CBCT parameter adjusting method based on any object

Abstract

The invention relates to a CBCT (cone beam computed tomography) parameter adjustment method based on any object, belongs to the technical field of CBCT, and solves the problem of difficulty in parameter adjustment of CBCT in a complex environment scene. The adjustment method comprises the following steps: judging whether a preset adjustment event occurs in the CBCT or not; if the CBCT has the preset adjustment event, CBCT parameters are adjusted through a preset parameter estimation method; if the CBCT does not have the preset adjustment event, judging whether the difference between the current time point and the time point of the last adjustment exceeds a preset adjustment time threshold value or not; if the difference between the current time point and the time point of the previous adjustment exceeds a preset adjustment time threshold, adjusting the CBCT parameters through a preset parameter estimation method; the preset parameter estimation method comprises the following steps: acquiring projection images of any object at each projection angle by using the CBCT, and adjusting the torsion angle and the transverse offset of the CBCT according to the projection images at each projection angle. The CBCT parameters in a complex environment scene can be adjusted in time.

Description

Technical Field

[0001] This invention relates to the field of cone-beam computed tomography (CBCT) technology, and more particularly to a method for adjusting CBCT parameters based on arbitrary objects. Background Technology

[0002] Cone beam computed tomography (CBCT) is an important three-dimensional nondestructive testing technique with advantages such as fast scanning speed and high spatial resolution. It is widely used in medical imaging, industrial nondestructive testing, materials science and other fields.

[0003] In existing technologies, when CBCT is applied to complex environmental scenes, it often leads to drastic changes in CBCT parameters. Furthermore, the commonly used reconstruction algorithm in CBCT, the Feldkamp-Davis-Kress (FDK) algorithm, has extremely high requirements for the accuracy of CBCT parameters. Any tiny geometric deviation that is not effectively corrected will introduce geometric artifacts, such as boundary blurring, ghosting, or geometric distortion, which seriously affects the quality of reconstructed images and the accuracy of quantitative analysis.

[0004] The existing technology lacks a technical solution for updating and adjusting CBCT parameters in complex environmental scenarios, making it difficult to achieve accurate reconstruction of objects based on CBCT in complex environmental scenarios.

[0005] Therefore, there is an urgent need for a technical solution to adjust CBCT parameters. Summary of the Invention

[0006] Based on the above analysis, the present invention aims to provide a method for adjusting CBCT parameters based on arbitrary objects, in order to solve the problem of difficult parameter adjustment of CBCT in complex environmental scenes.

[0007] This invention provides a method for adjusting CBCT parameters based on arbitrary objects, the method comprising:

[0008] The system monitors CBCT in real time to determine whether a preset adjustment event has occurred. If a preset adjustment event occurs, the CBCT parameters are adjusted using a preset parameter estimation method.

[0009] If no preset adjustment event occurs in CBCT, determine whether the difference between the current time point and the time point of the last adjustment exceeds the preset adjustment time threshold; if the difference between the current time point and the time point of the last adjustment exceeds the preset adjustment time threshold, adjust the CBCT parameters using the preset parameter estimation method.

[0010] The preset parameter estimation methods include:

[0011] CBCT is used to acquire projection images of any object at various projection angles, and the torsional angle and lateral offset of the CBCT are adjusted based on the projection images at each projection angle.

[0012] Based on further improvements to the above adjustment method, the preset adjustment event includes one or more of the following:

[0013] Replace or reinstall the X-ray tube;

[0014] Replace or reinstall the flat panel detector;

[0015] Repair or adjust the robotic arm;

[0016] Repair or adjust the slip ring;

[0017] CBCT is subjected to impact or severe vibration.

[0018] Based on a further improvement to the above adjustment method, the adjustment method further includes:

[0019] Calculate the temperature and humidity differences of the CBCT's operating environment between the current time point and the last adjustment;

[0020] It determines whether the absolute value of the temperature difference exceeds the preset temperature adjustment threshold and whether the absolute value of the humidity difference exceeds the preset relative humidity threshold. If the absolute value of the temperature difference exceeds the preset temperature adjustment threshold or the absolute value of the humidity difference exceeds the preset relative humidity threshold, the CBCT parameters are adjusted in conjunction with the preset parameter estimation method.

[0021] Based on a further improvement of the above adjustment method, the preset temperature adjustment threshold is 6℃;

[0022] The preset relative humidity threshold is 15%RH.

[0023] Based on a further improvement to the above adjustment method, the adjustment of the sway angle and lateral offset of the CBCT according to the projected images at various projection angles includes:

[0024] Identify the object regions in the projected images at each projection angle; determine the preset sampling rows based on the object regions in the projected images at each projection angle.

