CN118115415A - Ultrasonic image optimization processing method and system based on artificial intelligence - Google Patents
Ultrasonic image optimization processing method and system based on artificial intelligence Download PDFInfo
- Publication number
- CN118115415A CN118115415A CN202410523910.1A CN202410523910A CN118115415A CN 118115415 A CN118115415 A CN 118115415A CN 202410523910 A CN202410523910 A CN 202410523910A CN 118115415 A CN118115415 A CN 118115415A
- Authority
- CN
- China
- Prior art keywords
- sequence
- value
- segment
- abnormal
- ultrasonic image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 27
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 26
- 238000003672 processing method Methods 0.000 title claims abstract description 23
- 230000002159 abnormal effect Effects 0.000 claims abstract description 95
- 210000001685 thyroid gland Anatomy 0.000 claims abstract description 94
- 230000003902 lesion Effects 0.000 claims abstract description 68
- 238000000034 method Methods 0.000 claims abstract description 46
- 238000002604 ultrasonography Methods 0.000 claims abstract description 36
- 230000003044 adaptive effect Effects 0.000 claims abstract description 18
- 238000012216 screening Methods 0.000 claims abstract description 7
- 238000001514 detection method Methods 0.000 claims description 20
- 238000004364 calculation method Methods 0.000 claims description 18
- 230000000875 corresponding effect Effects 0.000 claims description 18
- 238000004422 calculation algorithm Methods 0.000 claims description 15
- 230000006870 function Effects 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 4
- 230000002596 correlated effect Effects 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 abstract description 3
- 230000000694 effects Effects 0.000 description 7
- 238000001914 filtration Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Ultra Sonic Daignosis Equipment (AREA)
Abstract
The invention relates to the technical field of image processing, in particular to an ultrasonic image optimization processing method and system based on artificial intelligence, comprising the following steps: acquiring a thyroid ultrasonic image and a gray value sequence; dividing the gray value sequence into sequence segments according to the gray value sequence; obtaining a fluctuation characteristic value according to the sequence segment; screening abnormal sequence segments according to the fluctuation characteristic values; obtaining an abnormal updating sequence segment according to the abnormal sequence segment; obtaining a real lesion area according to the abnormal updating sequence segment; obtaining a characteristic value of a lesion area according to the real lesion area; obtaining a self-adaptive significance threshold according to the characteristic value of the lesion area; obtaining a salient region and a non-salient region according to the adaptive saliency threshold; and obtaining a thyroid ultrasound enhanced image according to the salient region and the non-salient region. According to the method, the accurate and credible self-adaptive significance threshold is obtained, and the thyroid ultrasonic image is enhanced, so that the enhancing effect of the thyroid ultrasonic image is improved.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to an ultrasonic image optimization processing method and system based on artificial intelligence.
Background
Ultrasound images are a common medical image, but because the imaging process is interfered by various factors, such as scattering, attenuation, noise and the like, the quality of the ultrasound images is poor, so that denoising and enhancement processing are required for the ultrasound images, and the quality of the ultrasound images and the diagnostic accuracy of doctors can be effectively improved through denoising and enhancement of the ultrasound images. Histogram equalization is a common image enhancement method, but the histogram equalization is to enhance the whole image globally directly, and enhance noise in the image at the same time of enhancing the image, so that local detail information in the image is lost. Therefore, the target area needing enhancement needs to be found first, the histogram enhancement processing is carried out on the target area, and the filtering operation is carried out on the non-target area. And further an enhanced image can be obtained.
For thyroid ultrasound images, the target area is a lesion area of a structure associated with the thyroid. The saliency detection CA algorithm is a commonly used target area extraction method, and firstly, an image is divided into a plurality of areas, the saliency degree of each area is represented by setting a saliency value for the area, a threshold value is set, and the area with the saliency value larger than the threshold value is marked as the saliency area, so that a required target area can be obtained. For a thyroid ultrasound image, since a plurality of tissues are contained in the gray level image thereof, most of the tissues are presented with low gray level values and the corresponding low gray level values are not completely identical. If a fixed threshold is used in the conventional algorithm, a part of tissues is contained in the salient region, so that the subsequent enhancement effect is affected, and the optimization result of the ultrasonic image does not reach the standard.
Disclosure of Invention
The invention provides an ultrasonic image optimization processing method and system based on artificial intelligence, which are used for solving the existing problems.
The ultrasonic image optimization processing method and system based on artificial intelligence adopt the following technical scheme:
The invention provides an ultrasonic image optimization processing method based on artificial intelligence, which comprises the following steps:
acquiring a plurality of thyroid ultrasonic images; acquiring a plurality of gray value sequences in any thyroid ultrasonic image; dividing each gray value sequence into a plurality of sequence segments according to the element value of each gray value sequence;
Obtaining a fluctuation characteristic value of each sequence segment according to the element value of each sequence segment; screening abnormal sequence segments from all the sequence segments according to the fluctuation characteristic value of each sequence segment; obtaining an abnormal updating sequence segment according to the abnormal sequence segment;
obtaining a real lesion area according to the abnormal updating sequence segment;
Obtaining a lesion area characteristic value of each thyroid ultrasonic image according to the real lesion area in each thyroid ultrasonic image; obtaining an adaptive significance threshold of each thyroid ultrasonic image according to the characteristic value of the lesion area of each thyroid ultrasonic image;
performing saliency detection on each thyroid ultrasonic image based on the adaptive saliency threshold to obtain a salient region and a non-salient region; and respectively carrying out enhancement treatment on the salient region and the non-salient region to obtain a thyroid ultrasound enhanced image.
