Oral X-ray film lesion area intelligent identification system based on artificial intelligence

Abstract

The invention relates to the technical field of oral X-ray image recognition, in particular to an intelligent oral X-ray film lesion area recognition system based on artificial intelligence. The method comprises the following steps: obtaining gray level change characteristics of each tooth region in an oral X-ray image according to gray level and gradient distribution in each tooth region; obtaining an optimal factor according to the gray change feature similarity between different regions, and further selecting a standard tooth region; obtaining a ray absorption coefficient according to the gray difference of different areas and the difference of the optimal factors; obtaining caries area possibility of the pixel points according to the positions of the pixel points in the tooth area; and the caries area is identified. According to the method, the caries area in the oral X-ray image is accurately segmented, and finally accurate treatment can be carried out on a patient.

Description

Technical Field

[0001] The present invention relates to the technical field of oral X-ray image recognition, and in particular to an intelligent recognition system for lesion areas in oral X-ray films based on artificial intelligence. Background Art

[0002] Dental X-ray is an imaging method used to diagnose oral diseases. This method uses dental X-ray equipment to penetrate tissues through X-rays and form images on photosensitive films or electronic sensors to show the structure and pathological conditions of teeth, tooth roots, periodontal tissues and surrounding bones. One of the most common lesions in oral X-ray images is dental caries, which is caused by bacteria eroding the hard tissues of teeth in the acidic environment formed in the mouth. It is usually accompanied by tooth sensitivity and pain, and in severe cases, it can even lead to tooth loss. Therefore, it is particularly important to use X-rays to promptly detect early signs of dental caries and treat them.

[0003] When using oral X-ray images to identify dental caries lesions, the characteristics of early caries are not obvious on X-ray images, and may only appear as slight color changes or spots. In addition, different tooth tissues have differences in density and absorption capacity. For example, enamel, dentin and pulp usually show large grayscale differences in X-ray images. In addition, X-ray images can easily produce uneven brightness due to technical factors. Traditional artificial intelligence-based image threshold segmentation methods are difficult to accurately segment caries areas, which is not conducive to accurate diagnosis and treatment planning for patients. Summary of the invention

[0004] In order to solve the technical problem that the characteristics of early caries in oral X-ray images are not obvious, and the X-ray images are prone to uneven brightness due to technical factors, resulting in the inability to accurately segment the caries area and ultimately the inability to accurately treat the patient, the purpose of the present invention is to provide an intelligent recognition system for oral X-ray lesion areas based on artificial intelligence, and the technical solutions adopted are as follows:

[0005] An artificial intelligence-based intelligent recognition system for oral X-ray lesion areas, the system comprising:

[0006] An image acquisition module, used for acquiring an oral X-ray image of a patient;

[0007] A radiation absorption coefficient analysis module is used to obtain the tooth area corresponding to each tooth in the oral X-ray image; select any tooth area as a reference tooth area; obtain the grayscale change characteristics of the reference tooth area according to the grayscale distribution and gradient distribution in the reference tooth area; obtain the preferred factor of the reference tooth area according to the similarity of the grayscale change characteristics between the reference tooth area and each other tooth area; screen all tooth areas according to the preferred factors of all tooth areas to obtain the standard tooth area in all tooth areas; obtain the radiation absorption coefficient of the reference tooth area according to the grayscale difference and the preferred factor difference between the reference tooth area and the standard area;

[0008] A caries area possibility analysis module is used to select a pixel point in a tooth area as a reference pixel point; and obtain the caries area possibility of the reference pixel point according to the position distribution of the reference pixel point in the corresponding tooth area;

[0009] The caries area recognition module is used to identify the caries area in the oral X-ray image according to the radiation absorption coefficient of the tooth area where each pixel point in the oral X-ray image is located and the caries area possibility.

[0010] Furthermore, the method for acquiring the grayscale change feature includes:

[0011] Obtain the minimum circumscribed rectangle of the reference tooth area; obtain all midlines and all diagonals in the minimum circumscribed rectangle;

[0012] Select any line segment from all the midlines and all the diagonals in the minimum circumscribed rectangle as a reference line segment; average the grayscale differences between every two adjacent pixels in the reference line segment to obtain the grayscale change rate of the reference line segment;

[0013] Obtaining the serial number of the pixel point corresponding to the maximum gradient value in the reference line segment as the first value; calculating the ratio between the first value and the number of pixels in the reference line segment as the grayscale mutation factor;

[0014] A vector obtained by combining the grayscale change rate and the grayscale mutation factor is used as a line segment grayscale change feature corresponding to the reference line segment;

[0015] Traverse all the midlines and all the diagonals to obtain the grayscale change features of the line segments corresponding to all the midlines and all the diagonals;

[0016] The vector obtained by combining the grayscale change features of all midlines and all diagonal lines is used as the grayscale change feature of the reference tooth area.

