Lung nodule visual enhancement identification method
By constructing a threshold selection range and calculating the membership probability of pixels, combined with gamma transform and pseudo-color processing, the problem of nodules and tissue connectivity interference in lung CT images was solved, and the visual recognition accuracy of nodules was improved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-09
- Publication Date
- 2026-05-08
AI Technical Summary
In existing technologies, adaptive threshold segmentation methods for lung CT images suffer from connectivity issues, leading to nodules and tissues being connected, which interferes with the determination of nodule shape and area and affects diagnostic accuracy.
By constructing a threshold selection range, using grayscale histograms and morphological processing, the first and second membership probabilities of pixels are calculated. Combined with gamma transform and pseudo-color processing, the visual recognition effect of nodules is enhanced.
It improves the visual recognition accuracy of lung nodules, reduces noise interference, enhances the salience of nodules, and makes nodule areas easier to identify.
Smart Images
Figure CN121998880A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of image data processing technology, and more specifically to a method for visually enhanced recognition of lung nodules. Background Technology
[0002] A chest CT scan is an imaging technique that uses computed tomography (CT) scans to assess lung diseases. This technique can be used to diagnose lung diseases such as lung nodules, lung tumors, and lung inflammation.
[0003] Medical departments use lung CT images to determine if there are nodules in the lungs. First, the grayscale images of the lung CT images need to be preprocessed to extract the lung regions, which are then segmented. The probability of nodules is determined by their shape and area. Existing techniques use adaptive thresholding segmentation methods; however, due to connected component adhesion issues, nodules and their corresponding tissues are often connected and integrated, significantly interfering with subsequent determinations of nodule shape and area.
[0004] This method first segments the image to obtain the lung region. Then, it uses the gray-level histogram of the lung region to determine the approximate threshold range. Based on nodule characteristics, it analyzes the probability of nodules at each threshold to obtain the probability of nodules in the CT image. Finally, it enhances the obtained nodules with pseudo-color. Summary of the Invention
[0005] This invention provides a visual enhancement method for recognizing pulmonary nodules to solve existing problems.
[0006] The present invention provides a method for visually enhanced recognition of pulmonary nodules, which adopts the following technical solution:
[0007] One embodiment of the present invention provides a method for visually enhanced recognition of pulmonary nodules, the method comprising the following steps:
[0008] A CT scanner is used to acquire grayscale CT images of the lungs, and the threshold selection range in the CT image is obtained based on the grayscale range of the lung CT grayscale image.
[0009] Based on the threshold selection range, the threshold is traversed one by one, and each threshold is taken as the image segmentation threshold to segment the grayscale image of lung CT to obtain multiple image segmentation result images, and the perimeter and area of the connected region in the result image are obtained for the same pixel.
[0010] The first membership probability of a pixel in each result image is obtained based on the perimeter and area of the connected region where the pixel is located. The first membership probability sequence is constructed by the first membership probabilities of the same pixel in multiple result images. The threshold used in image thresholding is used as the horizontal axis and the first membership probability is used as the vertical axis. The first membership probability sequence is constructed into a curve of the change of the first membership probability with respect to the threshold, which is called the first curve.
[0011] The second curve is intercepted on the first curve, and the maximum and minimum points of the first membership probability are obtained based on the second curve. The first curve feature of the second curve is obtained based on the difference between the maximum and minimum points, and the second curve feature of the second curve is obtained based on the slope of the maximum and minimum points.
[0012] The second membership probability of a pixel is obtained based on the first curve feature and the second curve feature.
[0013] The lung CT grayscale image is subjected to gamma transformation based on the second membership probability of the pixel to obtain the gamma transformation result image, and then pseudo-color processing is performed on the gamma transformation result image to obtain the pseudo-color processing result image.
[0014] Furthermore, the specific steps for obtaining the threshold selection range in the CT image based on the grayscale range of the lung CT grayscale image are as follows:
[0015] A grayscale CT image of the lungs is acquired using a CT scanner, and the number of pixels with the same grayscale value in the image is counted. A grayscale distribution histogram is constructed with the grayscale value on the horizontal axis and the corresponding number of pixels on the vertical axis. The maximum and minimum grayscale values of the corresponding lung CT grayscale image are obtained from the grayscale distribution histogram. The grayscale range of the lung CT grayscale image is defined as the range between the maximum and minimum grayscale values. The grayscale values within the grayscale range of the lung CT grayscale image are then used as the threshold selection range.
[0016] Furthermore, the resulting image is preprocessed using morphological opening operations.