Further, according to the element value of each gray value sequence, dividing each gray value sequence into a plurality of sequence segments, including the following specific steps:
For any gray value sequence, calculating the difference value of each element value and the left element value and the right element value of the element value respectively in the interval from the second element value to the last element value to obtain the left difference value and the right difference value of each element value, and recording the element value with the product of the left difference value and the right difference value being greater than 0 as a sequence extremum; and marking a sequence interval between every two sequence extreme values as a sequence segment to obtain a plurality of sequence segments of the gray value sequence.
Further, the step of obtaining the fluctuation characteristic value of each sequence segment according to the element value of each sequence segment comprises the following specific steps:
In any one gray value sequence, for any one sequence segment, acquiring the largest two sequence extremum and the smallest two sequence extremum in the intervals of the sequence segment and the two adjacent sequence segments, marking the absolute value of the difference value of the largest two sequence extremum as the maximum value distance of the sequence segment, marking the absolute value of the difference value of the smallest two sequence extremum as the minimum value distance of the sequence segment, and calculating the fluctuation characteristic value of the j-th sequence segment in the gray value sequence by the following steps:
；
In the method, in the process of the invention, A fluctuation feature value representing a j-th sequence segment; /(I)Representing the average of all gray values in the kth sequence segment; /(I)Representing the average of all gray values in the k+1th sequence segment; /(I)Representing the minimum distance of the j-th sequence segment; representing the maximum distance of the j-th sequence segment; /(I) Representing the average value of all gray values in the gray value sequence of the j-th sequence segment; /(I)Representing an absolute value function; /(I)Representing the normalization function.
Further, the step of screening abnormal sequence segments from all the sequence segments according to the fluctuation characteristic value of each sequence segment comprises the following specific steps:
In any gray value sequence, arranging the fluctuation characteristic values according to the sequence of sequence segments to obtain a fluctuation characteristic value sequence; based on a plurality of element values in the fluctuation characteristic value sequence, obtaining abnormal element values in the fluctuation characteristic value sequence by utilizing a local abnormal factor detection algorithm, and marking a sequence segment corresponding to the abnormal element values as an abnormal sequence segment.
Further, the step of obtaining the abnormal update sequence segment according to the abnormal sequence segment comprises the following specific steps:
in any gray value sequence, a plurality of continuous abnormal sequence segments are combined into a new abnormal sequence segment, and the new abnormal sequence segment and the abnormal sequence segment which is not combined are recorded as abnormal updating sequence segments.
Further, the step of obtaining the real lesion area according to the abnormal updating sequence segment comprises the following specific steps:
(1) Recording the first gray value sequence as a target sequence;
(2) Recording the first abnormal updating sequence segment which does not participate in the correlation calculation on the target sequence as a target sequence segment, and performing the step (3); if no abnormal updating sequence segment which does not participate in the correlation calculation exists on the target sequence, marking the next gray value sequence of the target sequence as a new target sequence, repeating the step (2) until the target sequence is the last gray value sequence, and stopping repeating;
(3) Sequentially calculating the correlation degree of all the target sequence segments and each abnormal updating sequence segment in the next gray value sequence of the last target sequence segment until the correlation degree is greater than a preset correlation threshold value When the correlation degree is stopped being calculated, the correlation degree between the target sequence segments is larger than a preset correlation threshold/>, and the target sequence segments are correlated with the target sequence segmentsTaking the abnormal updating sequence section of the (c) as a target sequence section, repeating the step (3) until no correlation degree with all target sequence sections in the next gray value sequence is larger than a preset correlation threshold/>Stopping repeating the step (3) when the sequence segment is abnormally updated, and performing the step (4);
(4) And (3) marking the pixel point set corresponding to all the target sequence segments as a real lesion area, and repeating the step (2).
Further, the step of sequentially calculating the correlation between all the target sequence segments and each abnormal update sequence segment in the next gray value sequence of the gray value sequence where the last target sequence segment is located includes the following specific steps:
Acquiring sequence values of a first element value and a last element value of each abnormal updating sequence segment in a gray value sequence respectively, and recording the sequence values as a starting sequence value and an ending sequence value of the abnormal updating sequence segment respectively;
Marking the next gray value sequence of the last target sequence segment as a reference sequence, sequencing the target sequence segment and the abnormal updating sequence segment in any one abnormal updating sequence segment in all the target sequence segments and the reference sequence, and collectively grouping the target sequence segment and the abnormal updating sequence segment as sequence segments, calculating the absolute value of the difference value of the starting sequence values and the absolute value of the difference value of the ending sequence values of every two adjacent sequence segments, marking the absolute value of the maximum difference value of the starting sequence values as an initial difference value, and marking the absolute value of the maximum difference value of the ending sequence values as an ending difference value; the specific calculation method of the correlation degree between all the target sequence segments and the xth abnormal updating sequence segment in the reference sequence comprises the following steps:
；
In the method, in the process of the invention, Representing the correlation of all target sequence segments with the x-th abnormally updated sequence segment in the reference sequence,Representing initial difference values of the xth abnormal updating sequence segment in all the target sequence segments and the reference sequence; and the ending difference value of the xth abnormal updating sequence segment in the target sequence segment and the reference sequence is represented.