[0017] Furthermore, the method for obtaining the preferred factor of the reference tooth region includes:

[0018] The preferred factor is obtained according to the preferred factor calculation formula, and the preferred factor calculation formula is as follows:

[0019]

[0020] Where ω0 represents the preferred factor of the reference tooth region; N represents the number of tooth regions; f0 represents the grayscale change characteristics of the reference tooth region; f n represents the grayscale change characteristics of the nth other tooth area; cos(f0,f n ) represents the cosine similarity between the grayscale change characteristics of the reference tooth region and the grayscale change characteristics of the nth other tooth region.

[0021] Furthermore, the method for obtaining the standard tooth region includes:

[0022] The tooth region with the largest preference factor is taken as the standard tooth region.

[0023] Furthermore, the method for obtaining the radiation absorption coefficient includes:

[0024] The standard tooth area is obtained according to the calculation formula of the radiation absorption coefficient, and the calculation formula of the radiation absorption coefficient is as follows:

[0025]

[0026] Where, γ0 represents the radiation absorption coefficient of the reference tooth area; represents the mean gray value of the reference tooth area; represents the grayscale mean of the standard tooth area; ω0 represents the preferred factor of the reference tooth area; ω A represents the preferred factor of the standard tooth area; norm( ) represents the normalization function.

[0027] Furthermore, the method for obtaining the possibility of the caries area includes:

[0028] The boundary of each tooth region that contacts the tooth regions of other layers is regarded as the occlusal surface of the tooth region, and the other two boundaries are regarded as the adjacent surfaces of the tooth region;

[0029] Acquire the adjacent surface of the tooth region closest to the reference pixel point as the first adjacent surface, calculate the distance between the reference pixel point and the first adjacent surface as the first distance; calculate the distance between the reference pixel point and the occlusal surface of the tooth region as the second distance;

[0030] Performing least squares fitting on all pixel points of the occlusal surface of the tooth region to which the reference pixel point belongs, obtaining the maximum loss value among all pixel points of the occlusal surface and the pixel point corresponding to the maximum loss value;

[0031] The caries area possibility is obtained according to the caries area possibility calculation formula, and the caries area possibility calculation formula is as follows:

[0032]

[0033] In the formula, represents the possibility of the caries area of ​​the reference pixel; S represents the first distance of the reference pixel; R represents the second distance of the reference pixel; ε max represents the maximum loss value of the pixel point in the occlusal surface of the tooth area to which the reference pixel point belongs; L represents the distance between the reference pixel point and the pixel point corresponding to the maximum loss value; σ max represents the maximum gap width between the tooth region to which the reference pixel belongs and the adjacent tooth region; P represents the distance between the reference pixel and the adjacent surface corresponding to the maximum gap width; norm() represents the normalization function.

[0034] Further, according to the radiation absorption coefficient of the tooth area where each pixel point in the oral X-ray image is located and the possibility of the caries area, the caries area in the oral X-ray image is identified, including:

[0035] Performing image segmentation on the oral X-ray image to obtain a basic segmentation threshold in the oral X-ray image;

[0036] The product of the caries area possibility of the reference pixel point and the ray absorption coefficient of the tooth area to which the reference pixel point belongs is used as the segmentation correction coefficient of the reference pixel point;

[0037] The product of the segmentation correction coefficient of the reference pixel and the basic segmentation threshold is used as the final segmentation threshold of the reference pixel;

[0038] Traversing all pixel points in the oral X-ray image to obtain a final segmentation threshold value for all pixel points;

[0039] Taking the pixel points whose grayscale value is less than the final segmentation threshold as the caries lesion pixel points;

[0040] The area composed of all caries lesion pixels is regarded as the caries area.

[0041] An artificial intelligence-based intelligent identification method for oral X-ray lesion areas, the method comprising:

[0042] Obtain oral X-ray images of the patient;

[0043] Acquire the tooth region corresponding to each tooth in the oral X-ray image; select any tooth region as a reference tooth region; obtain the grayscale change characteristics of the reference tooth region according to the grayscale distribution and gradient distribution in the reference tooth region; obtain the preferred factor of the reference tooth region according to the grayscale change characteristics between the reference tooth region and each other tooth region; screen all tooth regions according to the preferred factors of all tooth regions to obtain the standard tooth region among all tooth regions; obtain the radiation absorption coefficient of the reference tooth region according to the grayscale difference and the preferred factor difference between the reference tooth region and the standard region;

[0044] Select a pixel point in a tooth area as a reference pixel point; obtain the caries area possibility of the reference pixel point according to the position distribution of the reference pixel point in the corresponding tooth area;

[0045] The caries area in the oral X-ray image is identified based on the radiation absorption coefficient of the tooth area where each pixel point in the oral X-ray image is located and the possibility of the caries area.