[0017] Furthermore, the first membership probability sequence is obtained as follows:
[0018] First membership probability of a pixel in a grayscale image of a lung CT scan:
[0019]
[0020] In the formula, A in C refers to the area of the connected region containing the i-th pixel in the n-th result image. ih This refers to the perimeter of the connected component containing the i-th pixel in the n-th morphologically processed image, m. inThis represents the first membership probability of the i-th pixel in the n-th result image;
[0021] The first membership probability of a pixel in all the resulting images constitutes the first membership probability sequence.
[0022] Furthermore, the first curve feature is obtained using the following method:
[0023]
[0024] In the formula, f i The fluctuation degree of the second curve representing the i-th pixel in the lung CT grayscale image is denoted as the first curve feature. H represents the number of first membership probabilities greater than the empirical probability threshold in the first membership probability sequence of the i-th pixel in the lung CT grayscale image. M ih The local mean factor represents the h-th first membership probability greater than the empirical probability threshold in the first membership probability sequence of the i-th pixel in a lung CT grayscale image. This represents the mean of the first membership probability in the second curve.
[0025] Furthermore, the second curve feature is obtained using the following method:
[0026] Find the pair of maximum and minimum points with the largest difference on the horizontal axis in the second curve;
[0027] The difference between the maximum and minimum values of the first membership probability in the second curve is denoted as the first difference.
[0028] The difference between the maximum value of the first membership probability in the second curve and the segmentation threshold corresponding to the minimum target value is recorded as the second difference.
[0029] The absolute value of the ratio of the first difference to the second difference is used as the feature of the second curve.
[0030] Furthermore, the specific steps for obtaining the second membership probability of a pixel based on the first curve feature and the second curve feature are as follows:
[0031] P i =exp(-βf i )*exp(-k i )
[0032] In the formula, exp() is an exponential function with the natural constant as the base; β is an adjustment parameter, and P i Let represent the second membership probability of the i-th pixel in the grayscale image of a lung CT scan, and let represent the fluctuation degree of the second curve of the i-th pixel in the grayscale image of a lung CT scan. Let k represent the second curve feature. i The slope of the second curve represents the i-th pixel in a grayscale CT image of the lungs.
[0033] Furthermore, the specific steps for obtaining the gamma-transformed image by performing gamma transformation on the lung CT grayscale image based on the second membership probability of the pixels include the following:
[0034]
[0035] Where, q i P represents the grayscale value of the i-th pixel in the image; i f represents the second membership probability of the i-th pixel; i This represents the grayscale value of the i-th pixel after gamma transformation. All pixels after gamma transformation constitute the resulting gamma-transformed image. Attached Figure Description
[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0037] Figure 1 This is a flowchart illustrating the steps of a visual enhancement method for recognizing pulmonary nodules according to the present invention.
[0038] Figure 2 Grayscale image of lung nodules;
[0039] Figure 3 The graph shows the first membership probability curves under different thresholds;
[0040] Figure 4 This is a cut-off plot of the first membership probability curve under different thresholds. Detailed Implementation
[0041] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a data management method for a security operation and maintenance system and a visual enhancement recognition method for lung nodules proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0043] The following description, in conjunction with the accompanying drawings, details a specific scheme for a visual enhancement recognition method for lung nodules provided by the present invention.
[0044] Please see Figure 1 The diagram illustrates a flowchart of a visual enhancement method for recognizing lung nodules according to an embodiment of the present invention. The method includes the following steps:
[0045] Step S001: Obtain a grayscale image of the lung CT scan. After grayscale processing, determine the approximate range of the threshold based on the histogram of the lung CT grayscale image.
[0046] This embodiment uses a threshold segmentation algorithm to segment the lung region to different degrees. Since the gray value selection range is between 0 and 255, the corresponding threshold selection range will be too large. In order to reduce the threshold selection range, based on the gray value of the lung (the gray value of the nodule part is relatively small, the gray value of the lung tissue part is relatively large, and the nodule part is a circle with a diameter of less than 30 mm), the gray value distribution histogram of the lung CT gray value image and the corresponding result image after some threshold segmentation are observed to obtain a suitable threshold selection range.