Further, the specific calculation method for obtaining the characteristic value of the lesion area of each thyroid ultrasonic image according to the real lesion area in each thyroid ultrasonic image comprises the following steps:
；
In the method, in the process of the invention, Representing the characteristic value of a lesion area of a v-th thyroid ultrasonic image; /(I)Representing the number of real lesion areas in the v-th thyroid ultrasound image; /(I)The number of pixel points of the y-th real lesion area in the v-th thyroid ultrasonic image is represented; /(I)And representing the number of all pixel points in the v-th thyroid ultrasonic image.
Further, the adaptive significance threshold value of each thyroid ultrasound image is obtained according to the feature value of the lesion area of each thyroid ultrasound image, and the specific calculation method comprises the following steps:
；
In the method, in the process of the invention, Representing an adaptive saliency threshold of the v-th thyroultrasonic image; /(I)Representing a preset experience significance threshold; /(I)An adaptive symbol value representing a saliency threshold of the v-th thyroultrasonic image; /(I)Representing the characteristic value of a lesion area of a v-th thyroid ultrasonic image; /(I)Representing the number of acquired thyroid ultrasound images; /(I)Representing a lesion area characteristic value of the r-th thyroid ultrasonic image; /(I)Representing an absolute value function;
The method for acquiring the self-adaptive symbol value of the significance threshold of the v-th thyroid ultrasonic image comprises the following steps:
。
The invention also provides an ultrasonic image optimization processing system based on artificial intelligence, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the computer program stored in the memory to realize the steps of the ultrasonic image optimization processing method based on artificial intelligence.
The technical scheme of the invention has the beneficial effects that: according to the invention, the fluctuation characteristic value of each sequence segment is obtained according to the element value of each sequence segment, so that a basis is provided for screening abnormal sequence segments, and the enhancement effect of an ultrasonic image is improved; the real lesion area is obtained according to the abnormal updating sequence segment, the position of the real lesion area is accurately judged, and the ultrasonic image enhancement effect is further improved; according to the characteristic value of the lesion area of each thyroid ultrasonic image, the self-adaptive saliency threshold of each thyroid ultrasonic image is obtained, a more accurate detection basis is provided for saliency detection, and the effect of enhancing the ultrasonic image is further improved. The invention carries out significance detection through the accurate and reliable self-adaptive significance threshold value, and obtains the thyroid ultrasound enhanced image with better enhanced effect.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of steps of an artificial intelligence based ultrasound image optimization processing method of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following detailed description is given below of the ultrasonic image optimization processing method and system based on artificial intelligence according to the invention, which are specific implementation, structure, characteristics and effects thereof, with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
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.
The following specifically describes a specific scheme of the ultrasonic image optimization processing method and system based on artificial intelligence provided by the invention with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of steps of an artificial intelligence-based ultrasound image optimization processing method according to an embodiment of the present invention is shown, the method includes the steps of:
Step S001: acquiring a plurality of thyroid ultrasonic images; acquiring a plurality of gray value sequences in any thyroid ultrasonic image; and dividing each gray value sequence into a plurality of sequence segments according to the element value of each gray value sequence.
It is to be noted that an ultrasound image is a medical image generated using an ultrasound technique, which creates an image by transmitting high-frequency sound waves into human tissue and recording the reflected sound waves. Ultrasound waves propagate and reflect in different ways in different tissues, producing images with different brightness and textures. This example uses a thyroid ultrasound image as an example for analysis.
Specifically, performing thyroid ultrasonic detection on a plurality of patients by using an ultrasonic instrument to obtain a plurality of thyroid ultrasonic images; the present embodiment acquires corresponding thyroid ultrasound images for 100 patients.
It should be noted that, for different thyroid ultrasound images, since the size and extent of the lesion region generated are different, when the target region is acquired using the saliency detection algorithm CA, the saliency threshold corresponding to different regions is different. Therefore, firstly, according to the distribution characteristics of the lesion area in the thyroid ultrasound gray level image, the characteristic value of the lesion area in the image is calculated. In the thyroid ultrasonic gray scale image, the pathological areas such as cystic nodule and solid node are brighter than normal thyroid tissues due to the liquid. Thus, when the saliency detection CA method is used for detection, the target region is a region of higher gray value in the image. Due to the difference in lesion level, when the significance detection CA is performed to identify the target area, the corresponding identification range should be different. Therefore, the thyroid ultrasonic image needs to be initially analyzed, and the characteristic value of the lesion area in the thyroid ultrasonic image is calculated.
