CN115019589B - Intelligent CT teaching simulation system based on optics - Google Patents

Abstract

The invention relates to the field of medical teaching equipment, and aims to propose an optical-based intelligent CT teaching simulation system to assist teaching and help students improve their understanding and research capabilities of CT imaging principles and new CT technologies. To this end, the technical solution adopted by the present invention is an optical-based intelligent CT teaching simulation system, including a light source module, a scanning module, a collection module, an intelligent control and reconstruction module and a display module. The light source module and scanning module are used to simulate CT scanning. The signals obtained by scanning are collected by the acquisition module and sent to the intelligent control and reconstruction module. The intelligent control and reconstruction module controls each module and sends the results to the display module for display. The invention is mainly used in the design and manufacture of medical teaching equipment.

Description

Optical-based intelligent CT teaching simulation system

Technical field

The invention relates to the field of medical teaching equipment, and in particular to an optical-based intelligent CT teaching simulation system.

Background technique

Currently, CT (Computed Tomography) plays a very important role in disease diagnosis and examination. In the teaching and training process of medical imaging majors, compared with pure theoretical explanations, experimental teaching can allow students to better understand and master the operation of CT, imaging principles and imaging quality control, which will improve the teaching effect and cultivate Medical imaging professionals are of great significance.

However, clinical CT equipment is expensive, carries radiation risks, and requires high protection and use environment. It is difficult for universities to invest the funds and conditions to meet the demand. Although the existing CT simulator technology has scanning and display modules, the projection images and CT images are stored data, not the actual collected projection images, nor the reconstructed tomograms, so they can only be used for students familiar with CT. Operating procedures. However, the principles of CT imaging are complicated. Currently, there is a lack of a system that can not only simulate operations, but also intuitively simulate the projection and reconstruction process of CT to facilitate students' understanding of the principles of CT imaging. In addition, with the development of artificial intelligence and CT technology, intelligent positioning and dose intelligent control are the development directions of CT technology. The development of intelligent CT teaching simulation systems is of great value for students to understand, learn and research new intelligent positioning and dose control technologies.

Contents of the invention

In order to overcome the shortcomings of the existing technology, the present invention aims to propose an optical-based intelligent CT teaching simulation system to assist teaching and help students improve their understanding and research capabilities of CT imaging principles and new CT technologies. To this end, the technical solution adopted by the present invention is an optical-based intelligent CT teaching simulation system, a light source module, a scanning module, an acquisition module, an intelligent control and reconstruction module and a display module;

The light source module includes a light source, a beam homogenizing device and an adjustable slit. The adjustable slit consists of four flat plates: upper, lower, left and right. It is used to adjust the size of the light-transmitting window to adjust the illumination field of view; among them, the light source is The switchable wavelength light source is used to simulate dual-energy CT or use different wavelengths to simulate different tube voltage adjustments. The light source module can adopt two structures: one is that the light source passes through the diffuse scattering sheet and continues to propagate through the adjustable slit; The light source is reflected by the diffuse reflector and then continues to propagate through the adjustable slit;

The scanning module includes a vertical translation stage, a rotating platform, an imaging phantom and a water tank. The imaging phantom is installed on the rotating platform, and the rotating platform is installed on the vertical translation stage. The rotating platform is used to rotate the imaging phantom for multi-angle projection. During acquisition, the vertical translation stage moves up and down with the rotating platform. When the vertical translation stage does not move, the rotating platform rotates to simulate the translational step-by-step scanning of CT. When the rotating translation stage rotates, the vertical translation stage moves vertically. Move downward to simulate the spiral scanning of CT. The imaging phantom is an imaging phantom with a certain light absorption distribution that allows light to pass through;

The two surfaces through which light propagates through the water tank are flat plates with light-transmitting windows;

The collection module includes an iris, a lens and a camera. The light emitted by the light source module passes through the imaging phantom, passes through the light-transmitting window of the water tank, passes through the iris and lens in turn, and is finally collected by the camera;

The intelligent control and reconstruction module includes a computer, and the rotating platform, vertical translation stage and camera are connected to the computer through electrical connecting lines. On the one hand, the computer uses the internal intelligent control module to control the rotation of the scanning module's rotating platform and the movement of the vertical translation stage, as well as to control the image acquisition of the camera. On the other hand, after the acquisition is completed, the internal reconstruction module uses the corresponding reconstruction algorithm to capture the camera. The projection image is reconstructed into a tomographic image;

In addition, the computer’s intelligent control module is also used for intelligent positioning and dose intelligent control;

The display module is used to display the projection image, sinogram and reconstructed image display at each angle to simulate the projection image, sinogram and reconstructed image display of the CT image.