[0047] The specific steps include:
[0048] Using a CT scanner to acquire grayscale CT images of the lungs, such as Figure 2 As shown;
[0049] The number of pixels with the same gray value in the image is counted, and a gray-level distribution histogram is constructed with the gray value on the horizontal axis and the corresponding number of pixels on the vertical axis. The gray-level distribution histogram represents the probability of each gray value appearing in the lung CT gray-level image. Based on the gray-level distribution histogram of the lung CT gray-level image, the maximum and minimum gray values with a non-zero probability of appearing in the gray-level distribution histogram are obtained. All gray values within the range from the minimum gray value to the maximum gray value are taken as the gray-level distribution range of the lung CT gray-level image, including the maximum and minimum gray values. Then, the gray-level distribution range is used as the threshold selection range for threshold segmentation of the lung CT gray-level image.
[0050] Thus, we have obtained Figure 2 The grayscale distribution range of lung CT grayscale images is 0-200, which means the threshold selection range is also 0-200.
[0051] Step S002: Based on the segmentation effect of the lung region under each threshold, and combined with nodule features (shape, size), obtain the first membership probability of the pixel in the connected domain.
[0052] The specific steps are as follows:
[0053] Because any selected threshold exceeding the optimal threshold range results in insufficiently distinct lung nodules in the segmented image, the probability of identifying a lung nodule within that threshold image is lower. Conversely, within a better threshold range, lung nodules are more clearly visible, increasing the probability of them being identified as lung nodules. For CT images without nodules, the probability of each corresponding nodule is relatively low across multiple thresholds; for images with nodules, the probability is higher at certain thresholds.
[0054] Furthermore, the clarity of the corresponding lung nodule regions varies in the threshold segmentation results obtained for different thresholds K. For CT images with good segmentation results, the nodule clarity is better, while for CT images with poor segmentation results, the nodule clarity is generally poor or even invisible. Therefore, the probability of nodules corresponding to different thresholds is also different.
[0055] Based on the analysis of the threshold segmentation result image corresponding to each threshold K, and combined with nodule features (shape, size), the probability that each connected component is a nodule is obtained. The specific process of obtaining the probability that each connected component is a nodule within a certain threshold selection range in this embodiment is as follows:
[0056] (1) Obtained in the above steps Figure 2 The grayscale distribution range is 0-200, therefore Figure 2 The threshold selection range is also 0-200. Within the threshold selection range, the values are traversed one by one from small to large to obtain the threshold K. The lung grayscale image is segmented using different threshold K. The pixel values of the lung CT grayscale image with gray values less than the threshold K are set to 0, and the pixel values of the pixel values with gray values greater than the threshold K are set to 1. Assuming that a total of N different thresholds are obtained by traversing within the threshold selection range, the segmentation result images of N lung nodules corresponding to different threshold K can be obtained, that is, the threshold segmentation result images.
[0057] Since different thresholds K result in different segmentation effects, the degree to which nodes are clearly presented in the connected components also varies.
[0058] Based on the characteristics of the nodules, it can be understood that the area of suspected nodules should be round or near-round in shape. In reality, nodules are solid tissues that are approximately spherical. Therefore, in the acquired grayscale images of lung CT, the display effect of nodules is generally consistent. However, the display effect of nodules is best in the segmentation effect image with a threshold value of around 90.
[0059] (2) Morphological opening operations are performed on the threshold segmentation result images obtained at different thresholds K to obtain N corresponding morphological processing result images. The influence of lung tissue, blood vessels, etc., is eliminated and noise in the images is removed. Connected component detection is then performed on the N morphological processing result images to obtain the first sequence {C} of the connected components where the pixels at the same position are located in the morphological processing result images corresponding to the segmentation result images at different thresholds. i1 C i2 C in C iN} and the second sequence {A i0， A i1 A in A iN}, where C in A represents the perimeter of the connected component containing the i-th pixel of the n-th resulting image in the morphological processing result image. in This represents the area of the connected region containing the i-th pixel in the n-th morphologically processed image. It is then compared with the characteristics of lung nodules (lung nodules are round or nearly round, with a radius between 30 mm) to obtain the probability that each pixel in the N morphologically processed images contains a suspected nodule.
[0060] Because the nodules are circular or near-circular in shape with a radius of 30 mm or less, and the noise points have varying shapes, the approximation of the nodules can be obtained from N morphologically processed images after segmentation and morphological processing at different thresholds K. This approximation is based on the shape characteristics of the connected components, and the first membership probability m of the connected component containing the i-th pixel in the n-th morphologically processed image is obtained based on the shape characteristics of the nodules. in .