[0046] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned artificial intelligence-based intelligent recognition system for oral X-ray lesion areas.

[0047] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the above-mentioned intelligent recognition system for lesion areas in oral X-rays based on artificial intelligence are implemented.

[0048] The present invention has the following beneficial effects:

[0049] The present invention obtains an oral X-ray image of a patient; since caries only exists in the tooth area, and each tooth is an independent entity, in order to reduce the interference of possible brightness unevenness on lesion identification, each tooth is divided into an independent area; since there is a phenomenon of uneven grayscale values ​​in the oral X-ray image, it is easy to be confused with the caries area, resulting in errors in subsequent image recognition, so according to the grayscale distribution and gradient distribution in the reference tooth area, the grayscale change characteristics of the reference tooth area are obtained; since the tooth area with a large deviation in the grayscale change characteristics indicates that the tooth area may be interfered by artifact noise, according to the grayscale change characteristics between the reference tooth area and each other tooth area, the preferred factor of the reference tooth area is obtained; the standard tooth area is screened out by the preferred factor for comparison with the other tooth areas; since the overall grayscale value of the tooth area interfered by noise is relatively Compared with the overall grayscale value of the tooth area not disturbed by noise, it is lower, so the reflection coefficient coefficient of the tooth area is analyzed by the grayscale difference and the preferred factor difference between each other tooth area and the standard tooth area, and the X-ray absorption degree in the tooth area is reflected by the reflection absorption coefficient; since caries mainly include occlusal caries and interproximal caries, both of which appear at the tooth boundary, and the larger the gap between adjacent teeth and the more uneven the tooth occlusal surface, the more likely it is that food residues and bacteria will gather, so these positions are also more likely to have caries lesions; so each pixel reference pixel of each tooth area is analyzed, and the possibility of the caries area of ​​the reference pixel is obtained according to the position distribution of the reference pixel in the tooth area to which it belongs; according to the ray absorption coefficient of the tooth area where each pixel in the oral X-ray image is located and the possibility of the caries area, the caries area in the oral X-ray image is identified. The present invention accurately segments the caries area in the oral X-ray image, and finally can accurately treat the patient. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0051] Figure 1 A flow chart of an artificial intelligence-based intelligent identification system for lesion areas in oral X-rays provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation method, structure, features and effects of an intelligent recognition system for oral X-ray lesion areas based on artificial intelligence proposed by the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0053] Unless defined otherwise, 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 belongs.

[0054] The following is a detailed description of a specific scheme of an intelligent identification system for lesion areas in oral X-rays based on artificial intelligence provided by the present invention in conjunction with the accompanying drawings.

[0055] See also Figure 1 , which shows an artificial intelligence-based oral X-ray lesion area intelligent recognition system provided by an embodiment of the present invention, the system includes: an image acquisition module S101, a ray absorption coefficient analysis module S102, a caries area possibility analysis module S103 and a caries area recognition module S104, and the specific operations are as follows:

[0056] The image acquisition module S101 is used to acquire the patient's oral X-ray image.

[0057] The embodiments of the present invention are mainly used in scenarios of identifying caries areas inside the oral cavity of a patient. In reality, medical personnel usually obtain oral X-ray images of the patient and further determine whether the patient has caries based on the oral X-ray images.

[0058] In one embodiment of the present invention, an oral X-ray image of a patient is obtained by a dental X-ray device. Since the oral X-ray image itself is a grayscale image, the grayscale value of the pixel reflects the degree of absorption of the X-ray by the object, wherein the stronger the absorption degree, the higher the grayscale value. The semantic segmentation technology is used to extract the periodontal part to avoid other irrelevant areas affecting the subsequent recognition effect.

[0059] The radiation absorption coefficient analysis module S102 is used to obtain the tooth area corresponding to each tooth in the oral X-ray image; select any tooth area as the reference tooth area; obtain the grayscale change characteristics of the reference tooth area according to the grayscale distribution and gradient distribution in the reference tooth area; obtain the preferred factor of the reference tooth area according to the similarity of the grayscale change characteristics between the reference tooth area and each other tooth area; screen all tooth areas according to the preferred factors of all tooth areas to obtain the standard tooth area among all tooth areas; obtain the radiation absorption coefficient of the reference tooth area according to the grayscale difference and the preferred factor difference between the reference tooth area and the standard area.