There are two implementation processes for the intelligent positioning function: the first one: 1) The user sets the inspection site in advance, which is the area of interest, 2) the camera collects an image, 3) determines the area of interest in the collected image through image recognition and segmentation technology, 4) Calculate the upper, lower and center coordinates of the area of interest in the vertical direction. 5) If the area of interest is larger than the system imaging field of view, control the vertical translation stage to move the imaging object so that the lower limit coordinates of the area of interest are slightly higher than the lower limit of the system imaging field of view. The coordinates are used as the starting position of the imaging scan; if the area of interest is less than or equal to the system imaging field of view, the vertical translation stage is controlled to move the imaging object so that the vertical center coordinates of the area of interest coincide with the center coordinates of the system imaging field of view as the imaging scanning position. That is to realize the intelligent positioning function; the second type: 1) the camera collects an image, 2) the user manually outlines the area of interest on the collected image, 3) calculates the upper and lower limits and center coordinates of the area of interest in the vertical direction, 4 ) If the area of interest is larger than the system imaging field of view, control the vertical translation stage to move the imaging object so that the lower limit coordinates of the area of interest are slightly higher than the lower limit coordinates of the system imaging field of view as the starting position of the imaging scan; if the area of interest is less than or equal to the system imaging field of view , then the vertical translation stage is controlled to move the imaging object so that the vertical center coordinates of the area of interest coincide with the center coordinates of the system imaging field of view as the imaging scanning position. That is to realize the intelligent positioning function.

The dose intelligent control function estimates the dose required for imaging based on the area to be imaged, thereby automatically controlling the intensity of the appropriate light source or the exposure time of the camera. The specific steps are: 1) Learn a large amount of data through deep learning methods and train from a single projection image The neural network mapped to the three-dimensional tomographic image determines the neural network model after training and verification. The input of the training data and verification data used in deep learning is a single projection image, and the output is the corresponding real three-dimensional tomographic image; 2) Collect a single projection image of the area to be imaged through the camera, and use the neural network model learned previously to estimate the three-dimensional tomographic image corresponding to the area to be imaged; 3) Use the estimated three-dimensional tomographic image to estimate the dose required for imaging; 4) The computer automatically controls the intensity of the light source or the exposure time of the camera based on the calculated dose estimate.

The dose intelligent control function estimates the dose required for imaging based on the area to be imaged, thereby automatically controlling the intensity of the appropriate light source or the exposure time of the camera. The detailed steps are:

1) The computer learns a large amount of data through deep learning methods to train a neural network that maps a single projection map to a three-dimensional tomographic image. After training and verification, the neural network model is determined. The inputs of training data and verification data used in deep learning are both A single projection image is output as the corresponding real three-dimensional tomographic image; 2) A single projection image of the area to be imaged is collected through the camera, and the computer uses the neural network model learned previously to estimate the three-dimensional tomographic image corresponding to the area to be imaged; 3) The computer uses the estimated three-dimensional tomographic image to estimate the dose required for imaging, that is, the light intensity of the light source with a fixed exposure time or the exposure time of the camera with a fixed light source intensity. For the light intensity I ₂ of the incident camera and the camera's response value I _3The relationship is I ₃ =I ₂ *D*E, where E is the exposure time, D is the conversion coefficient of I ₂ and I ₃ within the unit exposure time, the light intensity I ₁ of the light source and the light intensity passing through the phantom are also That is, the relationship between the light intensity I ₂ incident on the camera is I ₂ =I ₁ *(M*C _i ), where M is the attenuation of the light intensity from the light source to the incident phantom, and C _i is calculated using the estimated three-dimensional tomographic image. The light intensity at each angle is attenuated by the projection after the phantom, i=1, 2,..., n, n is the projection angle, then the relationship between the light source intensity I ₁ and the camera's response value I ₃ is I ₁ =I ₂ / (M*C _i )=I ₃ /(D*E)/(M*C _i )=I ₃ /(D*E*M*C _i ). In the first case, when the purpose is to obtain good imaging quality, when the exposure time E is constant, the maximum value of the light source intensity I ₁ should be less than I _3max / (D*E*M*C _max ), where I _3max is The maximum value I ₃ can take, C _max is the maximum value of C _i , that is, the corresponding light intensity attenuation of the incident light passing through the phantom is minimum, or when the light source intensity I ₁ is constant, the exposure time E should be less than I _3max /( D*I ₁ *M*C _max ); in the second case, when the purpose is to use a lower dose, when the exposure time E is constant, the maximum value of the light source intensity I ₁ should be less than I _3max /(D*E *M*C _max ) simultaneously takes a value greater than I _3min /(D*E*M*C _min ), where I _3min is the image noise value when the camera has no light input when the exposure time is E, and C _min is the value of C _i The minimum value corresponds to the maximum light intensity attenuation of the incident light passing through the phantom; or when the light source intensity I ₁ is constant, the exposure time E should be less than the simultaneous value of I _3max / (D*I ₁ *M*C _max ) Greater than I _3min / (D*I ₁ *M*C _min ); 4) The computer automatically controls the intensity of the light source or the exposure time of the camera based on the calculated and estimated dose.

The neural network used is a convolutional neural network.

The imaging process is as follows:

1) Fill the water tank in the scanning module with water, install the imaging phantom on the rotating platform, and manually or by computer control the vertical translation stage through the intelligent positioning function of the intelligent control module to move the imaging phantom to the appropriate imaging position;

2) Turn on the light source of the light source module. The light emitted by the light source passes through the beam homogenizing device to form a relatively uniform beam. After adjusting the light size through the adjustable slit, it passes through the light window on the side of the water tank and shines on the imaging phantom. The light passes through the phantom and then passes through the light window on the other side of the water tank. It passes through the iris and the lens and enters the camera. The intensity of the light source or the exposure of the camera is controlled manually or by computer through the dose intelligent control function of the intelligent control module. time for subsequent projection acquisition;