Specifically, in any thyroid ultrasound image, gray values of pixel points of each row are arranged according to a position sequence to obtain a gray value sequence of each row; for any gray value sequence, calculating the difference value of each element value and the left element value and the right element value of the element value respectively in the interval from the second element value to the last element value to obtain the left difference value and the right difference value of each element value, wherein the difference value calculation is that the element value is the subtracted number, and the element value on the left side or the right side is the subtracted number; the element value with the product of the left difference value and the right difference value being greater than 0 is recorded as a sequence extremum; and marking a sequence interval between every two sequence extreme values as a sequence segment to obtain a plurality of sequence segments of the gray value sequence. According to the method, a plurality of sequence segments in each gray value sequence are obtained.
Step S002: obtaining a fluctuation characteristic value of each sequence segment according to the element value of each sequence segment; screening abnormal sequence segments from all the sequence segments according to the fluctuation characteristic value of each sequence segment; and obtaining an abnormal updating sequence segment according to the abnormal sequence segment.
Since the normal tissue in the thyroid gray image shows a low gray value, the gray values of the pixels in the tissue are similar and distributed uniformly, and the lesion tissue and the noise show a high gray value. The gray value of the pixel points in the lesion area is higher than that in the normal tissue due to the existence of the lesion, and the gray value distribution of the pixel points in the lesion area is uneven. For any gray value sequence, the gray value of the pixel point of the normal tissue area in the sequence is smaller and the fluctuation is smaller; the fluctuation of the pixel points of the lesion tissue area should be larger and the gray value is higher than that of the normal tissue area; other noise areas and highlight areas are unified as non-target areas.
Specifically, in any one gray value sequence, for any one sequence segment, the largest two sequence extremum and the smallest two sequence extremum in the intervals of the sequence segment and the two adjacent sequence segments are obtained, the absolute value of the difference value of the largest two sequence extremum is recorded as the maximum value distance of the sequence segment, the absolute value of the difference value of the smallest two sequence extremum is recorded as the minimum value distance of the sequence segment, and the calculation mode of the fluctuation characteristic value of the j-th sequence segment in the gray value sequence is as follows:
；
In the method, in the process of the invention, A fluctuation feature value representing a j-th sequence segment; /(I)Representing the average of all gray values in the kth sequence segment; /(I)Representing the average of all gray values in the k+1th sequence segment; /(I)Representing the minimum distance of the j-th sequence segment; representing the maximum distance of the j-th sequence segment; /(I) Representing the average value of all gray values in the gray value sequence of the j-th sequence segment; /(I)Representing an absolute value function; /(I)Representing a linear normalization function, the normalization object is/> of all sequence segments in the gray value sequence; In particular, if there is no sequence segment adjacent to the left or adjacent to the right, the calculation is performed with the actually existing sequence segment.
It is to be noted that,The smaller the value of (c) represents the closer the overall gray values of the kth sequence segment and the (k+1) th sequence segment are. /(I)The smaller the value of (c) is, the smaller the difference degree of the sequence extremum near the jth sequence segment is, and the more regular the fluctuation of the corresponding three sequence segments is. /(I)The smaller the value of (c) is, the more stable the fluctuation of the element value near the jth sequence segment is, and the greater the possibility that the pixel point corresponding to the element value near the jth sequence segment is a normal tissue region.The smaller the value of (c) is, the closer the pixel point corresponding to the element value near the jth sequence segment is to the normal tissue region, and the smaller the fluctuation characteristic value of the jth sequence segment is.
Further, obtaining a fluctuation characteristic value of each sequence segment according to the method; in any gray value sequence, arranging the fluctuation characteristic values according to the sequence of sequence segments to obtain a fluctuation characteristic value sequence; based on a plurality of element values in the fluctuation characteristic value sequence, obtaining abnormal element values in the fluctuation characteristic value sequence by using an LOF abnormal detection algorithm, and marking a sequence segment corresponding to the abnormal element values as an abnormal sequence segment. The LOF anomaly detection algorithm is a well known technique, and specific methods are not described herein. The Chinese language of the LOF anomaly detection algorithm is called a local anomaly factor detection algorithm, and the English language is called Local Outlier Factor.
Step S003: and obtaining a real lesion area according to the abnormal updating sequence segment.
It should be noted that, since the fluctuation feature values of the normal tissue region and the background region are relatively small, all the pixel points corresponding to the calculated abnormal sequence segment belong to the suspected lesion region.
Specifically, in any one gray value sequence, a plurality of continuous abnormal sequence segments are combined into a new abnormal sequence segment, and the new abnormal sequence segment and the abnormal sequence segment which is not combined are recorded as abnormal update sequence segments. And acquiring the sequence values of the first element value and the last element value of each abnormal updating sequence segment in the gray value sequence respectively, and recording the sequence values as a starting sequence value and an ending sequence value of the abnormal updating sequence segment respectively.
It should be noted that, since the lesion area of thyroid tissue is generally circular, elliptical or variable irregular shape, noise has no obvious shape feature and the distribution is relatively dispersed. Therefore, the distribution positions of the suspected lesion areas on the plurality of fluctuation curve segments can be analyzed to further limit the lesion areas, and the influence of noise points on the judgment of the lesion areas is avoided.