[0061] The circularity of pixels in a connected region of an image is determined by calculating the circularity, which is also known as the first membership probability m. in The closer the circularity of a connected component is to 1, the closer its shape is to a circle. Therefore, the first membership probability m... in :
[0062]
[0063] In the formula A in C refers to the area of the connected region containing the i-th pixel in the n-th morphologically processed image. in This refers to the perimeter of the connected component containing the i-th pixel in the n-th morphologically processed image, according to 4*π*A. in The degree of fit to the circle is determined by the ratio of the circle's perimeter to the square of the perimeter of the white connected region. The first membership probability m... inThis represents the circularity of the connected region containing the i-th pixel in the n-th morphological processing result image. The closer the circularity of the connected region is to 1, the more likely it is to be a lung nodule region.
[0064] Connected component analysis is used to obtain the probability that the connected component containing each pixel at the current threshold belongs to a lung nodule. Similarly, the same approach is taken for other morphological processing results, obtaining the probability that the connected component containing each pixel at each threshold belongs to a nodule. That is, since a connected component contains multiple pixels, the first membership probability of the connected component is assigned to each pixel within the connected component, thus obtaining the first membership probability of each pixel. This yields the probability that each pixel in the morphologically processed image at each threshold K belongs to a nodule.
[0065] The first membership probability sequence {m} of the same pixel in the morphologically processed images after different threshold segmentation and morphological processing is obtained through the above calculations and statistics. i1， m i2， …, m in， …, m iN}, where m in Let represent the first membership probability of the i-th pixel in the n-th morphological processing result image. In addition, let the threshold used in the n-th threshold segmentation be the horizontal axis and the first membership probability of the i-th pixel be the vertical axis. Establish the curve of the first membership probability of the i-th pixel with respect to the threshold, which is called the first curve of the i-th pixel.
[0066] Step S003: Based on the segmentation results under different thresholds, determine the second membership probability of the lung nodules in the CT image.
[0067] Each threshold K segmentation of a lung CT grayscale image yields a segmented image. Through the above calculations, the probability of a pixel belonging to a nodule in the segmented image at each threshold K, and the resulting image after morphological processing, can be obtained. This nodule probability is for a single threshold. However, since a single threshold cannot effectively segment nodules, the probability of a pixel belonging to a nodule is determined by combining the segmentation probability at each threshold K and the resulting image after morphological processing. This allows for the calculation of the true nodule affiliation of each pixel. Because nodule regions are circular and relatively dense, the connected components of nodule regions after segmentation at different thresholds are more circular, with minimal variation in circularity. In contrast, the connected components formed by segmenting the trachea and other skeletal regions at certain thresholds are circular, but as the threshold changes, the segmented connected components exhibit irregular shapes, such as tree branches. Therefore, a probability change curve for pixels under multiple thresholds is constructed to calculate the true nodule probability of each pixel. This allows for visual enhancement of pixels with high true nodule probabilities. The process is as follows:
[0068] like Figure 3 As shown in the figure, the graph displays the first probability change curves for two pixels belonging to a nodule. These curves represent the first probability of two pixels. When the probability of a pixel belonging to a nodule is low at a certain threshold, it has little reference value. Therefore, a threshold is set to truncate the first curve. The empirical probability threshold is denoted as m′, and the empirical value is m′ = 0.65. The probability change curves are then truncated using this empirical probability threshold. Figure 4 The method for obtaining the extracted curve segment is shown below. Taking the i-th pixel as an example, the extracted curve segment is represented as {M}. i1， M i2， …, M ih， …, M iH The sequence of curve segments is called the second curve, where M... ih This represents the h-th element in the curve portion of the i-th pixel in a lung CT grayscale image where the first membership probability is greater than the empirical threshold.
[0069] Analysis reveals that, due to the circular shape and dense texture of the nodule region, the segmented connected regions of pixels belonging to the nodule region change slowly. In contrast, the tracheal region and other skeletal regions change with the segmentation threshold, resulting in irregularly shaped connected regions, such as a tree-branch pattern. This indicates a greater variation in the crack probability under different thresholds. Therefore, the degree of curve fluctuation for the portion exceeding the empirical threshold can be calculated.
[0070]
[0071] In the formula, the first curve characteristic f i Let M represent the fluctuation degree of the second curve of the i-th pixel in the grayscale image of a lung CT scan, H represent the number of first membership probabilities greater than the empirical probability threshold in the first membership probability sequence of the i-th pixel in the grayscale image of a lung CT scan, and M represent the number of first membership probabilities greater than the empirical probability threshold. ih The local mean factor represents the h-th first membership probability greater than the empirical probability threshold in the first membership probability sequence of the i-th pixel in a lung CT grayscale image. This represents the mean of the first membership probability in the second curve. The greater the fluctuation of the curve, the greater the change in the probability of nodules under different thresholds, that is, the lower the probability of nodules belonging to them.