3) For translational step-by-step scanning, there are two scanning modes. The first is that the computer controls the vertical translation stage to remain stationary. The rotating platform drives the imaging phantom to rotate. It stops at a certain angle every time it rotates, and the light emitted by the light source is collected through the camera. The projection through the imaging phantom is recorded as the projection at one angle. In this way, the projections from multiple angles are repeatedly collected. The rotating platform returns to the initial angle when scanning is started and stops. Then the vertical translation platform moves downward for a certain distance and continues to repeat. The above collection is to complete the projection of other parts of the phantom in the vertical direction; the second method is to control the vertical translation stage stationary by the computer, and the rotating platform drives the imaging phantom to rotate continuously, and the camera collects projections at an angle at certain intervals until the acquisition After completing the projection of multiple angles, the rotating platform returns to the initial angle when scanning and stops, and then the vertical translation stage moves downward for a certain distance, and continues to repeat the above collection to complete the projection of other parts of the phantom in the vertical direction; for the spiral Scanning, computer control drives the rotating platform to rotate the imaging phantom, and at the same time, the vertical translation stage moves downward, and the projections from multiple angles are collected through the camera. During the process of collecting projections, the projected images are displayed on the display module for real-time observation. and monitor the status of collected images;

4) The computer uses the multiple angle projections collected in 3) to perform reconstruction through the reconstruction module in the intelligent control and reconstruction module, and then displays the reconstruction results through the display module.

The characteristics and beneficial effects of the present invention are:

1) CT imaging uses X-rays to pass through the human body along a straight line, and reconstructs the image to obtain a cross-sectional image by collecting X-ray projections in multiple directions. The present invention uses a light source (such as visible light) to pass through a phantom of a certain transparent material along a straight line to perform projection reconstruction, and its physical process and reconstruction process are consistent with CT imaging. Therefore, the present invention can intuitively simulate the CT projection and reconstruction process, and can assist students to deeply understand the principles of CT imaging. The present invention uses light sources (such as visible light) for projection reconstruction, which completely avoids the radiation hazards of X-rays, reduces shielding and protection requirements, and is very suitable for teaching demonstrations of CT principles. At the same time, compared with the cost, maintenance and usage environment requirements of real CT, the cost and requirements of the present invention are much lower.

2) By adjusting the size of the adjustable slit, the present invention can control the irradiation range of the light source, thereby simulating the adjustment of the irradiation range of the CT tube; through the linkage control of the vertical translation stage and the rotating platform, it can simulate the translational step-by-step method of CT. Scanning and spiral scanning; the light source is a switchable wavelength light source, used to simulate dual-energy CT or simulate different tube voltage adjustments with different wavelengths.

3) The intelligent control and reconstruction module has intelligent positioning function and dose intelligent control function. The intelligent positioning function can automatically move the area of interest to the imaging area through computer control of the translation stage, reducing manual operations and improving imaging efficiency. The dose intelligent control function can automatically adjust the intensity of the light source and the exposure time of the camera through the neural network model to control the irradiation dose to obtain the best imaging quality or to ensure the imaging quality while using a lower irradiation dose to solve the problem of relying on manual experience to adjust the dose. Precision and regulation of complex problems.

4) The display module of the present invention can simulate the projection diagram, sinogram and reconstructed image display of CT images. The invention successfully integrates the teaching demonstration of scanning operation, CT imaging principle and CT image display into a system.

Picture description:

Figure 1 is an overall schematic diagram of the present invention. (a) is the module diagram of the optical-based intelligent CT teaching simulation system (b) is the structural schematic diagram of the optical-based intelligent CT teaching simulation system.

Figure 2 Two structures of light source modules. (a) The light source propagates backward through the diffuse scattering sheet. (b) The light source is reflected by the diffuse reflector and propagates backward. The arrow indicates the direction of light propagation.

Figure 3 is a schematic diagram of the adjustable slit.

Figure 4 is a schematic diagram of the scanning module. The arrow indicates the direction of light propagation.

Figure 5 is a flow chart of two intelligent positioning functions.

Figure 6 is a dose intelligent control flow chart.

Detailed ways

In view of the shortcomings of the existing technology, the present invention provides an optical-based intelligent CT teaching simulation system to assist teaching and help students improve their understanding and research capabilities of CT imaging principles and new CT technologies. CT imaging uses X-rays to pass through the human body in a straight line and reconstructs the image to obtain a cross-sectional image by collecting X-ray projections from multiple directions. The present invention uses a light source (such as visible light) to pass through a phantom of a certain transparent material along a straight line to perform projection reconstruction, and its physical process and reconstruction process are consistent with CT imaging. Therefore, the present invention can intuitively simulate the CT projection and reconstruction process, and can assist students to deeply understand the principles of CT imaging.

An optical-based intelligent CT teaching simulation system includes a light source module, a scanning module, an acquisition module, an intelligent control and reconstruction module, and a display module.

The light source module includes a light source 1, a beam homogenizing device 2 (such as a diffuser/reflective sheet) and an adjustable slit 3. The adjustable slit 3 consists of four flat plates, upper, lower, left and right, and is used to adjust the size of the light-transmitting window to adjust the illumination field of view. Among them, light source 1 is a switchable wavelength light source, used to simulate dual-energy CT or simulate different tube voltage adjustments with different wavelengths. The light source module can adopt two structures. One is that the light source 1 passes through the diffuse scattering sheet 2 and then continues to propagate through the adjustable slit 3. One is that the light source 1 is reflected by the diffuse reflection sheet 2 and then continues to propagate through the adjustable slit 3 .