Specifically, the method for obtaining the real lesion area according to the following method comprises the following steps:
(1) Recording the first gray value sequence as a target sequence;
(2) Recording the first abnormal updating sequence segment which does not participate in the correlation calculation on the target sequence as a target sequence segment, and performing the step (3); if no abnormal updating sequence segment which does not participate in the correlation calculation exists on the target sequence, marking the next gray value sequence of the target sequence as a new target sequence, repeating the step (2) until the target sequence is the last gray value sequence, and stopping repeating;
(3) Sequentially calculating the correlation degree of all the target sequence segments and each abnormal updating sequence segment in the next gray value sequence of the last target sequence segment until the correlation degree is greater than a preset correlation threshold value When the correlation degree is stopped being calculated, the correlation degree between the target sequence segments is larger than a preset correlation threshold/>, and the target sequence segments are correlated with the target sequence segmentsTaking the abnormal updating sequence section of the (c) as a target sequence section, repeating the step (3) until no correlation degree with all target sequence sections in the next gray value sequence is larger than a preset correlation threshold/>Stopping repeating the step (3) when the sequence segment is abnormally updated, and performing the step (4);
(4) And (3) marking the pixel point set corresponding to all the target sequence segments as a real lesion area, and repeating the step (2).
Preset correlation thresholdIn the description of this example, other values may be set in other embodiments, and the present example is not limited thereto.
The specific method for calculating the correlation degree comprises the following steps: and marking the next gray value sequence of the last target sequence segment as a reference sequence, sequencing the target sequence segment and the abnormal updating sequence segment in any one abnormal updating sequence segment in all the target sequence segments and the reference sequence, and collectively calculating the absolute value of the difference value of the starting sequence value and the absolute value of the difference value of the ending sequence value of each two adjacent sequence segments by the target sequence segment and the abnormal updating sequence segment, marking the absolute value of the maximum difference value of the starting sequence value as an initial difference value, and marking the absolute value of the maximum difference value of the ending sequence value as an ending difference value. When ordering any one of the abnormal update sequence segments in the target sequence segment and the corresponding reference sequence, labeling the target sequence segment from 1 according to the initial sequence, and taking any one of the abnormal update sequence segments in the reference sequence corresponding to the target sequence segment as the last sequence segment. For example, when the target sequence segment is e1, e2, e3, e4 and any one of the abnormal update sequence segments in the reference sequence corresponding to the target sequence segment is f1, the corresponding ordered sequence is: e1, e2, e3, e4, f1. And e1, e2, e3, e4, f1 are collectively referred to as sequence segments. The sequence value difference absolute value and the end sequence value difference absolute value of each two adjacent sequence segments are calculated, namely, the sequence value difference absolute values and the end sequence value difference absolute values of e1 and e2, e2 and e3, e3 and e4, e4 and f1 are calculated respectively. The specific calculation method of the correlation degree between all the target sequence segments and the xth abnormal updating sequence segment in the reference sequence comprises the following steps:
；
In the method, in the process of the invention, Representing the correlation of all target sequence segments with the x-th abnormally updated sequence segment in the reference sequence,Representing initial difference values of the xth abnormal updating sequence segment in all the target sequence segments and the reference sequence; and the ending difference value of the xth abnormal updating sequence segment in the target sequence segment and the reference sequence is represented.
According to the method, a plurality of real lesion areas are obtained.
Step S004: obtaining a lesion area characteristic value of each thyroid ultrasonic image according to the real lesion area in each thyroid ultrasonic image; and obtaining the self-adaptive significance threshold value of each thyroid ultrasonic image according to the characteristic value of the lesion area of each thyroid ultrasonic image.
Specifically, the calculation method of the characteristic value of the lesion area of the v-th thyroid ultrasonic image comprises the following steps:
；
In the method, in the process of the invention, Representing the characteristic value of a lesion area of a v-th thyroid ultrasonic image; /(I)Representing the number of real lesion areas in the v-th thyroid ultrasound image; /(I)The number of pixel points of the y-th real lesion area in the v-th thyroid ultrasonic image is represented; /(I)And representing the number of all pixel points in the v-th thyroid ultrasonic image.
According to the method, the characteristic value of the lesion area of each thyroid ultrasonic image is obtained.
Further, according to the feature value of the lesion area of each thyroid ultrasonic image, the adaptive saliency threshold value in the saliency detection algorithm CA is calculated, and the calculation mode of the adaptive saliency threshold value in the v-th thyroid ultrasonic image is as follows:
；
；
In the method, in the process of the invention, Representing an adaptive saliency threshold of the v-th thyroultrasonic image; /(I)Representing a preset empirical significance threshold,/>In the description of this example, other values may be set in other embodiments, and the present example is not limited thereto; /(I)An adaptive symbol value representing a saliency threshold of the v-th thyroultrasonic image; /(I)Representing the characteristic value of a lesion area of a v-th thyroid ultrasonic image; /(I)Representing the number of acquired thyroid ultrasound images; /(I)Representing a lesion area characteristic value of the r-th thyroid ultrasonic image; /(I)Representing an absolute value function.
It is to be noted that whenWhen the method is used, the area of the real lesion area on the v-th thyroid ultrasonic image is larger than that of other thyroid ultrasonic images, and the self-adaptive significance threshold of the v-th thyroid ultrasonic image is smaller, so that more significant areas are extracted; when/>And when the area of the real lesion area on the v-th thyroid ultrasonic image is smaller than that of other thyroid ultrasonic images, the adaptive significance threshold of the v-th thyroid ultrasonic image is larger, and the noise point is prevented from being extracted as a significant area.
