: run the system and enter the interactive interface;

[0065] S3. Interactive interface module operation: choose to operate directly or view the help guide according to the user's proficiency;

[0066] S301. Understand that system operators directly use files;

[0067] S302. Operators who do not understand the system need to check the help module;

[0068] S4. Open the image and perform preliminary screening: The system uses the image processing module to initially screen the image and remove blurry images;

[0069] S5. Determine whether the image is clear: Based on the image clarity, determine whether to enter the lighting correction;

[0070] S501. If the image is clear, select to display the spectrum:

[0071] S502. If the image is blurred, repeat the initial screening steps for the image:

[0072] S6. Eliminate uneven illumination: Extract illumination components using multi-scale Gaussian filtering and dynamically adjust the gamma value to achieve brightness balance.

[0073] S7. Output results: Finally, the processed image will be displayed to the staff.

[0074] Comparative experiment:

[0075] The following experiments compared the performance of the method of the present invention with that of unoptimized smartphone processing technology and traditional dermatoscope image processing to verify the system's advantages in image clarity screening, illumination correction, and diagnostic accuracy.

[0076] 1. Experimental Setup

[0077] Dataset

[0078] Group A (Technology group of the present invention): collected 1,000 skin disease images (mainly pigmented skin diseases) taken by smartphones, with a resolution of 1080p or above, covering different lighting conditions (low light, backlight, and shadow).

[0079] Group B (traditional dermatoscope group): 500 images of similar skin diseases were obtained using a professional dermatoscope.

[0080] Group C (non-optimized mobile phone group): 1,000 original mobile phone images without filtering and correction.

[0081] Parameter configuration

[0082] Blur detection: Fourier transform high-frequency energy threshold is 15% (if above the threshold, it is considered clear).

[0083] Multi-scale Gaussian filtering: the scale factor σ is [3, 5, 7], and the weight coefficient α is [0.4, 0.3, 0.3].

[0084] Adaptive gamma correction: initial γ = 1.5, dynamic adjustment range 0.8 to 2.2, average brightness L_avg = 128.

[0085] Evaluation Metrics

[0086] Image quality: PSNR (peak signal-to-noise ratio), SSIM (structural similarity).

[0087] Diagnostic consistency rate: Compare the consistency of doctors' diagnostic results before and after treatment (blind review by 3 senior dermatologists).

[0088] Algorithm efficiency: single image processing time (CPU: Intel i7-10750H; memory: 16GB; MATLAB R2021a).

[0089] 2. Experimental steps

[0090] Step 1: Image clarity screening test

[0091] Input: 1000 original images of Group A and Group C, including 30% blurred images.

[0092] Processing flow:

[0093] Group A used Fourier transform high frequency energy detection.

[0094] Group C directly accepted all images without screening.

[0095] Output: 700 clear images in group A and 300 blurred images in group C after screening.

[0096] Step 2: Lighting Correction Comparison Test

[0097] Input: Group A: 700 screened images; Group B: 500 dermoscopic images.

[0098] Processing flow:

[0099] Group A applied multi-scale Gaussian filtering + adaptive gamma correction.

[0100] Group B used standard dermoscopic image processing (histogram equalization).

[0101] No correction was performed for group C.

[0102] Output: Generate the corrected images for each group.

[0103] Step 3: Diagnostic agreement assessment

[0104] Input: 200 randomly selected corrected images from group A, 200 dermoscopic images from group B, and 200 unprocessed images from group C.

[0105] Evaluation Methodology:

[0106] Three doctors independently diagnosed the three sets of images and marked whether they were malignant lesions.

[0107] Using dermatoscopic biopsy results as the gold standard, the diagnostic sensitivity (true positive rate) and specificity (true negative rate) were calculated.

[0108] Step 4: Efficiency Assessment

[0109] The average processing time of each set of images (single image) is recorded, and the real-time performance improvement of the present invention is compared.

[0110] 3. Experimental Results

[0111]

[0112]

[0113] 4. Results Analysis

[0114] Image quality optimization:

[0115] The PSNR and SSIM of the present invention were significantly higher than those of group C (p<0.01*), and the clarity of the lesion edge after correction was close to the dermatoscope level.