[0060] In the prior art, when segmenting and identifying the dental caries area, due to the large difference in the grayscale representation of different tooth tissues in the oral X-ray image, and the fact that the oral X-ray image is prone to uneven brightness due to technical factors, there are many areas in the oral X-ray image that are similar to the features of the caries area, resulting in a high error rate in subsequent image recognition, delaying the best treatment period for patients. In reality, caries only exists in the tooth area, and each tooth is an independent entity. In order to reduce the interference of possible brightness unevenness on lesion identification, in the embodiment of the present invention, each tooth can first be divided into an independent area, and then each area can be analyzed separately.

[0061] In one embodiment of the present invention, since teeth usually have obvious boundaries and shape features, superpixel segmentation technology is used to segment the tooth area in the acquired oral X-ray image into several superpixel blocks, where each tooth belongs to a superpixel block. After the superpixel segmentation is completed, the superpixel blocks are marked using semantic segmentation technology to identify all the tooth areas therein, thereby achieving separation of different tooth areas. It should be noted that superpixel segmentation technology is a technical means well known to those skilled in the art and will not be described in detail here.

[0062] Due to the uneven brightness in oral X-ray images, they are easily confused with caries areas, resulting in errors in subsequent image recognition. Since each tooth has a strong visual consistency, the characteristics of the pixel grayscale in a single tooth area can be used to calculate the corresponding ray absorption coefficient, which is used to indicate its absorption degree of X-rays, so that different basic parameters can be set for different tooth areas when identifying caries lesions; and in order to avoid the influence of possible artifacts in the image on the ray absorption coefficient, the consistency of grayscale changes within the tooth area must also be considered during calculation to enhance the algorithm's anti-noise ability. Therefore, in the embodiment of the present invention, the grayscale change characteristics of the reference tooth area are obtained based on the grayscale distribution and gradient distribution in the reference tooth area.

[0063] Preferably, in one embodiment of the present invention, the method for acquiring grayscale change characteristics includes:

[0064] Since the shape of each tooth area is mostly rectangular, the minimum circumscribed rectangle of the reference tooth area is obtained; all the midlines and all the diagonals in the minimum circumscribed rectangle are obtained, and the grayscale change characteristics of the entire tooth area are analyzed through the grayscale changes on the two midlines and the two diagonals.

[0065] Select any line segment from all the midlines and all the diagonals in the minimum circumscribed rectangle as the reference line segment; calculate the average of the grayscale difference between every two adjacent pixels in the reference line segment to obtain the grayscale change rate of the reference line segment, wherein the grayscale change rate among all the midlines and all the diagonals can reflect the consistency of the grayscale change in the tooth area.

[0066] The serial number of the pixel point corresponding to the maximum gradient value in the reference line segment is obtained as the first value; the ratio between the first value and the number of pixels in the reference line segment is calculated as the grayscale mutation factor of the reference line segment, wherein the grayscale mutation factors in all midlines and all diagonals reflect the positional distribution of different structural tissues of the tooth.

[0067] The vector obtained by combining the grayscale change rate and the grayscale mutation factor is used as the corresponding line segment grayscale change feature of the reference line segment;

[0068] Traverse all the midlines and all the diagonals to obtain the grayscale change features of the line segments corresponding to all the midlines and all the diagonals;

[0069] The vector obtained by combining the grayscale change features of all midlines and all diagonal lines is used as the grayscale change feature of the reference tooth area.

[0070] Under normal circumstances, the grayscale change characteristics in the reference tooth area are similar to the grayscale change characteristics in each other tooth area. If a tooth area with a large deviation in the grayscale change characteristics appears, it means that the tooth area may be interfered by artifact noise. Therefore, in an embodiment of the present invention, the preferred factor of the reference tooth area is obtained based on the similarity of the grayscale change characteristics between the reference tooth area and each other tooth area.

[0071] Preferably, in one embodiment of the present invention, the method for obtaining the optimal factor includes:

[0072] The preferred factor is obtained according to the preferred factor calculation formula. The preferred factor calculation formula is as follows:

[0073]

[0074] Where ω0 represents the preferred factor of the reference tooth region; N represents the number of tooth regions; f0 represents the grayscale change characteristics of the reference tooth region; fn represents the grayscale change characteristics of the nth other tooth area; cos(f0,f n ) represents the cosine similarity between the grayscale change characteristics of the reference tooth region and the grayscale change characteristics of the nth other tooth region.