[0072] Because the nodule region is circular and relatively dense, the segmented connected components of pixels belonging to the nodule region change slowly in multiple segmentation result images. However, the connected components formed by the segmentation of the tracheal region and other skeletal regions will exhibit irregular shapes, such as tree branches, as the segmentation threshold changes. The first membership probability changes significantly under different thresholds, and the obtained connected components are easier to determine whether they correspond to the lung nodule region. Therefore, in this embodiment, we focus more on the points corresponding to the maximum segmentation threshold of the two minimum points. Thus, we obtain the maximum and minimum points of the curves that are greater than the empirical threshold. If there are multiple maximum and minimum points, we only retain the pair of maximum and minimum points with the largest absolute value of the difference in the corresponding horizontal coordinates.
[0073] The slope is calculated based on the selected maximum and minimum points. Since the nodule region changes slowly, the curve changes slowly (i.e., the slope approaches 0), while other regions change more rapidly, resulting in faster curve changes (i.e., the absolute value of the slope is larger). This is the second curve feature k of the i-th pixel. i :
[0074]
[0075] In the formula, the characteristic k of the second curve i This represents the overall slope of the second curve, with a maximum value of M. imax and minimum value M imin M represents the first membership probability corresponding to the maximum value and the first membership probability corresponding to the minimum value of the second curve for the i-th pixel in the lung CT grayscale image, respectively, and the first difference M. imax -M min This indicates the fluctuation range of the second curve; d imax and d imin d represents the segmentation thresholds corresponding to the maximum and minimum membership probabilities of the i-th pixel in the lung CT grayscale image, respectively. imax -d imin This is denoted as the second difference; the slope of the curve is the characteristic value of the second curve, k. i The closer a value is to 0, the greater the probability that the pixel belongs to a nodule.
[0076] The second membership probability of the corresponding pixel is obtained based on the first curve feature and the second curve feature, that is:
[0077] P i =exp(-βf i )*exp(-k i )
[0078] In the formula, exp() is an exponential function, used to normalize the values within the parentheses, mapping them to the interval (0, 1); β is an adjustment parameter, taken as an empirical value of β = 0.2, which can be manually adjusted; P i The second membership probability of the i-th pixel in a lung CT grayscale image is represented by the first curve feature f. i The second curve represents the degree of fluctuation of the i-th pixel in a lung CT grayscale image, and the feature k of the second curve is... i P represents the slope of the second curve representing the i-th pixel in a grayscale CT image of the lungs. i The larger the value, the smaller the fluctuation and the closer the slope of the second curve is to 0, thus the greater the probability that the pixel belongs to a nodule. Therefore, the second membership probability P is calculated for all pixels. i Through the second membership probability P i Apply pseudo-color enhancement to the image.
[0079] Step S004: Perform pseudo-color processing on the image to obtain a visually enhanced image.
[0080] The specific steps include:
[0081] After multi-threshold analysis, the second membership probability of pixels in the lung CT grayscale image is linearly normalized. Then, the grayscale of each pixel is transformed using the following gamma transform method to obtain the gamma transform result image:
[0082]
[0083] Where, q i P represents the grayscale value of the i-th pixel in the image; i f represents the second membership probability of the i-th pixel; i This represents the grayscale value of the i-th pixel after gamma transformation.
[0084] Second membership probability P i A higher grayscale value indicates that the pixel is more likely to belong to a nodule region, and the grayscale value of that pixel should be increased; conversely, it should be decreased. This achieves a visual enhancement effect for lung nodules, allowing the human eye to better identify lung nodule regions.
[0085] Finally, the gamma transform result image can be mapped to the colormap function for color processing to obtain a pseudo-color processing result image, which can better identify lung nodule regions.