[0025] Extract the same preset sampling rows from the projected images at each projection angle to obtain the sine graph corresponding to each preset sampling row;

[0026] The first lateral offset of each preset sampling row is determined based on the sine curve corresponding to each preset sampling row. The torsional angle of the CBCT is obtained by linear fitting based on the first lateral offset of each preset sampling row, and is used as the torsional angle of the adjusted CBCT.

[0027] The CBCT's torsion angle is used to perform rotation correction on the projected images at various projection angles. The sine curve corresponding to the center row is extracted from the projected images at each rotation-corrected projection angle, and the lateral offset of the center row is used as the lateral offset of the adjusted CBCT.

[0028] Based on the further improvement of the above adjustment method, the object regions in the projected images at various projection angles are determined by a preset semantic segmentation model. The preset semantic segmentation model adopts any of the following neural network models:

[0029] U-Net series;

[0030] SegNet series;

[0031] DeepLabv3+ series.

[0032] Based on a further improvement to the above adjustment method, the step of determining the preset sampling row according to the object region in the projected image at each projection angle includes:

[0033] Based on the object region in the projected image at each projection angle, determine the upper and lower limits of the row for the object region in the projected image at each projection angle, and find the maximum upper and minimum lower limits of the row.

[0034] Use the minimum row lower limit as the first preset sampling row;

[0035] Starting with the minimum row lower limit, every fixed number of rows is used as a preset sampling row until the maximum row upper limit is reached, thus obtaining all preset sampling rows.

[0036] Based on a further improvement to the above adjustment method, the step of determining the first lateral offset of each preset sampling row according to the sine curve corresponding to each preset sampling row includes:

[0037] Based on the sine curve corresponding to each preset sampling row, multiple second lateral offsets for each preset sampling row are determined by combining various preset lateral offset algorithms.

[0038] Remove the first outlier from the multiple second lateral biases of each preset sampling row to obtain the remaining multiple second lateral biases of each preset sampling row.

[0039] The average of the remaining multiple second lateral biases for each preset sampling row is used as the first lateral bias for each preset sampling row.

[0040] Based on a further improvement to the above adjustment method, the step of removing the first outlier from multiple second lateral biases in each preset sampling row includes:

[0041] Sort the multiple second lateral biases of each preset sampling row according to their size, determine the median of the multiple second lateral biases, and obtain the first median;

[0042] The second median is obtained by determining the median of the absolute values ​​of the differences between each second lateral offset and the first median.

[0043] The threshold range for the first outlier is determined based on the first median and the second median;

[0044] Based on the threshold range of the first outlier, multiple second lateral biases of each preset sampling row are eliminated to obtain multiple remaining second lateral biases of each preset sampling row.

[0045] Based on further improvements to the above adjustment method, the various preset lateral offset algorithms include at least three of the following lateral offset algorithms:

[0046] Center of mass method;

[0047] Cross-correlation method;

[0048] Extreme value method;

[0049] Symmetrical projection method;

[0050] Redundant information method.

[0051] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0052] 1. By monitoring in real time whether a preset adjustment event occurs in CBCT, and combining the difference between the current time point and the time point of the last adjustment with whether it exceeds the preset adjustment time threshold, it is determined whether to adjust the CBCT parameters. When it is necessary to adjust the CBCT parameters, it is only necessary to acquire the projection image of any object at various projection angles through CBCT. The adjustment of CBCT parameters can be achieved without the use of a special phantom, which improves the environmental adaptability of CBCT.

[0053] 2. By filtering the object regions in the projected images at various projection angles and excluding non-object regions, the amount of computation can be greatly reduced. At the same time, by extracting the same preset sampling rows from the projected images at various projection angles, the first lateral offset of each preset sampling row is determined using the sine curve corresponding to each preset sampling row. Based on the first lateral offset of each preset sampling row, linear fitting is performed to obtain the torsion angle of CBCT. The torsion angle of CBCT is then used to perform rotation correction on the projected images at various projection angles. The sine curve corresponding to the center row is extracted from the projected images at each rotation correction angle, and the lateral offset of the center row is used as the lateral offset of the adjusted CBCT.

[0054] 3. By using multiple lateral bias algorithms to calculate multiple second lateral biases for each preset sampling row, and then removing outliers with low correlation, the average of multiple second lateral biases with high correlation is used as the first lateral bias for each preset sampling row. This can greatly improve the accuracy of the lateral bias of each preset sampling row and further improve the accuracy of CBCT parameters.