The scanning module includes a vertical translation stage 4, a rotating platform 5, an imaging phantom 6 and a water tank 7. The imaging phantom 6 is installed on the rotating platform 5 , and the rotating platform 5 is installed on the vertical translation stage 4 . The rotating platform 5 is used to rotate the imaging phantom 6 to perform multi-angle projection collection. The vertical translation platform 4 can move up and down with the rotating platform 5. When the vertical translation stage 4 does not move, the rotating platform 5 rotates, which can simulate the translation step scanning of CT. When the rotary translation stage 5 rotates, the vertical translation stage 4 moves vertically downward, which can simulate the spiral scanning of CT. The imaging phantom 6 is an imaging phantom that has a certain light absorption distribution and allows light to pass through.

The two surfaces of the water tank 7 (the two surfaces through which light propagates) are flat plates with light-transmitting windows.

The acquisition module includes an iris diaphragm 8, a lens 9 and a camera 10. The light emitted by the light source module passes through the imaging phantom 6 and then passes through the light-transmitting window of the water tank 7 , passes through the variable diaphragm 8 and the lens 9 in sequence, and is finally collected by the camera 10 .

The intelligent control and reconstruction module includes a computer 11. The rotating platform 5, the vertical translation platform 4 and the camera 10 are connected to the computer 11 through electrical connecting lines. On the one hand, the computer 11 uses the internal intelligent control module to control the rotation of the scanning module rotating platform 5 and the movement of the vertical translation stage 4, and to control the image acquisition of the camera 10. On the other hand, after the acquisition is completed, the computer 11 uses the corresponding reconstruction through the internal reconstruction module. Algorithm (for example, translation step scanning uses filtered back-projection algorithm, spiral scan first uses the projection image of the layer adjacent to the slice to be reconstructed to linearly interpolate to the reconstructed slice to obtain the projection image of the reconstructed slice, and then uses the filtered back-projection algorithm to reconstruct) The projection image collected by the camera 10 is reconstructed into a tomographic image.

In addition, the intelligent control module of the computer 11 also includes intelligent positioning and dosage intelligent control functions.

The display module is used to display projection images, sinograms and reconstructed image displays at each angle (such as cross-sectional display, maximum density projection display and three-dimensional display) to simulate the projection images, sinograms and reconstructed image displays of CT images. Use the corresponding reconstruction algorithm to reconstruct the projection image collected by the camera into a tomographic image. Specifically, a translation step scanning is used and a filtered back-projection algorithm is used. The spiral scan first uses the projection image of the layer adjacent to the slice to be reconstructed to linearly interpolate to the reconstruction. The projection image of the reconstructed slice is obtained, and then the filtered back-projection algorithm is used to reconstruct it.

There are two implementation processes for the intelligent positioning function: the first one: 1) The user sets the inspection site, which is the area of interest, in advance, 2) the camera 10 collects an image, 3) the computer 11 determines the sense of the collected image through image recognition and segmentation technology. area of interest, 4) the computer 11 calculates the upper and lower limits and center coordinates of the area of interest in the vertical direction, 5) if the area of interest is larger than the system imaging field of view, the computer 11 controls the vertical translation stage 4 to move the imaging object to the lower limit of the area of interest. The coordinate slightly higher than the lower limit coordinate of the system imaging field of view is used as the starting position of the imaging scan; if the area of interest is less than or equal to the system imaging field of view, the computer 11 controls the vertical translation stage 4 to move the imaging object so that the vertical center coordinate of the area of interest is equal to The coordinates of the center of the system's imaging field of view coincide with each other as the imaging scanning position. That is to realize the intelligent positioning function; the second type: 1) the camera 10 collects an image, 2) the user manually outlines the area of interest on the collected image, 3) the computer 11 calculates the upper and lower limits and center of the area of interest in the vertical direction coordinates, 4) If the area of interest is larger than the system imaging field of view, the computer 11 controls the vertical translation stage 4 to move the imaging object so that the lower limit coordinates of the area of interest are slightly higher than the lower limit coordinates of the system imaging field of view as the starting position of the imaging scan; if the area of interest is less than or equal to the system imaging field of view, then the computer 11 controls the vertical translation stage 4 to move the imaging object so that the vertical center coordinates of the area of interest coincide with the system imaging field of view center coordinates as the imaging scanning position. That is to realize the intelligent positioning function.

The dose intelligent control function can estimate the dose required for imaging based on the area to be imaged, thereby automatically controlling the selection of the appropriate intensity of the light source 1 or the exposure time of the camera 10 . The specific steps are: 1) The computer 11 learns a large amount of data through deep learning methods to train a neural network that maps a single projection image to a three-dimensional tomographic image, and determines the neural network model after training and verification. The input of the training data and verification data used in deep learning is a single projection image, and the output is the corresponding real three-dimensional tomographic image. The network used is a convolutional neural network, such as U-net network, etc. 2) The camera 10 collects a single projection image of the area to be imaged, and the computer 11 uses the neural network model learned previously to estimate the three-dimensional tomographic image corresponding to the area to be imaged. 3) The computer 11 uses the estimated three-dimensional tomographic image to estimate the dose required for imaging. 4) The computer 11 automatically controls the intensity of the light source 1 or the exposure time of the camera 10 based on the calculated and estimated dose.