[0116] The dark details of low-light images (such as melanin distribution) are improved by more than 50% (SSIM>0.9).

[0117] Improved diagnostic accuracy:

[0118] The doctors' diagnostic sensitivity for group A images (93.2%) was 23.8% higher than that for group C (75.8%), and better than the 89.5% for the dermatoscope group.

[0119] The specificity verified the system's ability to suppress artifact interference (for example, the error in distinguishing erythema from normal skin was reduced by 15% after correction).

[0120] Processing efficiency:

[0121] The algorithm of the present invention takes 1.8 seconds per image, which is 28.6% faster than the traditional method and meets the needs of real-time remote diagnosis.

[0122] Applicability verification:

[0123] The algorithm still maintains PSNR ≥ 34dB in the test on low-end mobile phones (720p resolution), confirming its potential for lightweight deployment.

[0124] 5. Experimental Conclusion

[0125] By combining Fourier transform screening and adaptive gamma correction, this invention significantly improves the quality of skin disease images taken by smartphones, narrows the gap with professional dermatoscopes, and greatly improves the efficiency and accuracy of remote diagnosis, providing reliable technical support for primary medical care.

[0126] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A remote processing system for skin disease images, characterized in that: It includes an image acquisition module, which acquires skin disease images of patients. The image acquisition module is connected to an image processing module, which receives image information acquired from the image acquisition module and performs clarity and lighting processing on the image information. The image processing module is connected to an interactive interface module, which is responsible for managing and displaying the cleared image content.

2. A remote skin disease image processing system according to claim 1, characterized in that: The image processing module includes a defocus and blur detection module, which preliminarily screens and collects clearer skin disease images based on Fourier transform, removes pictures that are mixed with noise and have lost a lot of original image information, and transmits the processed pictures to the uneven illumination correction module.

3. A remote skin disease image processing system according to claim 2, characterized in that: The defocus blur detection module is connected to an uneven illumination correction module, which receives the preliminarily screened images and implements illumination correction of the skin image based on an uneven illumination adaptive algorithm based on a two-dimensional gamma function, thereby presenting the true information of the skin disease image to dermatologists to the greatest extent possible.

4. The remote skin disease image processing system according to claim 3, characterized in that: The illumination unevenness adaptive algorithm based on the two-dimensional gamma function includes: a first step: extracting illumination components using a multi-scale Gaussian function.

5. The remote skin disease image processing system according to claim 3, characterized in that: The illumination unevenness adaptive algorithm based on the two-dimensional gamma function further includes: a second step: adaptive brightness correction based on the two-dimensional gamma function.

6. The remote skin disease image processing system according to claim 1, characterized in that: The interactive interface module includes a file management module and a help module. The interactive interface module includes a visualization window, which facilitates users to delete or modify controls or modify codes through the visualization window. The file management module includes functions such as opening pictures, taking screenshots, saving and exiting the system. The file management module is connected to the image processing module.

7. The method for using the remote skin disease image processing system according to any one of claims 1-2, characterized in that: The following steps are involved: S1. Image acquisition: The user takes a photo of the patient's skin disease using a camera; S2. Open the system: run the system and enter the interactive interface; S3. Interactive interface module operation: choose to operate directly or view the help guide according to the user's proficiency; S4. Open the image and perform preliminary screening: The system uses the image processing module to initially screen the image and remove blurry images; S5. Determine whether the image is clear: Based on the image clarity, determine whether to enter the lighting correction; S6. Eliminate uneven illumination: Extract illumination components using multi-scale Gaussian filtering and dynamically adjust the gamma value to achieve brightness balance. S7. Output results: Finally, the processed image will be displayed to the staff.

8. The method for using the remote skin disease image processing system according to claim 7, characterized in that: The step S3 includes: S301. Understand that system operators use files directly.

9. The method for using the remote skin disease image processing system according to claim 7, characterized in that: The step S3 includes: S302. Operators who are not familiar with the system need to check the help module.

10. The method for using the remote skin disease image processing system according to claim 7, characterized in that: The step S5 includes: S501. If the image is clear, select to display the spectrum: S502. If the image is judged to be blurry, the preliminary screening step is repeated for the image.

Images