According to the method, the adaptive significance threshold of each thyroid ultrasonic image is obtained.
Step S005: performing saliency detection on each thyroid ultrasonic image based on the adaptive saliency threshold to obtain a salient region and a non-salient region; and respectively carrying out enhancement treatment on the salient region and the non-salient region to obtain a thyroid ultrasound enhanced image.
Further, a CA algorithm for performing saliency detection on each thyroid ultrasonic image is utilized to obtain a salient region and a non-salient region of each thyroid ultrasonic image. In each thyroid ultrasound image, the significant region is subjected to enhancement processing on the image by using a histogram equalization method, and the non-significant region is subjected to filtering processing, and in the embodiment, the Gaussian filtering algorithm is adopted for filtering processing, so that the thyroid ultrasound enhanced image is obtained. The CA algorithm, histogram equalization, and gaussian filtering are known techniques, and specific methods are not described herein. The Chinese language of the CA algorithm is called an image Context awareness algorithm, and the English language is called Context-Aware.
Thus, the thyroid ultrasound enhanced image is obtained, and the optimization processing of the ultrasound image is realized.
This embodiment is completed.
The invention also provides an ultrasonic image optimization processing system based on artificial intelligence, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the computer program stored in the memory to realize the steps of the ultrasonic image optimization processing method based on artificial intelligence.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the invention, but any modifications, equivalent substitutions, improvements, etc. within the principles of the present invention should be included in the scope of the present invention.
Claims (10)
1. The ultrasonic image optimization processing method based on artificial intelligence is characterized by comprising the following steps of:
acquiring a plurality of thyroid ultrasonic images; acquiring a plurality of gray value sequences in any thyroid ultrasonic image; dividing each gray value sequence into a plurality of sequence segments according to the element value of each gray value sequence;
Obtaining a fluctuation characteristic value of each sequence segment according to the element value of each sequence segment; screening abnormal sequence segments from all the sequence segments according to the fluctuation characteristic value of each sequence segment; obtaining an abnormal updating sequence segment according to the abnormal sequence segment;
obtaining a real lesion area according to the abnormal updating sequence segment;
Obtaining a lesion area characteristic value of each thyroid ultrasonic image according to the real lesion area in each thyroid ultrasonic image; obtaining an adaptive significance threshold of each thyroid ultrasonic image according to the characteristic value of the lesion area of each thyroid ultrasonic image;
performing saliency detection on each thyroid ultrasonic image based on the adaptive saliency threshold to obtain a salient region and a non-salient region; and respectively carrying out enhancement treatment on the salient region and the non-salient region to obtain a thyroid ultrasound enhanced image.
2. The ultrasonic image optimization processing method based on artificial intelligence according to claim 1, wherein the dividing each gray value sequence into a plurality of sequence segments according to the element value of each gray value sequence comprises the following specific steps:
For any gray value sequence, calculating the difference value of each element value and the left element value and the right element value of the element value respectively in the interval from the second element value to the last element value to obtain the left difference value and the right difference value of each element value, and recording the element value with the product of the left difference value and the right difference value being greater than 0 as a sequence extremum; and marking a sequence interval between every two sequence extreme values as a sequence segment to obtain a plurality of sequence segments of the gray value sequence.
3. The ultrasonic image optimization processing method based on artificial intelligence according to claim 1, wherein the step of obtaining the fluctuation feature value of each sequence segment according to the element value of each sequence segment comprises the following specific steps:
In any one gray value sequence, for any one sequence segment, acquiring the largest two sequence extremum and the smallest two sequence extremum in the intervals of the sequence segment and the two adjacent sequence segments, marking the absolute value of the difference value of the largest two sequence extremum as the maximum value distance of the sequence segment, marking the absolute value of the difference value of the smallest two sequence extremum as the minimum value distance of the sequence segment, and calculating the fluctuation characteristic value of the j-th sequence segment in the gray value sequence by the following steps:
；
In the method, in the process of the invention, A fluctuation feature value representing a j-th sequence segment; /(I)Representing the average of all gray values in the kth sequence segment; representing the average of all gray values in the k+1th sequence segment; /(I) Representing the minimum distance of the j-th sequence segment; /(I)Representing the maximum distance of the j-th sequence segment; /(I)Representing the average value of all gray values in the gray value sequence of the j-th sequence segment; /(I)Representing an absolute value function; /(I)Representing the normalization function.
4. The ultrasonic image optimization processing method based on artificial intelligence according to claim 1, wherein the step of screening abnormal sequence segments from all sequence segments according to the fluctuation characteristic value of each sequence segment comprises the following specific steps:
In any gray value sequence, arranging the fluctuation characteristic values according to the sequence of sequence segments to obtain a fluctuation characteristic value sequence; based on a plurality of element values in the fluctuation characteristic value sequence, obtaining abnormal element values in the fluctuation characteristic value sequence by utilizing a local abnormal factor detection algorithm, and marking a sequence segment corresponding to the abnormal element values as an abnormal sequence segment.