[0055] 4. For multiple first lateral offsets of multiple preset sampling rows, the torsion angle of CBCT is determined based on the least squares fitting method to improve the accuracy of the torsion angle of CBCT. The torsion angle of CBCT is combined with the rotation correction of the projected image under each projection angle to make the accuracy of the lateral offset of CBCT even higher.

[0056] In this invention, the above-described technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of this invention will be set forth in the following description, and some advantages may become apparent from the description or be learned by practicing the invention. The objects and other advantages of this invention can be realized and obtained from what is particularly pointed out in the description and drawings. Attached Figure Description

[0057] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0058] Figure 1 This is a flowchart illustrating a method for adjusting CBCT parameters based on arbitrary objects, provided in an embodiment of the present invention. Detailed Implementation

[0059] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0060] A specific embodiment of the present invention discloses a method for adjusting CBCT parameters based on arbitrary objects, such as... Figure 1 As shown, the adjustment method includes:

[0061] Step S1: Determine whether a preset adjustment event has occurred in CBCT by real-time monitoring; if a preset adjustment event has occurred in CBCT, adjust the CBCT parameters using a preset parameter estimation method;

[0062] Step S2: If no preset adjustment event occurs in CBCT, determine whether the difference between the current time point and the time point of the last adjustment exceeds the preset adjustment time threshold; if the difference between the current time point and the time point of the last adjustment exceeds the preset adjustment time threshold, adjust the CBCT parameters using the preset parameter estimation method.

[0063] The preset parameter estimation methods include:

[0064] CBCT is used to acquire projection images of any object at various projection angles, and the torsional angle and lateral offset of the CBCT are adjusted based on the projection images at each projection angle.

[0065] Specifically, when CBCT is applied in complex environments, it is more likely to face various unexpected events. In step S1, CBCT is monitored in real time to determine whether a preset adjustment event has occurred.

[0066] Specifically, such as Figure 1 As shown, the preset adjustment event represents one or more mandatory events that will cause changes in CBCT parameters when the event occurs. The changes in CBCT parameters will severely affect the accuracy of the reconstructed image obtained from the projection image acquired by CBCT, making CBCT unusable.

[0067] Preferably, the preset adjustment event includes one or more of the following:

[0068] Replace or reinstall the X-ray tube;

[0069] Replace or reinstall the flat panel detector;

[0070] Repair or adjust the robotic arm;

[0071] Repair or adjust the slip ring;

[0072] CBCT is subjected to impact or severe vibration.

[0073] Specifically, when CBCT is applied in complex environments, it may become unusable due to impacts or severe vibrations.

[0074] Specifically, changes to the X-ray tube, flat panel detector, robotic arm, or slip ring in a CBCT may render the CBCT unusable.

[0075] When a preset adjustment event occurs in CBCT, the CBCT parameters need to be adjusted using a preset parameter estimation method to improve the quality of the reconstructed image obtained from the projection image acquired by CBCT.

[0076] Specifically, such as Figure 1As shown, in step S2, when no preset adjustment event occurs in CBCT, it is determined how long CBCT has not been adjusted and whether the CBCT parameters need to be adjusted, based on the preset adjustment time threshold.

[0077] Specifically, in step S2, if the difference between the current time point and the previous adjustment time point exceeds the preset adjustment time threshold, the CBCT parameters need to be adjusted using a preset parameter estimation method.

[0078] It is worth noting that CBCT parameters are affected by temperature, mainly due to the thermal expansion and contraction of the mechanical structure and the drift of detector performance. This can cause changes in CBCT parameters, which in turn can lead to geometric artifacts or grayscale drift in the reconstructed images, affecting the quality of the reconstructed images and the accuracy of quantitative analysis.

[0079] like Figure 1 As shown, preferably, the adjustment method further includes:

[0080] Step S3: Calculate the temperature and humidity differences of the CBCT's operating environment between the current time point and the last adjustment;

[0081] Step S4: Determine whether the absolute value of the temperature difference exceeds the preset temperature adjustment threshold and whether the absolute value of the humidity difference exceeds the preset relative humidity threshold; if the absolute value of the temperature difference exceeds the preset temperature adjustment threshold or the absolute value of the humidity difference exceeds the preset relative humidity threshold, then adjust the CBCT parameters in conjunction with the preset parameter estimation method.

[0082] Specifically, in step S3, the temperature and humidity of the CBCT's working environment at the current time point are collected in real time using temperature and humidity sensors. At the same time, the temperature and humidity of the working environment at the time of each CBCT parameter adjustment are recorded.