The imaging process is as follows:

1) Fill the water tank 7 in the scanning module with water, and install the imaging phantom 6 on the rotating platform 5. The human or computer 11 controls the vertical translation stage 4 to move the imaging phantom 6 to a suitable imaging position through the intelligent positioning function of the intelligent control module.

2) Turn on light source 1 of the light source module. The light emitted by the light source 1 forms a relatively uniform beam after passing through the beam homogenizing device 2. After passing through the adjustable slit 3 to adjust the light size, it passes through the light window on the side of the water tank 7 and shines on the imaging phantom 6. The light passes through the phantom 6 and then passes through the light window on the other side of the water tank 7 , passes through the iris 8 in turn, and enters the camera 10 through the lens 9 . The intensity of the light source 1 or the exposure time of the camera 10 is controlled manually or by the computer 11 through the dose intelligent control function of the intelligent control module for subsequent projection collection.

3) For translational step scanning, there are two scanning modes. The first one is that the computer 11 controls the vertical translation stage 4 to remain stationary, and the rotating platform 5 drives the imaging phantom 6 to rotate. It stops at a certain angle every time it rotates, and is collected by the camera 10 The light emitted by the primary light source 1 passes through the projection of the imaging phantom 6, which is recorded as a projection at one angle. In this way, the projections from multiple angles are repeatedly collected. The rotating platform 5 returns to the initial angle when scanning is started and stops, and then the platform is translated vertically. 4 moves downward a certain distance and continues to repeat the above collection to complete the projection of other parts of the phantom 6 in the vertical direction; the second is that the computer 11 controls the vertical translation stage 4 to remain stationary, and the rotating platform 5 drives the imaging phantom 6 to rotate continuously. , the camera 10 collects projections from one angle at regular intervals until all projections from multiple angles are collected. The rotating platform 5 returns to the initial angle when scanning and stops, and then the vertical translation platform 4 moves downward for a certain distance, and continues to repeat the above collection to complete the projection of other parts of the phantom 6 in the vertical direction. For spiral scanning, the computer 11 controls that while the rotating platform 5 drives the imaging phantom 6 to rotate, the vertical translation stage 4 moves downward to collect projections from multiple angles through the camera 10. During the process of collecting projections, the projected image is displayed Display in the module to observe and monitor the status of the collected images in real time.

4) The computer 11 uses the multiple angle projections collected in 3) to perform reconstruction through the reconstruction module in the intelligent control and reconstruction module, and then displays the reconstruction result through the display module.

An optical-based intelligent CT teaching simulation system includes a light source module, a scanning module, an acquisition module, an intelligent control and reconstruction module, and a display module.

The light source module includes a light source 1, a beam homogenizing device 2 and an adjustable slit 3. The adjustable slit 3 consists of four flat plates, upper, lower, left and right, and is used to adjust the size of the light-transmitting window to adjust the illumination field of view. The light source can be an LED with multiple wavelengths, an LED array with multiple wavelengths, or a halogen lamp and a filter to obtain different emission wavelengths. The beam homogenizing device 2 can be a diffuse scattering sheet or a diffuse reflection sheet. The light source module can adopt two structures. One is that the light source 1 passes through the diffuse scattering sheet 2 and then continues to propagate through the adjustable slit 3. One is that the light source 1 is reflected by the diffuse reflection sheet 2 and then continues to propagate through the adjustable slit 3 .

The scanning module includes a vertical translation stage 4, a rotating platform 5, an imaging phantom 6 and a water tank 7. The imaging phantom 6 is installed on the rotating platform 5 , and the rotating platform 5 is installed on the vertical translation stage 4 . The imaging phantom 6 is an imaging phantom with a certain light absorption distribution that allows light to pass through, such as a transparent colloid partially containing dye.

The two surfaces of the water tank 7 (the two surfaces through which light propagates) are flat plates with light-transmitting windows.

The acquisition module includes an iris diaphragm 8, a lens 9 and a camera 10. The light emitted by the light source module passes through the imaging phantom 6 and then passes through the light-transmitting window of the water tank 7 , passes through the variable diaphragm 8 and the lens 9 in sequence, and is finally collected by the camera 10 .

The intelligent control and reconstruction module includes a computer 11. The rotating platform 5, the vertical translation platform 4 and the camera 10 are connected to the computer 11 through electrical connecting lines. On the one hand, the computer 11 uses the internal intelligent control module to control the rotation of the scanning module rotating platform 5 and the movement of the vertical translation stage 4, and to control the image acquisition of the camera 10. On the other hand, after the acquisition is completed, the computer 11 uses the corresponding reconstruction through the internal reconstruction module. Algorithm (for example, translation step scanning uses filtered back-projection algorithm, spiral scan first uses the projection image of the layer adjacent to the slice to be reconstructed to linearly interpolate to the reconstructed slice to obtain the projection image of the reconstructed slice, and then uses the filtered back-projection algorithm to reconstruct) The projection image collected by the camera 10 is reconstructed into a tomographic image.

The intelligent control module of the computer 11 also includes intelligent positioning and dosage intelligent control functions.