5. The ultrasonic image optimization processing method based on artificial intelligence according to claim 1, wherein the obtaining the abnormal update sequence segment according to the abnormal sequence segment comprises the following specific steps:
in any gray value sequence, a plurality of continuous abnormal sequence segments are combined into a new abnormal sequence segment, and the new abnormal sequence segment and the abnormal sequence segment which is not combined are recorded as abnormal updating sequence segments.
6. The ultrasonic image optimization processing method based on artificial intelligence according to claim 1, wherein the obtaining the real lesion area according to the abnormal update sequence segment comprises the following specific steps:
(1) Recording the first gray value sequence as a target sequence;
(2) Recording the first abnormal updating sequence segment which does not participate in the correlation calculation on the target sequence as a target sequence segment, and performing the step (3); if no abnormal updating sequence segment which does not participate in the correlation calculation exists on the target sequence, marking the next gray value sequence of the target sequence as a new target sequence, repeating the step (2) until the target sequence is the last gray value sequence, and stopping repeating;
(3) Sequentially calculating the correlation degree of all the target sequence segments and each abnormal updating sequence segment in the next gray value sequence of the last target sequence segment until the correlation degree is greater than a preset correlation threshold value When the correlation degree is stopped being calculated, the correlation degree between the target sequence segments is larger than a preset correlation threshold/>, and the target sequence segments are correlated with the target sequence segmentsTaking the abnormal updating sequence section of the (c) as a target sequence section, repeating the step (3) until no correlation degree with all target sequence sections in the next gray value sequence is larger than a preset correlation threshold/>Stopping repeating the step (3) when the sequence segment is abnormally updated, and performing the step (4);
(4) And (3) marking the pixel point set corresponding to all the target sequence segments as a real lesion area, and repeating the step (2).
7. The ultrasonic image optimization processing method based on artificial intelligence according to claim 6, wherein the step of sequentially calculating the correlation between all the target sequence segments and each abnormal update sequence segment in the sequence of gray values next to the sequence of gray values in which the last target sequence segment is located comprises the following specific steps:
Acquiring sequence values of a first element value and a last element value of each abnormal updating sequence segment in a gray value sequence respectively, and recording the sequence values as a starting sequence value and an ending sequence value of the abnormal updating sequence segment respectively;
Marking the next gray value sequence of the last target sequence segment as a reference sequence, sequencing the target sequence segment and the abnormal updating sequence segment in any one abnormal updating sequence segment in all the target sequence segments and the reference sequence, and collectively grouping the target sequence segment and the abnormal updating sequence segment as sequence segments, calculating the absolute value of the difference value of the starting sequence values and the absolute value of the difference value of the ending sequence values of every two adjacent sequence segments, marking the absolute value of the maximum difference value of the starting sequence values as an initial difference value, and marking the absolute value of the maximum difference value of the ending sequence values as an ending difference value; the specific calculation method of the correlation degree between all the target sequence segments and the xth abnormal updating sequence segment in the reference sequence comprises the following steps:
；
In the method, in the process of the invention, Representing the correlation of all target sequence segments with the x-th abnormally updated sequence segment in the reference sequence,Representing initial difference values of the xth abnormal updating sequence segment in all the target sequence segments and the reference sequence; and the ending difference value of the xth abnormal updating sequence segment in the target sequence segment and the reference sequence is represented.
8. The ultrasonic image optimizing processing method based on artificial intelligence according to claim 1, wherein the obtaining the characteristic value of the lesion area of each thyroid ultrasonic image according to the real lesion area in each thyroid ultrasonic image comprises the following specific calculation method:
；
In the method, in the process of the invention, Representing the characteristic value of a lesion area of a v-th thyroid ultrasonic image; /(I)Representing the number of real lesion areas in the v-th thyroid ultrasound image; /(I)The number of pixel points of the y-th real lesion area in the v-th thyroid ultrasonic image is represented; /(I)And representing the number of all pixel points in the v-th thyroid ultrasonic image.