[0075] In the preferred factor calculation formula, the greater the cosine similarity between the grayscale change characteristics of the reference tooth area and the grayscale change characteristics of each other tooth area, the more representative the grayscale change characteristics of the reference tooth area are among all tooth areas. At this time, the reference tooth area is more referenceable, that is, the preferred factor of the reference tooth area is larger.

[0076] Preferably, in one embodiment of the present invention, since the overall grayscale value of the tooth area affected by noise is lower than the overall grayscale value of the tooth area not affected by noise, the tooth area with the largest preferred factor is used as the standard tooth area, and the reflection coefficient of the tooth area is analyzed by the grayscale difference and the preferred factor difference between each other tooth area and the standard tooth area, and the reflection absorption coefficient is used to reflect the degree of X-ray absorption in the tooth area.

[0077] Preferably, in one embodiment of the present invention, the method for obtaining the radiation absorption coefficient includes:

[0078] The standard tooth area is obtained according to the calculation formula of the radiation absorption coefficient. The calculation formula of the radiation absorption coefficient is as follows:

[0079]

[0080] Where, γ0 represents the radiation absorption coefficient of the reference tooth area; represents the mean gray value of the reference tooth area; represents the grayscale mean of the standard tooth area; ω0 represents the preferred factor of the reference tooth area; ω A represents the preferred factor of the standard tooth area; norm( ) represents the normalization function.

[0081] In the calculation formula of the radiation absorption coefficient, the overall grayscale difference between the reference tooth area and the standard tooth area The larger the value, the greater the overall gray value of the reference tooth area, and the greater the radiation absorption coefficient of the reference tooth area; the ratio of the preferred factor of the reference tooth area to the preferred factor of the standard tooth area The larger it is, the smaller the difference in grayscale characteristics between the reference tooth area and the standard tooth area is. At this time, the possibility that the reference tooth area is affected by noise is less, and the degree of radiation absorption by the reference tooth area is higher, that is, the radiation absorption coefficient of the reference tooth area is larger.

[0082] The caries area possibility analysis module S103 is used to select a pixel point in a tooth area as a reference pixel point; and obtain the caries area possibility of the reference pixel point according to the position distribution of the reference pixel point in the corresponding tooth area.

[0083] Generally, the cause of dental caries is the presence of food residues or bacterial residues on the tooth surface. These residues form dental plaque on the tooth surface and produce acidic substances that cause demineralization and organic matter collapse, ultimately causing dental caries. Therefore, when identifying and detecting dental caries lesion areas, more attention should be paid to areas that are prone to the accumulation of food residues and bacteria. Dental caries mainly include occlusal caries and interproximal caries, both of which occur at the tooth boundaries. The larger the gap between adjacent teeth and the more uneven the occlusal surface of the teeth, the more likely it is that food residues and bacteria will accumulate. Therefore, these locations are also more likely to have dental caries lesions. Therefore, in the embodiment of the present invention, each pixel reference pixel of each tooth area is analyzed, and the possibility of the dental caries area of ​​the reference pixel is obtained according to the position distribution of the reference pixel in the tooth area to which it belongs.

[0084] Preferably, in one embodiment of the present invention, the method for obtaining the possibility of a caries area includes:

[0085] The boundary of each tooth region that contacts the tooth regions of other layers is regarded as the occlusal surface of the tooth region, and the other two boundaries are regarded as the adjacent surfaces of the tooth region.

[0086] The adjacent surface of the tooth area closest to the reference pixel point is obtained as the first adjacent surface, and the distance between the reference pixel point and the first adjacent surface is calculated as the first distance; the distance between the reference pixel point and the occlusal surface of the tooth area is calculated as the second distance.

[0087] The least squares method is used to fit all the pixels of the occlusal surface of the tooth area to which the reference pixel belongs, and the maximum loss value among all the pixels of the occlusal surface and the pixel corresponding to the maximum loss value are obtained. The larger the loss value of the pixel of the occlusal surface, the deeper the depression of the occlusal surface and the greater the possibility of caries.

[0088] The caries area possibility is obtained according to the caries area possibility calculation formula. The caries area possibility calculation formula is as follows:

[0089]

[0090] In the formula, represents the possibility of the caries area of ​​the reference pixel; S represents the first distance of the reference pixel; R represents the second distance of the reference pixel; ε max represents the maximum loss value of the pixel point in the occlusal surface of the tooth area to which the reference pixel point belongs; L represents the distance between the reference pixel point and the pixel point corresponding to the maximum loss value; σ maxrepresents the maximum gap width between the tooth region to which the reference pixel belongs and the adjacent tooth region; P represents the distance between the reference pixel and the adjacent surface corresponding to the maximum gap width; norm() represents the normalization function.