[0086] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for visually enhanced recognition of pulmonary nodules, characterized in that, The method includes the following steps: A CT scanner is used to acquire grayscale CT images of the lungs, and the threshold selection range in the CT image is obtained based on the grayscale range of the lung CT grayscale image. Based on the threshold selection range, the threshold is traversed one by one, and each threshold is taken as the image segmentation threshold to segment the grayscale image of lung CT to obtain multiple image segmentation result images, and the perimeter and area of the connected region in the result image are obtained for the same pixel. The first membership probability of a pixel in each result image is obtained based on the perimeter and area of the connected region where the pixel is located. The first membership probability sequence is constructed by the first membership probabilities of the same pixel in multiple result images. The threshold used in image thresholding is used as the horizontal axis and the first membership probability is used as the vertical axis. The first membership probability sequence is constructed into a curve of the change of the first membership probability with respect to the threshold, which is called the first curve. The second curve is intercepted on the first curve, and the maximum and minimum points of the first membership probability are obtained based on the second curve. The first curve feature of the second curve is obtained based on the difference between the maximum and minimum points, and the second curve feature of the second curve is obtained based on the slope of the maximum and minimum points. The second membership probability of a pixel is obtained based on the first curve feature and the second curve feature. The lung CT grayscale image is subjected to gamma transformation based on the second membership probability of the pixel to obtain the gamma transformation result image, and then pseudo-color processing is performed on the gamma transformation result image to obtain the pseudo-color processing result image.
2. The method for visually enhanced recognition of pulmonary nodules according to claim 1, characterized in that, The specific steps for obtaining the threshold selection range in the CT image based on the grayscale range of the lung CT image are as follows: A grayscale CT image of the lungs is acquired using a CT scanner, and the number of pixels with the same grayscale value in the image is counted. A grayscale distribution histogram is constructed with the grayscale value on the horizontal axis and the corresponding number of pixels on the vertical axis. The maximum and minimum grayscale values of the corresponding lung CT grayscale image are obtained from the grayscale distribution histogram. The grayscale range of the lung CT grayscale image is defined as the range between the maximum and minimum grayscale values. The grayscale values within the grayscale range of the lung CT grayscale image are then used as the threshold selection range.
3. The method for visually enhanced recognition of pulmonary nodules according to claim 1, characterized in that, The resulting image is preprocessed using morphological opening operations.
4. The method for visually enhanced recognition of pulmonary nodules according to claim 1, characterized in that, The first membership probability sequence is obtained as follows: First membership probability of a pixel in a grayscale image of a lung CT scan: In the formula, A in C refers to the area of the connected region containing the i-th pixel in the n-th result image. in This refers to the perimeter of the connected component containing the i-th pixel in the n-th morphologically processed image, m. in This represents the first membership probability of the i-th pixel in the n-th result image; The first membership probability of a pixel in all the resulting images constitutes the first membership probability sequence.
5. The method for visually enhanced recognition of pulmonary nodules according to claim 1, characterized in that, The first curve feature is obtained as follows: In the formula, f i The fluctuation degree of the second curve representing the i-th pixel in the lung CT grayscale image is denoted as the first curve feature. H represents the number of first membership probabilities greater than the empirical probability threshold in the first membership probability sequence of the i-th pixel in the lung CT grayscale image. M ih The local mean factor represents the h-th first membership probability greater than the empirical probability threshold in the first membership probability sequence of the i-th pixel in a lung CT grayscale image. This represents the mean of the first membership probability in the second curve.
6. The method for visually enhanced recognition of pulmonary nodules according to claim 1, characterized in that, The second curve feature is obtained as follows: Find the pair of maximum and minimum points with the largest difference on the horizontal axis in the second curve; The difference between the maximum and minimum values of the first membership probability in the second curve is denoted as the first difference. The difference between the maximum value of the first membership probability in the second curve and the segmentation threshold corresponding to the minimum target value is recorded as the second difference. The absolute value of the ratio of the first difference to the second difference is used as the feature of the second curve.
7. The method for visually enhanced recognition of pulmonary nodules according to claim 1, characterized in that, The specific steps for obtaining the second membership probability of a pixel based on the first curve feature and the second curve feature are as follows: P i =exp(-βf i )*exp(-k i ) In the formula, exp() is an exponential function with the natural constant as the base; β is an adjustment parameter, and P i The second membership probability of the i-th pixel in a lung CT grayscale image is represented by the first curve feature f. i The second curve represents the degree of fluctuation of the i-th pixel in a lung CT grayscale image, and the feature k of the second curve is... i The slope of the second curve represents the i-th pixel in a grayscale CT image of the lungs.
8. The method for visually enhanced recognition of pulmonary nodules according to claim 1, characterized in that, The specific steps involved in performing a gamma transform on the lung CT grayscale image based on the second membership probability of the pixels to obtain the gamma transform result image are as follows: Where, q i P represents the grayscale value of the i-th pixel in the image; i f represents the second membership probability of the i-th pixel; i This represents the grayscale value of the i-th pixel after gamma transformation. All pixels after gamma transformation constitute the resulting gamma-transformed image.