9. The ultrasonic image optimization processing method based on artificial intelligence according to claim 1, wherein the obtaining the adaptive saliency threshold of each thyroid ultrasonic image according to the feature value of the lesion area of each thyroid ultrasonic image comprises the following specific calculation modes:
；
In the method, in the process of the invention, Representing an adaptive saliency threshold of the v-th thyroultrasonic image; /(I)Representing a preset experience significance threshold; /(I)An adaptive symbol value representing a saliency threshold of the v-th thyroultrasonic image; /(I)Representing the characteristic value of a lesion area of a v-th thyroid ultrasonic image; /(I)Representing the number of acquired thyroid ultrasound images; /(I)Representing a lesion area characteristic value of the r-th thyroid ultrasonic image; /(I)Representing an absolute value function;
The method for acquiring the self-adaptive symbol value of the significance threshold of the v-th thyroid ultrasonic image comprises the following steps:
。
10. An artificial intelligence based ultrasound image optimization processing system comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the computer program when executed by the processor implements the steps of the artificial intelligence based ultrasound image optimization processing method of any of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410523910.1A CN118115415B (en) | 2024-04-29 | Ultrasonic image optimization processing method and system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410523910.1A CN118115415B (en) | 2024-04-29 | Ultrasonic image optimization processing method and system based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118115415A true CN118115415A (en) | 2024-05-31 |
| CN118115415B CN118115415B (en) | 2024-07-05 |
Family
ID=
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103077550A (en) * | 2012-12-10 | 2013-05-01 | 华北电力大学（保定） | Vascular four-dimensional reconstruction method in NOT gate-controlled ICUS (intravascular ultrasound) image sequence |
| CN106780448A (en) * | 2016-12-05 | 2017-05-31 | 清华大学 | A kind of pernicious sorting technique of ultrasonic Benign Thyroid Nodules based on transfer learning Yu Fusion Features |
| CN107093184A (en) * | 2017-04-11 | 2017-08-25 | 湖北理工学院 | A kind of ultrasonic image sequence dividing method based on sparse features and shape correlation |
| WO2021169452A1 (en) * | 2020-09-27 | 2021-09-02 | 平安科技（深圳）有限公司 | Thyroid nodule image processing method and apparatus, and computer device |
| CN114926433A (en) * | 2022-05-19 | 2022-08-19 | 常州市第二人民医院 | Ultrasonic classification method based on ultrasonic dynamic sequence image characteristics |
| CN117078671A (en) * | 2023-10-13 | 2023-11-17 | 陕西秒康医疗科技有限公司 | Thyroid ultrasonic image intelligent analysis system |
| CN117392155A (en) * | 2023-12-11 | 2024-01-12 | 吉林大学 | High-throughput gene sequencing data processing method based on image processing |
| CN117649357A (en) * | 2024-01-29 | 2024-03-05 | 深圳市恩普电子技术有限公司 | Ultrasonic image processing method based on image enhancement |
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103077550A (en) * | 2012-12-10 | 2013-05-01 | 华北电力大学（保定） | Vascular four-dimensional reconstruction method in NOT gate-controlled ICUS (intravascular ultrasound) image sequence |
| CN106780448A (en) * | 2016-12-05 | 2017-05-31 | 清华大学 | A kind of pernicious sorting technique of ultrasonic Benign Thyroid Nodules based on transfer learning Yu Fusion Features |
| CN107093184A (en) * | 2017-04-11 | 2017-08-25 | 湖北理工学院 | A kind of ultrasonic image sequence dividing method based on sparse features and shape correlation |
| WO2021169452A1 (en) * | 2020-09-27 | 2021-09-02 | 平安科技（深圳）有限公司 | Thyroid nodule image processing method and apparatus, and computer device |
| CN114926433A (en) * | 2022-05-19 | 2022-08-19 | 常州市第二人民医院 | Ultrasonic classification method based on ultrasonic dynamic sequence image characteristics |
| CN117078671A (en) * | 2023-10-13 | 2023-11-17 | 陕西秒康医疗科技有限公司 | Thyroid ultrasonic image intelligent analysis system |
| CN117392155A (en) * | 2023-12-11 | 2024-01-12 | 吉林大学 | High-throughput gene sequencing data processing method based on image processing |
| CN117649357A (en) * | 2024-01-29 | 2024-03-05 | 深圳市恩普电子技术有限公司 | Ultrasonic image processing method based on image enhancement |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116152505B (en) | Bone target identification and segmentation method based on X-ray data | |
| CN117078671B (en) | Thyroid ultrasonic image intelligent analysis system | |
| CN111383259B (en) | Image analysis method, computer device, and storage medium | |
| CN110009656B (en) | Target object determination method and device, storage medium and electronic device | |
| CN111260701B (en) | Multi-mode retina fundus image registration method and device | |
| CN110880177A (en) | Image identification method and device | |
| CN114332132A (en) | Image segmentation method and device and computer equipment | |
| CN109544528B (en) | Lung nodule image identification method and device | |
| WO2020168647A1 (en) | Image recognition method and related device | |
| CN109241898B (en) | Method and system for positioning target of endoscopic video and storage medium | |
| CN111815610B (en) | Lesion detection method and device for lesion image | |
| CN111166362A (en) | Medical image display method and device, storage medium and electronic equipment | |
| CN114677525B (en) | Edge detection method based on binary image processing | |
| CN112396565A (en) | Method and system for enhancing and segmenting blood vessels of images and videos of venipuncture robot | |
| CN113327225A (en) | Method for providing airway information | |
| CN112529918B (en) | Method, device and equipment for segmenting brain room area in brain CT image | |
| Shelke et al. | Automated segmentation and detection of brain tumor from MRI | |
| CN118115415B (en) | Ultrasonic image optimization processing method and system based on artificial intelligence | |
| CN118115415A (en) | Ultrasonic image optimization processing method and system based on artificial intelligence | |
| CN116485819A (en) | Ear-nose-throat examination image segmentation method and system | |
| CN113192099B (en) | Tissue extraction method, device, equipment and medium | |
| Kandeel et al. | A study of a modified histogram based fast enhancement algorithm (MHBFE) | |
| CN114387257A (en) | Segmentation method, system, device and medium for lung lobe region in lung image | |
| CN117974692B (en) | Ophthalmic medical image processing method based on region growing | |
| Obayya et al. | Lung cancer recognition using radon transform and adaptive neuro fuzzy inference system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |