Automatic exposure method for near-infrared first-area fluorescence imaging

Abstract

The invention provides an automatic exposure method for near-infrared first-area fluorescence imaging, which comprises the following steps of: imaging an indocyanine green sample which excites near-infrared first-area rays by using different exposure levels to obtain a series of fluorescence image data under different exposure conditions, each exposure level is determined by a specific gain and shutter speed combination; establishing a mapping relation between the gray scale and the exposure level of the fluorescence image; predicting the performance of the current fluorescence image under any exposure level based on the established mapping relation; and according to the prediction result, selecting the exposure level capable of generating the optimal image quality under the current condition as the optimal exposure level. According to the method, the fluorescence imaging automatic exposure method is specifically constructed, the change of the exposure environment can be responded in real time in the process of near-infrared first-area fluorescence molecular image surgical navigation, and correct fluorescence imaging exposure is carried out.

Description

Technical Field

[0001] The invention relates to the technical field of medical imaging and image processing, and in particular to an automatic exposure method for near-infrared first-zone fluorescence imaging. Background Art

[0002] Dynamic imaging of the lesion area during surgery can provide surgeons with precise surgical guidance. Traditional imaging detection technologies (such as CT, MRI, PET, SPECT, etc.) are difficult to provide real-time and clear lesion localization during surgery. With the continuous improvement of camera imaging technology and production technology, color visible light cameras can already meet the requirements of doctors for resolution, imaging quality and frame rate in intraoperative scenarios, but visible light cameras alone cannot collect spectral information outside the visible light band. Therefore, in the case of only visible light cameras, doctors can only locate lesions by naked eye observation and empirical judgment. Under the irradiation of the excitation light source, if targeted fluorescent probes (such as indocyanine green) are used to mark the diseased cells, near-infrared emission light in a specific band can be generated. Then, the near-infrared camera can be used to collect and image the excited near-infrared emission light, presenting details that are difficult to clearly distinguish with visible light cameras (such as polyps, tumors and even blood vessels). This imaging method is near-infrared fluorescent molecular imaging. Based on near-infrared fluorescent molecular imaging, tumor lesions can be marked during surgery to assist surgeons in resection, which can achieve precision medicine. Near-infrared fluorescence molecular imaging has the advantages of high resolution, no ionizing radiation, and real-time imaging. It is suitable for highlighting the diseased area marked by fluorescent probes in real time during surgery. Using a near-infrared zone 1 camera alone is also unable to provide the state of human organs under visible light (including color, shape, surface texture, etc.). If a visible light camera and a near-infrared camera are used for multiple spectral synchronous imaging at the same time, it is possible to capture the natural morphology of organs under visible light irradiation, and accurately locate the pathological area marked with fluorescence, thereby presenting rich pathological details.

[0003] Therefore, fast and accurate automatic exposure of near-infrared zone 1 fluorescence imaging can help doctors make better judgments. However, current autofocus methods are designed for natural images, and there are obvious differences between natural images and fluorescence images. So far, there is no autofocus method designed for fluorescence imaging. Summary of the invention

[0004] The present invention is directed to a near-infrared fluorescence imaging surgical navigation device and provides an automatic exposure method for near-infrared zone 1 fluorescence imaging. The method can infer the optimal exposure setting in the current environment based on the current fluorescence image and perform correct fluorescence imaging exposure.

[0005] The present invention achieves the above-mentioned purpose through the following technical solutions:

[0006] An automatic exposure method for near-infrared first-region fluorescence imaging comprises the following steps:

[0007] Using different exposure levels to image an indocyanine green sample that excites near-infrared radiation, a series of fluorescence image data under different exposure conditions are obtained, wherein each exposure level is determined by a specific gain and shutter speed combination;

[0008] Establishing a mapping relationship between grayscale and exposure level of fluorescent image;

[0009] Based on the established mapping relationship, the performance of the current fluorescent image at any exposure level is predicted;

[0010] According to the prediction result, the exposure level that can produce the best image quality under the current conditions is selected as the optimal exposure level.