[0091] In the calculation formula of the probability of caries area, the distance between the reference pixel point and the adjacent surface of the tooth area and the occlusal surface of the tooth area The smaller it is, the higher the possibility of caries at the location of the reference pixel, that is, the greater the possibility of caries area of ​​the reference pixel; the maximum loss value ε of the pixel in the occlusal surface max The larger the value, the greater the degree of depression of the occlusal surface of the tooth area, and the greater the distance between the reference pixel point and the maximum depression of the occlusal surface (the location of the occlusal pixel corresponding to the maximum loss value), the greater the possibility of dental caries at the location of the reference pixel point, that is, the greater the possibility of dental caries area of ​​the reference pixel point; the maximum gap width σ between the tooth area to which the reference pixel point belongs and the adjacent tooth area max The larger the value is, the more likely a caries area is to appear in this gap, and the smaller the distance P between the reference pixel and the adjacent surface corresponding to the maximum gap width is, the greater the possibility of caries occurring at the location of the reference pixel, that is, the greater the possibility of a caries area at the reference pixel.

[0092] The caries area identification module S104 is used to identify the caries area in the oral X-ray image according to the radiation absorption coefficient of the tooth area where each pixel point in the oral X-ray image is located and the possibility of the caries area.

[0093] The image segmentation threshold of the conventional image segmentation algorithm cannot accurately segment the caries area. Therefore, preferably, in the embodiment of the present invention, the caries area in the oral X-ray image is identified according to the radiation absorption coefficient of the tooth area where each pixel point in the oral X-ray image is located and the possibility of the caries area, including:

[0094] Image segmentation is performed on the oral X-ray image to obtain a basic segmentation threshold in the oral X-ray image. It should be noted that, in one embodiment of the present invention, the Otsu method is used to perform image segmentation on the oral X-ray image, and the method of using the Otsu method to obtain the basic segmentation threshold in the oral X-ray image is a technical means well known to those skilled in the art and will not be elaborated here.

[0095] The product of the caries area possibility of the reference pixel point and the radiation absorption coefficient of the tooth area to which the reference pixel point belongs is used as the segmentation correction coefficient of the reference pixel point; the product of the segmentation correction coefficient of the reference pixel point and the basic segmentation threshold is used as the final segmentation threshold of the reference pixel point; in one embodiment of the present invention, the final segmentation threshold calculation formula is as follows:

[0096]

[0097] Where, F represents the final segmentation threshold of the reference pixel; B represents the basic segmentation threshold in the oral X-ray image; γ represents the radiation absorption coefficient of the reference tooth area; Indicates the possibility of caries area at the reference pixel.

[0098] In the final segmentation threshold calculation formula, the greater the probability that the radiation absorption coefficient of the tooth area where the reference pixel is located is the caries area of ​​the reference pixel, the greater the image segmentation threshold required for the reference pixel. Therefore, at this time, the final segmentation threshold of the reference pixel is larger.

[0099] Traverse all pixels in the oral X-ray image to obtain the final segmentation threshold of all pixels; take pixels with gray values ​​less than the final segmentation threshold as caries lesion pixels; and take the area composed of all caries lesion pixels as the caries area.

[0100] The caries area in the oral X-ray image is displayed on the monitor through image visualization method, so that medical personnel can view the abnormalities in the oral cavity more intuitively.

[0101] At this point, the intelligent recognition of caries areas in oral X-ray images is completed.

[0102] In summary, an oral X-ray image of the patient is obtained; a tooth area corresponding to each tooth in the oral X-ray image is obtained; one tooth area is selected as a reference tooth area; the grayscale change characteristics of the reference tooth area are obtained according to the grayscale distribution and gradient distribution in the reference tooth area; the preferred factor of the reference tooth area is obtained according to the grayscale change characteristics between the reference tooth area and each other tooth area; all tooth areas are screened according to the preferred factors of all tooth areas to obtain a standard tooth area in all tooth areas; the ray absorption coefficient of the reference tooth area is obtained according to the grayscale difference and the preferred factor difference between the reference tooth area and the standard area; a pixel point in a tooth area is selected as a reference pixel point; the possibility of the caries area of ​​the reference pixel point is obtained according to the position distribution of the reference pixel point in the tooth area to which it belongs; the caries area in the oral X-ray image is identified according to the ray absorption coefficient of the tooth area where each pixel point in the oral X-ray image is located and the possibility of the caries area.

[0103] The second object of the present invention is to provide an artificial intelligence-based intelligent identification method for oral X-ray lesion areas, comprising:

[0104] Obtain oral X-ray images of the patient;

[0105] Obtain the tooth region corresponding to each tooth in the oral X-ray image; select any tooth region as the reference tooth region; obtain the grayscale change characteristics of the reference tooth region according to the grayscale distribution and gradient distribution in the reference tooth region; obtain the preferred factor of the reference tooth region according to the grayscale change characteristics between the reference tooth region and each other tooth region; screen all tooth regions according to the preferred factors of all tooth regions to obtain the standard tooth region in all tooth regions; obtain the radiation absorption coefficient of the reference tooth region according to the grayscale difference and the preferred factor difference between the reference tooth region and the standard region;

[0106] Select a pixel point in a tooth area as a reference pixel point; obtain the caries area possibility of the reference pixel point according to the position distribution of the reference pixel point in the corresponding tooth area;

[0107] The caries area in the oral X-ray image is identified based on the radiation absorption coefficient of the tooth area where each pixel point in the oral X-ray image is located and the possibility of the caries area.

[0108] The third purpose of an embodiment of the present invention is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the above-mentioned artificial intelligence-based intelligent recognition system for lesion areas in oral X-ray films are implemented.

[0109] The fourth purpose of an embodiment of the present invention is to provide a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the above-mentioned artificial intelligence-based oral X-ray lesion area intelligent recognition system.

[0110] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0111] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. An artificial intelligence-based intelligent recognition system for oral X-ray lesion areas, characterized in that: The system comprises: An image acquisition module, used for acquiring an oral X-ray image of a patient; A radiation absorption coefficient analysis module is used to obtain the tooth area corresponding to each tooth in the oral X-ray image; select any tooth area as a reference tooth area; obtain the grayscale change characteristics of the reference tooth area according to the grayscale distribution and gradient distribution in the reference tooth area; obtain the preferred factor of the reference tooth area according to the similarity of the grayscale change characteristics between the reference tooth area and each other tooth area; screen all tooth areas according to the preferred factors of all tooth areas to obtain the standard tooth area in all tooth areas; obtain the radiation absorption coefficient of the reference tooth area according to the grayscale difference and the preferred factor difference between the reference tooth area and the standard area; A caries area possibility analysis module is used to select a pixel point in a tooth area as a reference pixel point; and obtain the caries area possibility of the reference pixel point according to the position distribution of the reference pixel point in the corresponding tooth area; The caries area recognition module is used to identify the caries area in the oral X-ray image according to the radiation absorption coefficient of the tooth area where each pixel point in the oral X-ray image is located and the caries area possibility.

2. The artificial intelligence-based intelligent recognition system for oral X-ray lesion areas according to claim 1, characterized in that: The method for acquiring the grayscale change feature includes: Obtain the minimum circumscribed rectangle of the reference tooth area; obtain all midlines and all diagonals in the minimum circumscribed rectangle; Select any line segment from all the midlines and all the diagonals in the minimum circumscribed rectangle as a reference line segment; average the grayscale differences between every two adjacent pixels in the reference line segment to obtain the grayscale change rate of the reference line segment; Obtaining the serial number of the pixel point corresponding to the maximum gradient value in the reference line segment as the first value; calculating the ratio between the first value and the number of pixels in the reference line segment as the grayscale mutation factor; A vector obtained by combining the grayscale change rate and the grayscale mutation factor is used as a line segment grayscale change feature corresponding to the reference line segment; Traverse all the midlines and all the diagonals to obtain the grayscale change features of the line segments corresponding to all the midlines and all the diagonals; The vector obtained by combining the grayscale change features of all midlines and all diagonal lines is used as the grayscale change feature of the reference tooth area.

3. The artificial intelligence-based intelligent recognition system for oral X-ray lesion areas according to claim 1, characterized in that: The method for obtaining the preferred factor of the reference tooth region comprises: The preferred factor is obtained according to the preferred factor calculation formula, and the preferred factor calculation formula is as follows: Where ω0 represents the preferred factor of the reference tooth region; N represents the number of tooth regions; f0 represents the grayscale change characteristics of the reference tooth region; f n represents the grayscale change characteristics of the nth other tooth area; cos(f0, f n ) represents the cosine similarity between the grayscale change characteristics of the reference tooth region and the grayscale change characteristics of the nth other tooth region.

4. The artificial intelligence-based intelligent recognition system for oral X-ray lesion areas according to claim 1, characterized in that: The method for obtaining the standard tooth region comprises: The tooth region with the largest preference factor is taken as the standard tooth region.

5. The artificial intelligence-based intelligent recognition system for oral X-ray lesion areas according to claim 1, characterized in that: The method for obtaining the radiation absorption coefficient includes: The standard tooth area is obtained according to the calculation formula of the radiation absorption coefficient, and the calculation formula of the radiation absorption coefficient is as follows: Where, γ0 represents the radiation absorption coefficient of the reference tooth area; represents the mean gray value of the reference tooth area; represents the grayscale mean of the standard tooth area; ω0 represents the preferred factor of the reference tooth area; ω A represents the preferred factor of the standard tooth area; norm() represents the normalization function.

6. The artificial intelligence-based intelligent recognition system for oral X-ray lesion areas according to claim 1, characterized in that: The method for obtaining the possibility of the caries area includes: The boundary of each tooth region that contacts the tooth regions of other layers is regarded as the occlusal surface of the tooth region, and the other two boundaries are regarded as the adjacent surfaces of the tooth region; Acquire the adjacent surface of the tooth region closest to the reference pixel point as the first adjacent surface, calculate the distance between the reference pixel point and the first adjacent surface as the first distance; calculate the distance between the reference pixel point and the occlusal surface of the tooth region as the second distance; Performing least squares fitting on all pixel points of the occlusal surface of the tooth region to which the reference pixel point belongs, obtaining the maximum loss value among all pixel points of the occlusal surface and the pixel point corresponding to the maximum loss value; The caries area possibility is obtained according to the caries area possibility calculation formula, and the caries area possibility calculation formula is as follows: In the formula, represents the possibility of the caries area of ​​the reference pixel; S represents the first distance of the reference pixel; R represents the second distance of the reference pixel; ε max represents the maximum loss value of the pixel point in the occlusal surface of the tooth area to which the reference pixel point belongs; L represents the distance between the reference pixel point and the pixel point corresponding to the maximum loss value; σ max It represents the maximum gap width between the tooth region to which the reference pixel belongs and the adjacent tooth region; P represents the distance between the reference pixel and the adjacent surface corresponding to the maximum gap width; norm() represents the normalization function.

7. The artificial intelligence-based intelligent recognition system for oral X-ray lesion areas according to claim 1, characterized in that: According to the radiation absorption coefficient of the tooth area where each pixel point in the oral X-ray image is located and the possibility of the caries area, the caries area in the oral X-ray image is identified, including: Performing image segmentation on the oral X-ray image to obtain a basic segmentation threshold in the oral X-ray image; The product of the caries area possibility of the reference pixel point and the ray absorption coefficient of the tooth area to which the reference pixel point belongs is used as the segmentation correction coefficient of the reference pixel point; The product of the segmentation correction coefficient of the reference pixel and the basic segmentation threshold is used as the final segmentation threshold of the reference pixel; Traversing all pixel points in the oral X-ray image to obtain a final segmentation threshold value for all pixel points; Taking the pixel points whose grayscale value is less than the final segmentation threshold as the caries lesion pixel points; The area composed of all caries lesion pixels is regarded as the caries area.

8. An artificial intelligence-based intelligent identification method for lesion areas in oral X-rays, characterized in that: The method comprises: Obtain oral X-ray images of the patient; Acquire the tooth region corresponding to each tooth in the oral X-ray image; select any tooth region as a reference tooth region; obtain the grayscale change characteristics of the reference tooth region according to the grayscale distribution and gradient distribution in the reference tooth region; obtain the preferred factor of the reference tooth region according to the grayscale change characteristics between the reference tooth region and each other tooth region; screen all tooth regions according to the preferred factors of all tooth regions to obtain the standard tooth region among all tooth regions; obtain the radiation absorption coefficient of the reference tooth region according to the grayscale difference and the preferred factor difference between the reference tooth region and the standard region; Select a pixel point in a tooth area as a reference pixel point; obtain the caries area possibility of the reference pixel point according to the position distribution of the reference pixel point in the corresponding tooth area; The caries area in the oral X-ray image is identified based on the radiation absorption coefficient of the tooth area where each pixel point in the oral X-ray image is located and the possibility of the caries area.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, the steps of the artificial intelligence-based intelligent recognition system for lesion areas in oral X-rays as described in any one of claims 1 to 7 are implemented.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the artificial intelligence-based intelligent identification system for oral X-ray lesion areas as described in any one of claims 1 to 7 are implemented.

Images