There are two implementation processes for the intelligent positioning function: the first one: 1) The user sets the inspection site, which is the area of interest, in advance, 2) the camera 10 collects an image, 3) the computer 11 determines the sense of the collected image through image recognition and segmentation technology. area of interest, 4) the computer 11 calculates the upper and lower limits and center coordinates of the area of interest in the vertical direction, 5) if the area of interest is larger than the system imaging field of view, the computer 11 controls the vertical translation stage 4 to move the imaging object to the lower limit of the area of interest. The coordinate slightly higher than the lower limit coordinate of the system imaging field of view is used as the starting position of the imaging scan; if the area of interest is less than or equal to the system imaging field of view, the computer 11 controls the vertical translation stage 4 to move the imaging object so that the vertical center coordinate of the area of interest is equal to The coordinates of the center of the system's imaging field of view coincide with each other as the imaging scanning position. That is to realize the intelligent positioning function; the second type: 1) the camera 10 collects an image, 2) the user manually outlines the area of interest on the collected image, 3) the computer 11 calculates the upper and lower limits and center of the area of interest in the vertical direction coordinates, 4) If the area of interest is larger than the system imaging field of view, the computer 11 controls the vertical translation stage 4 to move the imaging object so that the lower limit coordinates of the area of interest are slightly higher than the lower limit coordinates of the system imaging field of view as the starting position of the imaging scan; if the area of interest is less than or equal to the system imaging field of view, then the computer 11 controls the vertical translation stage 4 to move the imaging object so that the vertical center coordinates of the area of interest coincide with the system imaging field of view center coordinates as the imaging scanning position. That is to realize the intelligent positioning function.

The dose intelligent control function can estimate the dose required for imaging based on the area to be imaged, thereby automatically controlling the selection of the appropriate intensity of the light source 1 or the exposure time of the camera 10 . The specific steps are: 1) The computer 11 learns a large amount of data through deep learning methods to train a neural network that maps a single projection image to a three-dimensional tomographic image, and determines the neural network model after training and verification. The input of the training data and verification data used in deep learning is a single projection image, and the output is the corresponding real three-dimensional tomographic image. The network used is a convolutional neural network, such as U-net network, etc. 2) The camera 10 collects a single projection image of the area to be imaged, and the computer 11 uses the neural network model learned previously to estimate the three-dimensional tomographic image corresponding to the area to be imaged. 3) The computer 11 uses the estimated three-dimensional tomographic image to estimate the dose required for imaging, that is, the light intensity of the light source 1 with a fixed exposure time or the exposure time of the camera 10 with a fixed light source intensity. Example of calculation method: Assume that the relationship between the light intensity I ₂ incident on the camera 10 and the response value I ₃ of the camera 10 is I ₃ =I ₂ *D*E, where E is the exposure time and D is I ₂ and I ₃ within the unit exposure time. The transformation coefficient is a constant and can be measured experimentally. The relationship between the light intensity I ₁ of the light source 1 and the light intensity passing through the phantom 6 , that is, the light intensity I ₂ incident on the camera 10 is I ₂ =I ₁ *(M*C _i ), where M is the light intensity of the light source 1 The attenuation before the incident phantom 6 is a constant and can be measured. C _i is the projection attenuation of the light intensity at various angles after passing through the phantom 6 calculated using the estimated three-dimensional tomographic image ((i=1, 2, ..., n, n is the projection angle). Then the relationship between the light intensity I ₁ of the light source 1 and the response value I ₃ of the camera 10 is I ₁ =I ₂ /(M*C _i )=I ₃ /(D*E)/ (M*C _i )=I ₃ /(D*E*M*C _i ). If you want to obtain better imaging quality, when the exposure time E is constant, the maximum value of the light intensity I ₁ of the light source 1 should be less than I _3max /(D*E*M*C _max ), where I _3max is the maximum value that I ₃ can take, and C _max is the maximum value of C _i , that is, the corresponding light intensity attenuation of the incident light passing through the phantom 6 is minimum. Or when When the light intensity I ₁ of light source 1 is constant, the exposure time E should be less than I _3max / (D*I ₁ *M*C _max ). If you want to use a lower dose, when the exposure time E is constant, the light intensity I 1 of light source ₁ The maximum value should be less than I _3max / (D*E*M*C _max ) and the value should be slightly greater than I _3min / (D*E*M*C _min ), where I _3min is the camera without light when the exposure time is E The image noise value at input, C _min is the minimum value of C _i , and the corresponding light intensity attenuation of the incident light passing through the phantom 6 is the largest. Or when the light intensity I ₁ of the light source 1 is constant, the exposure time E should be less than I _3max / The simultaneous value of (D*I ₁ *M*C _max ) is slightly greater than I _3min / (D*I ₁ *M*C _min ). 4) The computer 11 automatically controls the intensity of the light source 1 or the camera according to the calculated estimated dose 10 exposure times.

The display module is used to display projection images, sinograms and reconstructed image displays at each angle (such as cross-sectional display, maximum density projection display and three-dimensional display) to simulate the projection images, sinograms and reconstructed image displays of CT images.

The imaging process is as follows:

1) Fill the water tank 7 in the scanning module with water, and install the imaging phantom 6 on the rotating platform 5. The human or computer 11 controls the vertical translation stage 4 to move the imaging phantom 6 to a suitable imaging position through the intelligent positioning function of the intelligent control module.

2) Turn on light source 1 of the light source module. The light emitted by the light source 1 forms a relatively uniform beam after passing through the beam homogenizing device 2. After passing through the adjustable slit 3 to adjust the light size, it passes through the light window on the side of the water tank 7 and shines on the imaging phantom 6. The light passes through the phantom 6 and then passes through the light window on the other side of the water tank 7 , passes through the iris 8 in turn, and enters the camera 10 through the lens 9 . The intensity of the light source 1 or the exposure time of the camera 10 is controlled manually or by the computer 11 through the dose intelligent control function of the intelligent control module for subsequent projection collection.