[0011] According to an automatic exposure method for near-infrared region 1 fluorescence imaging provided by the present invention, the mapping relationship between grayscale and fluorescence image exposure level is established, including:

[0012] In this series of fluorescence image data, the area with grayscale values ​​uniformly distributed between [0, 255] is selected as the fixed area, and all pixel values ​​in the fixed area are intercepted for each photo as the basic data points for constructing the mathematical model.

[0013] According to an automatic exposure method for near-infrared zone 1 fluorescence imaging provided by the present invention, a mapping relationship between grayscale and exposure level is fitted using the least squares method based on an objective function, which is expressed as formula (1):

[0014]

[0015] The response equation of fluorescence exposure and grayscale mapping is fitted and expressed as formula (2):

[0016] E ij =f(I ij ) (2)

[0017] The response equation of grayscale and fluorescence exposure mapping is fitted and expressed as formula (3):

[0018] I ij =g(E ij ) (3)

[0019] Where N represents the number of pixels in the selected area, P represents the number of exposure levels of the photo, I represents the grayscale, and E represents the exposure level; I ij Indicates the grayscale value of the pixel at position (i, j), E ij Represents the exposure level at position (i, j).

[0020] According to an automatic exposure method for near-infrared first-zone fluorescence imaging provided by the present invention, based on function f and function g, the fluorescence image at the current exposure level is used to predict its condition at a specific exposure level, which is expressed as formula (4):

[0021]

[0022] Predict what the fluorescent photo will look like at any exposure level.

[0023] According to an automatic exposure method for near-infrared first-zone fluorescence imaging provided by the present invention, the grayscale distribution histogram of the fluorescence image is corrected:

[0024] Only those that meet [I min , I max ] range of pixels, where I min and I max is a number between 1 and 254, I min Less than I max .

[0025] According to an automatic exposure method for near-infrared fluorescence imaging provided by the present invention, the optimal exposure level is determined by calculating the mean loss, which is expressed as formula (5):

[0026] L=|I ref -I mean | (5)

[0027] Among them I ref is the reference mean of the grayscale value distribution of the designed optimal exposure, I mean It is the corrected mean value of the gray value distribution of the current exposure.

[0028] According to an automatic exposure method for near-infrared fluorescence imaging provided by the present invention, for predicting the mean loss L of all exposure levels all ={L0, L1, L2, ...}, select the exposure level with the lowest mean loss as the output level for the next exposure.

[0029] According to an automatic exposure method for near-infrared zone 1 fluorescence imaging provided by the present invention, the set A of gain and shutter speed combinations = {gain 0 and shutter 1 / 4000, gain 0 and shutter 1 / 2000, gain 0 and shutter 1 / 1000, gain 0 and shutter 1 / 500, gain 0 and shutter 1 / 250, gain 0 and shutter 1 / 125, gain 0 and shutter 1 / 100, gain 3 and shutter 1 / 100, gain 6 and shutter 1 / 100, gain 9 and shutter 1 / 100, gain 12 and shutter 1 / 100, gain 15 and shutter 1 / 100, gain 18 and shutter 1 / 100, gain 21 and shutter 1 / 100, gain 24 and shutter 1 / 100}.

[0030] According to an automatic exposure method for near-infrared zone 1 fluorescence imaging provided by the present invention, the exposure levels are set to Ae={0.00025, 0.0005, 0.001, 0.002, 0.004, 0.008, 0.01, 0.012, 0.014, 0.016, 0.018, 0.020, 0.022, 0.024, 0.026}.

[0031] According to an automatic exposure method for near-infrared zone 1 fluorescence imaging provided by the present invention, a feedback system is established to fine-tune the exposure level according to the imaging results, and the adjustment results are incorporated into the training data of the machine learning model.

[0032] It can be seen that compared with the prior art, the method provided by the present invention has the following beneficial effects:

[0033] 1. The present invention can respond in real time to changes in the exposure environment during near-infrared region I fluorescence molecular imaging surgical navigation and perform correct fluorescence imaging exposure.

[0034] 2. The present invention specifically constructs an automatic exposure method for fluorescence imaging, which can quickly reach the optimal exposure setting when there are changes in the exposure environment during surgery, thereby performing better surgery and benefiting patients.

[0035] 3. The present invention can ensure the best image quality for each imaging by predicting the performance of fluorescent images at different exposure levels and automatically selecting the optimal exposure level, avoiding problems such as image blur, loss of details or increased noise caused by underexposure or overexposure. The automatic exposure method provided by the present invention reduces the interference of human factors and makes the imaging results more stable and consistent.

[0036] 4. The automatic exposure method provided by the present invention can quickly adjust the exposure parameters and complete the imaging, which greatly shortens the imaging time. The user does not need to manually adjust the exposure level. The user only needs to place the sample in the imaging system, and the system can automatically complete the setting of the exposure parameters and the imaging process, which greatly simplifies the operation process.

[0037] 5. This method is applicable to ICG samples of different thicknesses and properties, as well as different imaging environments. By adjusting exposure parameters and imaging strategies, it can meet a variety of imaging requirements.

[0038] 6. The present invention is based on the established mapping relationship between grayscale and fluorescent image exposure level. This method can intelligently predict image performance under different exposure levels and provide a scientific basis for the selection of exposure parameters.

[0039] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 The present invention is a flowchart of an automatic exposure method for near-infrared region 1 fluorescence imaging.

[0041] Figure 2 It is a schematic diagram of collecting fluorescence data at different exposure levels in an automatic exposure method embodiment of the present invention for near-infrared zone 1 fluorescence imaging for fitting a curve about grayscale and exposure level.

[0042] Figure 3 It is a first curve schematic diagram of exposure level and gray scale in an embodiment of an automatic exposure method for near-infrared region 1 fluorescence imaging of the present invention.

[0043] Figure 4 It is a second curve schematic diagram of exposure level and gray scale in an embodiment of an automatic exposure method for near-infrared region 1 fluorescence imaging of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0046] See also Figures 1 to 4 The present invention provides an automatic exposure method for near-infrared region 1 fluorescence imaging, the method comprising the following steps:

[0047] Step S1, using different exposure levels to image an indocyanine green sample that excites near-infrared radiation, to obtain a series of fluorescence image data under different exposure conditions, wherein each exposure level is determined by a specific gain and shutter speed combination.

[0048] Step S2, establishing a mapping relationship between grayscale and fluorescent image exposure level.

[0049] Step S3, predicting the performance of the current fluorescent image at any exposure level based on the established mapping relationship.

[0050] Step S4: According to the prediction result, the exposure level that can produce the best image quality under the current conditions is selected as the optimal exposure level.

[0051] In the above step S1, first, the CMOS sensor for near-infrared region 1 fluorescence imaging needs to be calibrated, that is, in a fixed scene, all exposure levels are used to collect data for the indocyanine green sample that excites near-infrared region 1 rays, and a series of fluorescence photos collected at different exposure levels can be obtained.

[0052] In the above step S2, in this series of fluorescence image data, an area with grayscale values ​​uniformly distributed between [0, 255] is selected as a fixed area, and all pixel values ​​in the fixed area are intercepted for each photo as basic data points for constructing a mathematical model.

[0053] In this embodiment, the mapping relationship between grayscale and exposure level is fitted using the least square method based on the objective function, which is expressed as formula (1):

[0054]

[0055] The response equation of fluorescence exposure and grayscale mapping is fitted and expressed as formula (2):

[0056] E ij =f(I ij ) (2)

[0057] The response equation of grayscale and fluorescence exposure mapping is fitted and expressed as formula (3):

[0058] I ij =g(E ij ) (3)

[0059] Where N represents the number of pixels in the selected area, P represents the number of exposure levels of the photo, I represents the grayscale, and E represents the exposure level; I ij Indicates the grayscale value of the pixel at position (i, j), E ij Represents the exposure level at position (i, j).

[0060] In this embodiment, based on the function f and the function g, the fluorescent image at the current exposure level is used to predict the situation at a specific exposure level, which is expressed as formula (4):

[0061]

[0062] Similarly, the fluorescence photograph can be predicted at any exposure level.

[0063] For judging whether the exposure is reasonable, the best way in this embodiment is to use a histogram to judge. If the grayscale distribution histogram of the image is left-biased, it means that it is underexposed; if the grayscale distribution histogram of the image is right-biased, it means that it is overexposed; if the grayscale distribution histogram of the image is bell-shaped, it means that it is in an ideal exposure state. However, there is a significant difference between the grayscale histogram distribution of the fluorescent image and the grayscale distribution histogram of the natural image. The reason is that the background of the fluorescent image is completely black and there is no information, which causes the grayscale histogram distribution of the fluorescent image to be severely left-biased, while the background of the natural image is not completely black and there is still information. Therefore, it is necessary to make certain corrections to the grayscale distribution histogram statistics of the fluorescent image.

[0064] In this embodiment, the grayscale distribution histogram of the fluorescent image is corrected:

[0065] Only those that meet [I min , I max ] range of pixels, where I min and I max is a number between 1 and 254, I min Less than I max .

[0066] In this embodiment, the optimal exposure level is determined by calculating the mean loss, which is expressed as formula (5):

[0067] L=|I ref -I mean | (5)

[0068] Among them I ref is the reference mean of the grayscale value distribution of the designed optimal exposure, I mean It is the corrected mean value of the gray value distribution of the current exposure.

[0069] In this embodiment, for predicting the mean loss L for all exposure levels, all ={L0, L1, L2, ...}, select the exposure level with the lowest mean loss as the output of the next exposure level, because its predicted grayscale distribution mean is closest to our preset ideal value.

[0070] In this embodiment, the set of gain and shutter speed combinations A={gain 0 and shutter 1 / 4000, gain 0 and shutter 1 / 2000, gain 0 and shutter 1 / 1000, gain 0 and shutter 1 / 500, gain 0 and shutter 1 / 250, gain 0 and shutter 1 / 125, gain 0 and shutter 1 / 100, gain 3 and shutter 1 / 100, gain 6 and shutter 1 / 100, gain 9 and shutter 1 / 100, gain 12 and shutter 1 / 100, gain 15 and shutter 1 / 100, gain 18 and shutter 1 / 100, gain 21 and shutter 1 / 100, gain 24 and shutter 1 / 100} collects fluorescence data at different exposure levels, such as Figure 2 shown.

[0071] In this embodiment, the exposure levels are set to Ae={0.00025, 0.0005, 0.001, 0.002, 0.004, 0.008, 0.01, 0.012, 0.014, 0.016, 0.018, 0.020, 0.022, 0.024, 0.026} respectively.

[0072] In practical applications, the size of the collected data is 1920*1080, and the area with uniform grayscale distribution is selected for each exposure level data. Figure 2 In the case of , the coordinates of the upper left corner and the lower right corner of the region are [500, 500] and [900, 1200] respectively, and the cropped pixel region is used for fitting of formula 1. After fitting, we can obtain the exposure level corresponding to each grayscale from 0 to 255. Furthermore, in this embodiment, the functions of formulas (2) and (3) can be obtained by polynomial fitting, that is, the function f is a mapping from grayscale to exposure level, as shown in Figure 3 As shown, the g function is the mapping from exposure level to grayscale, such as Figure 4 shown.

[0073] Assume that the current exposure level is 0.022, that is, the gain shutter combination is 18 and 1 / 100, and it is necessary to predict the optimal exposure level under the current situation. To predict the grayscale of the pixel at different exposure levels, the grayscale under different exposure registrations can be calculated using formula 4. Note the predicted grayscale distribution at this exposure level as I 曝光等级 By analogy, all exposure levels are traversed to obtain the predicted grayscale distribution and calculate its mean. Then, a set of predicted grayscale distribution means at different exposure levels can be obtained, which is denoted as B = {I 0.00025 , I 0.0005 , I 0.001 , I 0.002 , I 0.004 , I 0.008 , I 0.01 , I 0.012 , I 0.014 , I 0.016 , I 0.018 , I 0.020 , I 0.022 , I 0.024 , I 0.026}.

[0074] By using formula (5) and the preset fluorescence grayscale histogram distribution statistical correction range [10, 245], that is, I min and I max 10 and 245 respectively, and the absolute value difference between the predicted grayscale distribution mean and the optimal exposure grayscale distribution mean, i.e., the mean loss, can be obtained as C = {L 0.00025 , L 0.0005 , L 0.001 , L 0.002 , L 0.004 , L 0.008 , L 0.01 , L 0.012 , L 0.014 , L 0.016 , L 0.018 , L 0.020 , L 0.022 , L 0.024 , L 0.026 The exposure level with the smallest L value in the set is the recommended exposure level of this method under the exposure condition. 0.012 The minimum is the output gain exposure combination corresponding to 0.012, which is to predict that under the current exposure environment, the optimal gain is 3 and the shutter is 1 / 100.

[0075] In this embodiment, the following is also performed: establishing a feedback system, fine-tuning the exposure level according to the imaging results, and incorporating the adjustment results into the training data of the machine learning model.

[0076] Specifically, establishing an imaging and exposure level feedback mechanism includes: using a high-resolution camera or image sensor to capture images in real time, and ensuring that the imaging device has the function of automatically or manually adjusting the exposure level, such as aperture, shutter speed, and ISO. Using an image processing algorithm, the exposure level of the captured image is evaluated, and a standard range or threshold for the exposure level is set to determine whether the exposure of the current image is appropriate. When the exposure level of the image does not meet the preset standard, the system should automatically or prompt the user to make fine adjustments. Among them, fine-tuning can be achieved by adjusting the camera's exposure parameters (such as aperture, shutter speed, ISO), or by adjusting the brightness, contrast, etc. of the image through post-image processing software.

[0077] After fine-tuning the exposure level, collect the image data before and after the adjustment and its corresponding exposure parameters to ensure that the data is diverse and representative to cover different lighting conditions, shooting scenes, and exposure requirements.

[0078] The collected image data is annotated, including whether the exposure level is appropriate, the direction that needs to be adjusted (such as brighter, darker), and the specific amount of adjustment. The collected image data is preprocessed, including image cropping, scaling, normalization, and other operations to ensure data consistency and comparability. The exposure parameters and annotation information are converted into a format that the machine learning model can understand, such as a numerical feature vector or label.

[0079] Choose an appropriate machine learning model, such as a convolutional neural network (CNN), support vector machine (SVM), or random forest, to learn the relationship between image features and exposure levels.

[0080] Use the preprocessed data to train the model and continuously adjust the model's parameters and structure to improve its prediction accuracy and generalization ability. Use an independent test set to evaluate the trained model to check whether its prediction performance meets the requirements. Further optimize the model based on the evaluation results, such as adjusting model parameters, increasing training data, or using more advanced algorithms.

[0081] Integrate the feedback system and machine learning model into a unified platform or system to achieve automated and intelligent exposure adjustment, ensuring that the system can receive imaging data, perform exposure evaluation, perform fine-tuning operations and update the training data of the machine learning model in real time.

[0082] Through the above steps, an effective feedback system can be established to fine-tune the exposure level according to the imaging results, and incorporate the adjustment results into the training data of the machine learning model, thereby continuously improving the intelligence level of the system and the imaging quality.

[0083] In summary, the present invention can respond to changes in the exposure environment in real time and perform correct fluorescence imaging exposure during near-infrared zone 1 fluorescence molecular imaging surgical navigation. The present invention specifically constructs an automatic exposure method for fluorescence imaging, and can quickly reach the optimal exposure setting when there are changes in the exposure environment during surgery, thereby performing better surgery and benefiting patients.

[0084] Furthermore, the present invention can ensure that the best image quality can be obtained each time by predicting the performance of fluorescent images under different exposure levels and automatically selecting the optimal exposure level, avoiding problems such as image blur, loss of details or increased noise caused by underexposure or overexposure. The automatic exposure method provided by the present invention reduces the interference of human factors and makes the imaging results more stable and consistent.

[0085] Furthermore, the automatic exposure method provided by the present invention can quickly adjust exposure parameters and complete imaging, greatly shortening the imaging time. The user does not need to manually adjust the exposure level, but only needs to place the sample in the imaging system, and the system can automatically complete the setting of exposure parameters and the imaging process, greatly simplifying the operation process.

[0086] Furthermore, this method is applicable to ICG samples of different thicknesses and properties, as well as different imaging environments. By adjusting exposure parameters and imaging strategies, a variety of imaging requirements can be met.

[0087] Furthermore, based on the established mapping relationship between grayscale and fluorescent image exposure level, the present invention can intelligently predict image performance under different exposure levels, providing a scientific basis for the selection of exposure parameters.

[0088] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0089] The above-mentioned embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and substitutions made by technicians in this field on the basis of the present invention shall fall within the scope of protection required by the present invention.

Claims

1. An automatic exposure method for near-infrared region 1 fluorescence imaging, characterized in that: The following steps are involved: Using different exposure levels to image an indocyanine green sample that excites near-infrared radiation, a series of fluorescence image data under different exposure conditions are obtained, wherein each exposure level is determined by a specific gain and shutter speed combination; Establishing a mapping relationship between grayscale and exposure level of fluorescent image; Based on the established mapping relationship, the performance of the current fluorescent image at any exposure level is predicted; According to the prediction result, the exposure level that can produce the best image quality under the current conditions is selected as the optimal exposure level.

2. The method according to claim 1, characterized in that The step of establishing a mapping relationship between the grayscale and the exposure level of the fluorescent image includes: In this series of fluorescence image data, an area with grayscale values ​​evenly distributed between [0, 255] is selected as a fixed area, and all pixel values ​​in the fixed area are captured for each photo as basic data points for constructing a mathematical model.

3. The method according to claim 2, characterized in that: Based on the objective function, the least squares method is used to fit the mapping relationship between grayscale and exposure level, which is expressed as formula (1): The response equation of fluorescence exposure and grayscale mapping is fitted and expressed as formula (2): HAVE BEEN ij =f(I ij ) (2) The response equation of grayscale and fluorescence exposure mapping is fitted and expressed as formula (3): I ij =g(E ij ) (3) Where N represents the number of pixels in the selected area, P represents the number of exposure levels of the photo, I represents the grayscale, and E represents the exposure level; I ij Indicates the grayscale value of the pixel at position (i, j), E ij Represents the exposure level at position (i, j).

4. The method according to claim 3, characterized in that: Based on the function f and the function g, the fluorescence image at the current exposure level is used to predict its condition at a specific exposure level, which is expressed as formula (4): Predict what the fluorescence photo will look like at any exposure level.

5. The method according to claim 1, characterized in that: Correct the grayscale distribution histogram of the fluorescence image: Only those that meet [I min , I max ] range of pixels, where I min and I max is a number between 1 and 254, I min Less than I max .

6. The method according to claim 1, characterized in that: The optimal exposure level is determined by calculating the mean loss, which is expressed as formula (5): L=|I ref -I mean | (5) Among them I ref is the reference mean of the grayscale value distribution of the designed optimal exposure, I mean It is the corrected mean value of the gray value distribution of the current exposure.

7. The method according to claim 6, characterized in that: The mean loss L for predicting all exposure levels all ={L0, L1, L2, ...}, select the exposure level with the lowest mean loss as the output level for the next exposure.

8. The method according to claim 1, characterized in that: The set A of gain and shutter speed combinations = {gain 0 with shutter 1 / 4000, gain 0 with shutter 1 / 2000, gain 0 with shutter 1 / 1000, gain 0 with shutter 1 / 500, gain 0 with shutter 1 / 250, gain 0 with shutter 1 / 125, gain 0 with shutter 1 / 100, gain 3 with shutter 1 / 100, gain 6 with shutter 1 / 100, gain 9 with shutter 1 / 100, gain 12 with shutter 1 / 100, gain 15 with shutter 1 / 100, gain 18 with shutter 1 / 100, gain 21 with shutter 1 / 100, gain 24 with shutter 1 / 100}.

9. The method according to claim 8, characterized in that: The exposure levels are set to Ae={0.00025, 0.0005, 0.001, 0.002, 0.004, 0.008, 0.01, 0.012, 0.014, 0.016, 0.018, 0.020, 0.022, 0.024, 0.026} respectively.

10. The method according to any one of claims 1 to 9, characterized in that: Establish a feedback system to fine-tune the exposure level based on the imaging results, and incorporate the adjustment results into the training data of the machine learning model.

Images