[0083] Specifically, in step S3, the difference between the temperature of the CBCT's working environment at the current time and the temperature of the CBCT's working environment at the last adjustment is calculated as the temperature difference; at the same time, the difference between the humidity of the CBCT's working environment at the current time and the humidity of the CBCT's working environment at the last adjustment is calculated as the humidity difference.

[0084] Specifically, such as Figure 1 As shown, in step S4, it is determined whether the absolute value of the temperature difference exceeds the preset temperature adjustment threshold. If the absolute value of the temperature difference exceeds the preset temperature adjustment threshold, it is considered that the CBCT parameters have changed, and this change results in a lower quality of the reconstructed image based on the projection image acquired by CBCT. At this time, it is necessary to adjust the CBCT parameters using the preset parameter estimation method.

[0085] Specifically, such as Figure 1 As shown, in step S4, it is determined whether the absolute value of the humidity difference exceeds the preset relative humidity threshold.

[0086] Specifically, after obtaining the absolute value of the humidity difference, the ratio of the absolute value of the humidity difference to the humidity of the working environment of CBCT during the last adjustment is further calculated. If the ratio exceeds the preset relative humidity threshold, it is considered that the CBCT parameters have changed, and this change has resulted in a lower quality of the reconstructed image based on the projection image acquired by CBCT. In this case, the CBCT parameters are adjusted in conjunction with the preset parameter estimation method.

[0087] Preferably, the preset temperature adjustment threshold is 6°C;

[0088] The preset relative humidity threshold is 15%RH.

[0089] Specifically, RH is used to represent the relative humidity of the CBCT's working environment at the current time point and the relative humidity of the CBCT's working environment at the time of the last adjustment. If the absolute value of the humidity difference and the ratio of the humidity of the CBCT's working environment at the time of the last adjustment exceed the preset relative humidity threshold, the CBCT parameters are adjusted in combination with the preset parameter estimation method.

[0090] Preferably, adjusting the sway angle and lateral offset of the CBCT based on the projected images at various projection angles includes:

[0091] Identify the object regions in the projected images at each projection angle; determine the preset sampling rows based on the object regions in the projected images at each projection angle.

[0092] Extract the same preset sampling rows from the projected images at each projection angle to obtain the sine graph corresponding to each preset sampling row;

[0093] The first lateral offset of each preset sampling row is determined based on the sine curve corresponding to each preset sampling row. The torsional angle of the CBCT is obtained by linear fitting based on the first lateral offset of each preset sampling row, and is used as the torsional angle of the adjusted CBCT.

[0094] The CBCT's torsion angle is used to perform rotation correction on the projected images at various projection angles. The sine curve corresponding to the center row is extracted from the projected images at each rotation-corrected projection angle, and the lateral offset of the center row is used as the lateral offset of the adjusted CBCT.

[0095] Specifically, the CBCT parameters involved in the embodiments of the present invention include the lateral offset of the CBCT and the yaw angle of the CBCT. The lateral offset of the CBCT and the yaw angle of the CBCT are adjusted by a method for adjusting CBCT parameters based on arbitrary objects provided in the embodiments of the present invention, and the adjusted yaw angle and the adjusted lateral offset of the CBCT are determined.

[0096] Specifically, in CBCT, any object is placed on a rotating platform, and the X-ray source and detector are placed on both sides of the rotating platform. X-rays are emitted towards the object through the X-ray source, and the projected images are received by the detector to obtain the projected images of the object at various projection angles.

[0097] Specifically, when using CBCT to acquire projected images of an object, the rotating platform can be rotated, thereby changing the projection angle of the object relative to the X-ray source and detector. At each projection angle, the CBCT detector acquires the projection image at that projection angle, thus obtaining the projection image of the object at each projection angle.

[0098] Preferably, when using CBCT to acquire projection images of any object at various projection angles, the angle difference between any two adjacent projection angles is the same.

[0099] Specifically, when the rotating platform rotates, it rotates by a fixed angle each time, so that the angle difference between any two adjacent projection angles is the same. At each projection angle, the detector collects the projection image of the object at that projection angle.

[0100] For example, in CBCT, the total projection angle includes 360°. If a total of 720 projection images are acquired, the difference between two adjacent projection angles is 0.5°, that is, a projection image is acquired every 0.5° interval.

[0101] Specifically, the size of the projected image is the same at each projection angle, for example, 2340*2882.

[0102] Specifically, the projected images at various projection angles are detected to identify object regions and non-object regions within those images. Object regions represent areas containing projected content, while non-object regions represent areas without projected content.

[0103] Preferably, the object regions in the projected images at various projection angles are determined by a preset semantic segmentation model, wherein the preset semantic segmentation model adopts any of the following neural network models:

[0104] U-Net series;

[0105] SegNet series;

[0106] DeepLabv3+ series.

[0107] Specifically, non-object regions in a projected image typically have high brightness and uniform color, such as white or light colors, while object regions may have various colors and textures. To improve the recognition accuracy of object and non-object regions, this embodiment of the invention uses a preset semantic segmentation model to identify object and non-object regions in the projected image, thereby determining the object regions in the projected image at various projection angles.

[0108] Specifically, the U-Net series features a simple structure, is suitable for binary segmentation, and performs well on small datasets.

[0109] Specifically, the SegNet series, with its unique pooling indexing mechanism and memory-efficient design, is suitable for real-time segmentation and resource-constrained scenarios.

[0110] Specifically, the DeepLabv3+ series is suitable for scenarios requiring higher precision and has sufficient computing resources.

[0111] It is worth noting that when implementing the preset parameter estimation method provided in the embodiments of the present invention, a preset semantic segmentation model can be reasonably selected according to the specific usage requirements and hardware requirements, so as to quickly obtain the adjusted CBCT parameter values.

[0112] Specifically, after determining the object regions in the projected images at each projection angle, a preset sampling row is determined by combining the object regions in the projected images at all projection angles.

[0113] Preferably, determining the preset sampling row based on the object region in the projected image at each projection angle includes:

[0114] Based on the object region in the projected image at each projection angle, determine the upper and lower limits of the row for the object region in the projected image at each projection angle, and find the maximum upper and minimum lower limits of the row.

[0115] The preset sampling rows are determined based on the maximum row upper limit and the minimum row lower limit.

[0116] It is understandable that the area occupied by an object in a projected image at different angles often varies, that is, the upper and lower limits of the row it occupies are often different.

[0117] Specifically, after determining the object region in the projected image at each projection angle, the upper and lower limits of the rows occupied by the object region are further determined. The upper limits of the rows of the object region in the projected image at all projection angles are compared to find the maximum upper limit of the rows, and the lower limits of the rows of the object region in the projected image at all projection angles are compared to find the minimum lower limit of the rows.

[0118] Specifically, the preset sampling rows are determined based on the maximum row upper limit and the minimum row lower limit. For example, the size of the projected image at each projection angle is 2340*2882, that is, 2340 rows and 2882 columns. If the maximum row upper limit is 2000 and the minimum row lower limit is 200, then the preset sampling rows are selected from rows 200-2000.

[0119] Preferably, determining the preset sampling rows based on the maximum row upper limit and the minimum row lower limit includes:

[0120] Use the minimum row lower limit as the first preset sampling row;

[0121] Starting with the minimum row lower limit, every fixed number of rows is used as a preset sampling row until the maximum row upper limit is reached, thus obtaining all preset sampling rows.

[0122] Specifically, the minimum row limit of 200 is taken as the first preset sampling row, i.e. the first preset sampling row; the row difference between two adjacent preset sampling rows is set to a fixed row of 10, then the second preset sampling row is 20, and so on to obtain all preset sampling rows of 10, 20, 30...1990, 2000.

[0123] It is worth noting that if, after selecting a preset sampling row, the difference between the selected preset sampling row and the maximum row limit is less than 10 rows, then the selected preset sampling row will be used as the last preset sampling row, and no more preset sampling rows will be selected.

[0124] Specifically, after determining the preset sampling row, the same preset sampling row is extracted from the projection images under each projection angle, that is, the same preset sampling row is extracted from the projection images under each projection angle, and the combination is used as the sine graph corresponding to the preset sampling row.

[0125] It is worth noting that the more projection images there are, the more data in the preset sampling rows there are, and the higher the accuracy of the final CBCT parameters will be, but at the same time, the processing cost will also increase.

[0126] Preferably, when extracting the same preset sampling row from the projection images at each projection angle, it is also necessary to preprocess the projection images at each projection angle. The preprocessing includes one or more of the following:

[0127] Dark field correction;

[0128] Brightness correction;

[0129] Logarithmic transformation;

[0130] Dead pixel correction;

[0131] Noise filtering;

[0132] Beam hardening correction;

[0133] Scattering correction.

[0134] Specifically, dark field correction, bright field correction, logarithmic transformation, bad pixel correction, noise filtering, beam hardening correction, and scattering correction can be reasonably set according to the actual situation to reduce the noise impact on the projected image.

[0135] Specifically, a first lateral offset is corresponding to the sine graph of a preset sampling row. The first lateral offset of each preset sampling row is calculated based on the sine graph of each preset sampling row.

[0136] Preferably, determining the first lateral offset of each preset sampling row based on the sine wave corresponding to each preset sampling row includes:

[0137] Based on the sine curve corresponding to each preset sampling row, multiple second lateral offsets for each preset sampling row are determined by combining various preset lateral offset algorithms.

[0138] Remove the first outlier from the multiple second lateral biases of each preset sampling row to obtain the remaining multiple second lateral biases of each preset sampling row.

[0139] The average of the remaining multiple second lateral biases for each preset sampling row is used as the first lateral bias for each preset sampling row.

[0140] Specifically, multiple preset lateral bias algorithms are used to determine multiple second lateral biases based on the sine curves corresponding to preset sampling rows. That is, each preset lateral bias algorithm determines a second lateral bias based on the sine curves corresponding to preset sampling rows.

[0141] Preferably, the plurality of preset lateral offset algorithms include at least three of the following lateral offset algorithms:

[0142] Center of mass method;

[0143] Cross-correlation method;

[0144] Extreme value method;

[0145] Symmetrical projection method;

[0146] Redundant information method.

[0147] Specifically, the centroid method, cross-correlation method, extreme value method, symmetric projection method, and redundant information method are existing algorithms for calculating lateral bias, which will not be elaborated here.

[0148] Specifically, after obtaining the second lateral bias determined by each preset lateral bias algorithm, outliers in the multiple second lateral biases are removed, that is, the first outlier is removed from the multiple second lateral biases of each preset sampling row to obtain the remaining multiple second lateral biases of each preset sampling row.

[0149] For example, when using five preset lateral bias algorithms, the number of outliers can be reasonably determined according to the situation. The number of outliers can be 1, resulting in 4 remaining second lateral biases for each preset sampling row.

[0150] Preferably, removing the first outlier from multiple second lateral biases in each preset sampling row includes:

[0151] Sort the multiple second lateral biases of each preset sampling row according to their size, determine the median of the multiple second lateral biases, and obtain the first median;

[0152] The second median is obtained by determining the median of the absolute values ​​of the differences between each second lateral offset and the first median.

[0153] The threshold range for the first outlier is determined based on the first median and the second median;

[0154] Based on the threshold range of the first outlier, multiple second lateral biases of each preset sampling row are eliminated to obtain multiple remaining second lateral biases of each preset sampling row.

[0155] Specifically, if the number of first outliers is set to 1, then one outlier will be removed from multiple second lateral biases.

[0156] Specifically, the multiple second lateral offsets are sorted in ascending order, and the second lateral offset at the middle position is selected as the first median.

[0157] Specifically, the median of the absolute values ​​of the differences between the second lateral bias and the first median is calculated and used as the second median.

[0158] Specifically, the threshold range for the first outlier can be set using the common rule of "3 standard deviations" in simulated statistics. For example, the first median is Med. U If the second median is MAD, then the normal range of the second lateral bias is set to MedU±(3×MAD / 0.6745), and values ​​outside the normal range are taken as the threshold range of the first outlier.

[0159] Specifically, after determining the threshold range of the first outlier, the threshold ranges of the first outlier are removed from the multiple second lateral biases to obtain the remaining multiple second lateral biases, which are used as the remaining multiple second lateral biases of the preset sampling row.

[0160] Specifically, after determining the remaining multiple second lateral offsets for each preset sampling row, the average value of the remaining multiple second lateral offsets for each preset sampling row is taken as the first lateral offset for each preset sampling row.

[0161] Specifically, the CBCT torsion angle is obtained by linear fitting based on the first lateral offset of each preset sampling row.

[0162] Specifically, based on multiple first lateral offsets of multiple preset sampling rows, the torsional angle of CBCT is determined by the least squares fitting method.

[0163] Preferably, the yaw angle of the CBCT is determined by the following formula:

[0164]

[0165] Where α represents the yaw angle of the CBCT, β represents an undetermined constant, and u1 k z represents the first lateral offset of the k-th preset sampling row. k represents the row number of the k-th preset sampling row, and r represents the number of rows in the remaining preset sampling rows.

[0166] Specifically, substitute multiple values ​​of α and β into formula u1. k -(α×z k +β) to obtain multiple values, compare the magnitudes of the multiple values, and take the α corresponding to the minimum value as the torsional angle of CBCT.

[0167] Specifically, the CBCT's yaw angle is used to perform rotational correction on the projected images at various projection angles.

[0168] Specifically, after obtaining the torsion angle, geometric correction is applied to the projected images under each original projection angle. The coordinates of each pixel on the detector are rotated and corrected with the detector center as the origin, based on the torsion angle.

[0169] Specifically, the sine curves corresponding to the center row are extracted from the projection images at each projection angle after rotation correction, and the lateral offset of the center row is used as the lateral offset of the adjusted CBCT.

[0170] Preferably, the step of extracting the sine wave corresponding to the center row from the projection image at each projection angle after rotation correction, and using the lateral offset of the center row as the lateral offset of CBCT, includes:

[0171] Based on the sine curve corresponding to the center row, multiple third lateral offsets of the center row are determined by combining various preset lateral offset algorithms;

[0172] Remove the third outlier from the multiple third lateral offsets of the center row to obtain the remaining multiple third lateral offsets of the center row;

[0173] The average of the remaining third lateral offsets of the center row is used as the lateral offset of the center row.

[0174] Specifically, the sine wave corresponding to the center row is determined, and multiple third lateral offsets of the center row are determined using various preset lateral offset algorithms. Each preset lateral offset algorithm determines one third lateral offset.

[0175] Outliers in the multiple third lateral offsets are removed to obtain the remaining multiple third lateral offsets in the center row.

[0176] Preferably, the step of removing the third outlier from the multiple third lateral offsets of the center row to obtain the remaining multiple third lateral offsets of the center row includes:

[0177] Sort the multiple third horizontal offsets of the center row according to their size, determine the median of the multiple third horizontal offsets, and obtain the third median;

[0178] The fourth median is obtained by determining the median of the absolute values ​​of the differences between each third lateral offset and the third median.

[0179] The threshold range for the third outlier is determined based on the third and fourth medians;

[0180] Based on the threshold range of the third outlier, multiple third lateral offsets of the center row are removed to obtain the remaining multiple third lateral offsets of the center row.

[0181] It is worth noting that the method used to determine the threshold range of the third outlier based on the third and fourth medians is the same as the method used to determine the threshold range of the first outlier based on the first and second medians, and will not be repeated here.

[0182] Specifically, after determining the remaining third lateral offsets of the center row, the average of the remaining third lateral offsets of the center row is taken as the lateral offset of the center row.

[0183] Specifically, the lateral offset of the center line is used as the lateral offset of CBCT, thus obtaining the adjusted lateral offset of CBCT.

[0184] Compared with existing technologies, the CBCT parameter adjustment method based on arbitrary objects provided in this invention monitors in real time whether a preset adjustment event occurs in CBCT, and determines whether to adjust CBCT parameters based on whether the difference between the current time point and the previous adjustment time point exceeds a preset adjustment time threshold. When CBCT parameters need to be adjusted, it is only necessary to acquire projection images of arbitrary objects at various projection angles using CBCT, without the need for a dedicated phantom, thus improving the environmental adaptability of CBCT. Furthermore, by filtering preset sampling rows through object regions in the projection images at various projection angles and excluding non-object regions, the computational load can be greatly reduced. At the same time, by extracting the same preset sampling rows from the projection images at various projection angles, the first lateral offset of each preset sampling row is determined using the sine curve corresponding to each preset sampling row, and linear fitting is performed based on the first lateral offset of each preset sampling row. The torsion angle of the CBCT is obtained, and the torsion angle is used to perform rotation correction on the projected images at various projection angles. The sine curve corresponding to the center row is extracted from the projected images at each rotation-corrected projection angle, and the lateral offset of the center row is used as the adjusted lateral offset of the CBCT. At the same time, multiple second lateral offsets for each preset sampling row are calculated using various lateral offset algorithms. Outliers with low correlation are then removed, and the average of multiple second lateral offsets with high correlation is used as the first lateral offset for each preset sampling row. This can greatly improve the accuracy of the lateral offset of each preset sampling row, and further improve the accuracy of the CBCT parameters. Finally, the torsion angle of the CBCT is determined based on the least squares fitting method for multiple first lateral offsets of multiple preset sampling rows, improving the accuracy of the CBCT torsion angle. Combining the torsion angle of the CBCT with the rotation correction of the projected images at various projection angles makes the obtained CBCT lateral offset even more accurate.

[0185] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0186] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for adjusting CBCT parameters based on arbitrary objects, characterized in that, The adjustment method includes: The system monitors CBCT in real time to determine whether a preset adjustment event has occurred. If a preset adjustment event occurs, the CBCT parameters are adjusted using a preset parameter estimation method. If no preset adjustment event occurs in CBCT, determine whether the difference between the current time point and the time point of the last adjustment exceeds the preset adjustment time threshold; if the difference between the current time point and the time point of the last adjustment exceeds the preset adjustment time threshold, adjust the CBCT parameters using the preset parameter estimation method. The preset parameter estimation methods include: CBCT is used to acquire projection images of any object at various projection angles, and the torsional angle and lateral offset of the CBCT are adjusted based on the projection images at each projection angle.

2. The adjustment method according to claim 1, characterized in that, The preset adjustment events include one or more of the following: Replace or reinstall the X-ray tube; Replace or reinstall the flat panel detector; Repair or adjust the robotic arm; Repair or adjust the slip ring; CBCT is subjected to impact or severe vibration.

3. The adjustment method according to claim 1, characterized in that, The adjustment method further includes: Calculate the temperature and humidity differences of the CBCT's operating environment between the current time point and the last adjustment; It determines whether the absolute value of the temperature difference exceeds the preset temperature adjustment threshold and whether the absolute value of the humidity difference exceeds the preset relative humidity threshold. If the absolute value of the temperature difference exceeds the preset temperature adjustment threshold or the absolute value of the humidity difference exceeds the preset relative humidity threshold, the CBCT parameters are adjusted in conjunction with the preset parameter estimation method.

4. The adjustment method according to claim 3, characterized in that, The preset temperature adjustment threshold is 6℃; The preset relative humidity threshold is 15%RH.

5. The adjustment method according to any one of claims 1-4, characterized in that, The adjustment of the CBCT's yaw angle and lateral offset based on the projected images at various projection angles includes: Identify the object regions in the projected images at each projection angle; determine the preset sampling rows based on the object regions in the projected images at each projection angle. Extract the same preset sampling rows from the projected images at each projection angle to obtain the sine graph corresponding to each preset sampling row; The first lateral offset of each preset sampling row is determined based on the sine curve corresponding to each preset sampling row. The torsional angle of the CBCT is obtained by linear fitting based on the first lateral offset of each preset sampling row, and is used as the torsional angle of the adjusted CBCT. The CBCT's torsion angle is used to perform rotation correction on the projected images at various projection angles. The sine curve corresponding to the center row is extracted from the projected images at each rotation-corrected projection angle, and the lateral offset of the center row is used as the lateral offset of the adjusted CBCT.

6. The adjustment method according to claim 5, characterized in that, The object regions in the projected images at various projection angles are determined by a preset semantic segmentation model, wherein the preset semantic segmentation model adopts any of the following neural network models: U-Net series; SegNet series; DeepLabv3+ series.

7. The adjustment method according to claim 6, characterized in that, The step of determining the preset sampling row based on the object region in the projected image at each projection angle includes: Based on the object region in the projected image at each projection angle, determine the upper and lower limits of the row for the object region in the projected image at each projection angle, and find the maximum upper and minimum lower limits of the row. Use the minimum row lower limit as the first preset sampling row; Starting with the minimum row lower limit, every fixed number of rows is used as a preset sampling row until the maximum row upper limit is reached, thus obtaining all preset sampling rows.

8. The adjustment method according to claim 5, characterized in that, The step of determining the first lateral offset of each preset sampling row based on the sine curve corresponding to each preset sampling row includes: Based on the sine curve corresponding to each preset sampling row, multiple second lateral offsets for each preset sampling row are determined by combining various preset lateral offset algorithms. Remove the first outlier from the multiple second lateral biases of each preset sampling row to obtain the remaining multiple second lateral biases of each preset sampling row; The average of the remaining multiple second lateral biases for each preset sampling row is used as the first lateral bias for each preset sampling row.

9. The adjustment method according to claim 8, characterized in that, The step of removing the first outlier from multiple second lateral biases in each preset sampling row includes: Sort the multiple second lateral biases of each preset sampling row according to their size, determine the median of the multiple second lateral biases, and obtain the first median; The second median is obtained by determining the median of the absolute values ​​of the differences between each second lateral offset and the first median. The threshold range for the first outlier is determined based on the first median and the second median; Based on the threshold range of the first outlier, multiple second lateral biases of each preset sampling row are eliminated to obtain multiple remaining second lateral biases of each preset sampling row.

10. The adjustment method according to claim 8, characterized in that, The various preset lateral offset algorithms include at least three of the following lateral offset algorithms: Center of mass method; Cross-correlation method; Extreme value method; Symmetrical projection method; Redundant information method.

Images