3) For translational step scanning, there are two scanning modes. The first one is that the computer 11 controls the vertical translation stage 4 to remain stationary, and the rotating platform 5 drives the imaging phantom 6 to rotate. It stops at a certain angle every time it rotates, and is collected by the camera 10 The light emitted by the primary light source 1 passes through the projection of the imaging phantom 6, which is recorded as a projection at one angle. In this way, the projections from multiple angles are repeatedly collected. The rotating platform 5 returns to the initial angle when scanning is started and stops, and then the platform is translated vertically. 4 moves downward a certain distance and continues to repeat the above collection to complete the projection of other parts of the phantom 6 in the vertical direction; the second is that the computer 11 controls the vertical translation stage 4 to remain stationary, and the rotating platform 5 drives the imaging phantom 6 to rotate continuously. , the camera 10 collects projections from one angle at regular intervals until all projections from multiple angles are collected. The rotating platform 5 returns to the initial angle when scanning and stops, and then the vertical translation platform 4 moves downward for a certain distance, and continues to repeat the above collection to complete the projection of other parts of the phantom 6 in the vertical direction. For spiral scanning, the computer 11 controls that while the rotating platform 5 drives the imaging phantom 6 to rotate, the vertical translation stage 4 moves downward to collect projections from multiple angles through the camera 10. During the process of collecting projections, the projected image is displayed Display in the module to observe and monitor the status of the collected images in real time.

4) The computer 11 uses the multiple angle projections collected in 3) to perform reconstruction through the reconstruction module in the intelligent control and reconstruction module, and then displays the reconstruction result through the display module.

The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present invention. All are covered by the protection scope of the present invention.

Claims (4)

1. An optical-based intelligent CT teaching simulation system, characterized by including a light source module, a scanning module, an acquisition module, an intelligent control and reconstruction module and a display module; The light source module includes a light source, a beam homogenizing device and an adjustable slit. The adjustable slit consists of four flat plates: upper, lower, left and right. It is used to adjust the size of the light-transmitting window to adjust the illumination field of view; among them, the light source is The switchable wavelength light source is used to simulate dual-energy CT or use different wavelengths to simulate different tube voltage adjustments. The light source module adopts one of the following two structures: one is that the light source continues to propagate through the adjustable slit after passing through the diffuse scattering sheet; One is that the light source continues to propagate through the adjustable slit after being reflected by the diffuse reflector; The scanning module includes a vertical translation stage, a rotating platform, an imaging phantom and a water tank. The imaging phantom is installed on the rotating platform, and the rotating platform is installed on the vertical translation stage. The rotating platform is used to rotate the imaging phantom for multi-angle projection. During acquisition, the vertical translation stage moves up and down with the rotating platform. When the vertical translation stage does not move, the rotating platform rotates to simulate the translational step-by-step scanning of CT. When the rotating translation stage rotates, the vertical translation stage moves vertically. Move downward to simulate the spiral scanning of CT. The imaging phantom is an imaging phantom with a certain light absorption distribution that allows light to pass through; The two surfaces through which light propagates through the water tank are flat plates with light-transmitting windows; The collection module includes an iris, a lens and a camera. The light emitted by the light source module passes through the imaging phantom, passes through the light-transmitting window of the water tank, passes through the iris and lens in turn, and is finally collected by the camera; The intelligent control and reconstruction module includes a computer. The rotating platform, vertical translation stage and camera are connected to the computer through electrical connecting lines. On the one hand, the computer uses the internal intelligent control module to control the rotation of the scanning module's rotating platform and the movement of the vertical translation stage, and Control the image acquisition of the camera. On the other hand, after the acquisition is completed, the internal reconstruction module uses the corresponding reconstruction algorithm to reconstruct the projection image collected by the camera into a tomographic image; The computer's intelligent control module is also used for intelligent positioning and dosage intelligent control; The display module is used to display the projection image, sinogram and reconstructed image display at each angle to simulate the projection image, sinogram and reconstructed image display of CT images; The dose intelligent control function estimates the dose required for imaging based on the area to be imaged, thereby automatically controlling the intensity of the appropriate light source or the exposure time of the camera. The detailed steps are: 1) The computer learns a large amount of data through deep learning methods to train a neural network that maps a single projection map to a three-dimensional tomographic image. After training and verification, the neural network model is determined. The inputs of training data and verification data used in deep learning are both A single projection image is output as the corresponding real three-dimensional tomographic image; 2) The camera collects a single projection image of the area to be imaged, and the computer uses the neural network model learned previously to estimate the three-dimensional tomographic image corresponding to the area to be imaged; 3) The computer uses the estimated three-dimensional tomographic image to estimate the dose required for imaging, that is, the light intensity of the light source with a fixed exposure time or the exposure time of the camera with a fixed light source intensity. For the light intensity I ₂ incident on the camera, the response of the camera The relationship between numerical value I ₃ is I ₃ =I ₂ *D*E, where E is the exposure time, D is the conversion coefficient of I ₂ and I ₃ within the unit exposure time, the light intensity I ₁ of the light source and the light intensity passing through the phantom , that is, the relationship between the light intensity I ₂ incident on the camera is I ₂ =I ₁ *(M*C _i ), where M is the attenuation of the light intensity from the light source to the incident phantom, and C _i is the estimated three-dimensional tomographic image Calculate the projection attenuation of the light intensity at each angle after the phantom, i = 1, 2,..., n, n is the projection angle, then the relationship between the light source intensity I ₁ and the camera's response value I ₃ is I ₁ =I ₂ /(M*C _i )=I ₃ /(D*E)/(M*C _i )=I ₃ /(D*E*M*C _i ). In the first case, good imaging quality can be obtained. For the purpose, when the exposure time E is constant, the maximum value of the light source intensity I ₁ should be less than I _3max / (D*E*M*C _max ), where I _3max is the maximum value that I ₃ can take, and C _max is C The maximum value of _i means that the corresponding light intensity attenuation of the incident light passing through the phantom is minimum, or when the light source intensity I ₁ is constant, the exposure time E should be less than I _3max / (D*I ₁ *M*C _max ); In the second case, when the purpose is to use a lower dose, when the exposure time E is constant, the maximum value of the light source intensity I ₁ should be less than I _3max / (D*E*M*C _max ) while the value is greater than I _3min /(D*E*M*C _min ), where I _3min is the image noise value when the camera has no light input when the exposure time is E, C _min is the minimum value of C _i , and the corresponding incident light passes through the phantom _{The light intensity of _ _} *C _min ); 4) The computer automatically controls the intensity of the light source or the exposure time of the camera based on the calculated and estimated dose. 2. The optical-based intelligent CT teaching simulation system as claimed in claim 1, characterized in that the intelligent positioning function has two implementation processes: the first one: 1) the user sets the inspection site, that is, the area of interest, in advance, 2) The camera collects an image, 3) determines the area of interest in the collected image through image recognition and segmentation technology, 4) calculates the upper and lower limits and center coordinates of the area of interest in the vertical direction, 5) if the area of interest is larger than the system imaging field of view, Then control the vertical translation stage to move the imaging object so that the lower limit coordinate of the area of interest is slightly higher than the lower limit coordinate of the system imaging field of view as the starting position of the imaging scan; if the area of interest is less than or equal to the system imaging field of view, control the vertical translation stage to move the imaging object Make the vertical center coordinates of the area of interest coincide with the center coordinates of the system imaging field of view as the imaging scanning position, that is, realize the intelligent positioning function; second type: 1) the camera collects an image, 2) the user manually outlines and determines the collected image Region of interest, 3) Calculate the upper and lower limits and center coordinates of the region of interest in the vertical direction, 4) If the region of interest is larger than the system imaging field of view, control the vertical translation stage to move the imaging object so that the lower limit coordinates of the region of interest are slightly higher than The lower limit coordinates of the system imaging field of view are used as the starting position of the imaging scan; if the area of interest is less than or equal to the system imaging field of view, the vertical translation stage is controlled to move the imaging object so that the vertical center coordinates of the area of interest coincide with the center coordinates of the system imaging field of view as Imaging scans the position to achieve intelligent positioning function. 3. The optical-based intelligent CT teaching simulation system according to claim 1, characterized in that the neural network used is a convolutional neural network. 4. The optical-based intelligent CT teaching simulation system as claimed in claim 1, characterized in that the imaging process is as follows: 1) Fill the water tank in the scanning module with water, install the imaging phantom on the rotating platform, and manually or by computer control the vertical translation stage through the intelligent positioning function of the intelligent control module to move the imaging phantom to the appropriate imaging position; 2) Turn on the light source of the light source module. The light emitted by the light source passes through the beam homogenizing device to form a relatively uniform beam. After adjusting the light size through the adjustable slit, it passes through the light window on the side of the water tank and shines on the imaging phantom. The light passes through the phantom and then passes through the light window on the other side of the water tank. It passes through the iris and the lens and enters the camera. The intensity of the light source or the exposure of the camera is controlled manually or by computer through the dose intelligent control function of the intelligent control module. time for subsequent projection acquisition; 3) For translational step-by-step scanning, there are two scanning modes. The first is that the computer controls the vertical translation stage to remain stationary. The rotating platform drives the imaging phantom to rotate. It stops at a certain angle every time it rotates, and the light emitted by the light source is collected through the camera. The projection through the imaging phantom is recorded as the projection at one angle. In this way, the projections from multiple angles are repeatedly collected. The rotating platform returns to the initial angle when scanning is started and stops. Then the vertical translation platform moves downward for a certain distance and continues to repeat. The above collection is to complete the projection of other parts of the phantom in the vertical direction; the second method is to control the vertical translation stage stationary by the computer, and the rotating platform drives the imaging phantom to rotate continuously, and the camera collects projections at an angle at certain intervals until the acquisition After completing the projection of multiple angles, the rotating platform returns to the initial angle when scanning and stops, and then the vertical translation stage moves downward for a certain distance, and continues to repeat the above collection to complete the projection of other parts of the phantom in the vertical direction; for the spiral Scanning, computer control drives the rotating platform to rotate the imaging phantom, and at the same time, the vertical translation stage moves downward, and the projections from multiple angles are collected through the camera. During the process of collecting projections, the projected images are displayed on the display module for real-time observation. and monitor the status of collected images; 4) The computer uses the multiple angle projections collected in 3) to perform reconstruction through the reconstruction module in the intelligent control and reconstruction module, and then displays the reconstruction results through the display module.