Osteoporosis detection system based on nuclear magnetic resonance

Abstract

The invention discloses an osteoporosis detection system based on nuclear magnetic resonance, and relates to the technical field of intelligent detection, the osteoporosis detection system comprises a pre-operation module, an image processing module, an analysis module and a precision module, an image segmentation model is trained to obtain an intervertebral disc image, the preprocessed image is segmented to obtain a region-of-interest image, and the region-of-interest image is obtained; and judging a first osteoporosis level of the region of interest, inputting the region of interest image to the expert platform to obtain a second osteoporosis level, and executing a second operation. Through cooperative work of the pre-operation module, the image processing module, the analysis module and the precision module, the detection precision and efficiency are improved, through the image segmentation model and the analysis module, more comprehensive osteoporosis evaluation is provided, an expert platform is introduced, the accuracy and reliability of a detection result are ensured, and the detection efficiency is improved. The MRI technology is used for osteoporosis detection, radiation exposure is avoided, and the method is a non-invasive and safe detection method.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent detection, and in particular to an osteoporosis detection system based on nuclear magnetic resonance. Background Art

[0002] In recent years, the osteoporosis detection technology based on nuclear magnetic resonance is constantly developing and improving, and has broad application prospects. MRI technology has clinical value in the early diagnosis of osteoporosis, the prediction of fracture risk and the follow-up after treatment. MRI technology has the advantages of being non-invasive, radiation-free and having high tissue resolution. Compared with the assessment of osteoporosis based solely on bone mineral density (BMD), MRI technology can assess factors such as bone marrow composition, microcirculation, and bone microstructure, thereby more comprehensively reflecting bone strength.

[0003] At present, a Chinese invention patent with publication number CN119014815A discloses an osteoporosis detection method based on multi-mode photoacoustics and a detection system thereof. The method establishes an inversion model of the physical properties of bone tissue based on an inverse solution model, inputs the bone measurement photoacoustic wave signal parameters into the inversion model, outputs the measured bone tissue physical property information data, inputs the osteoporosis clinical standard data, compares the osteoporosis clinical standard data with the measurement data, and gives a bone diagnosis result. However, the related technology does not establish and analyze the model based on historical detection data, lacks the scientific nature of auxiliary diagnosis, does not eliminate invalid data, and does not eliminate the errors of the acquisition part, which is easy to cause wrong judgments, lacks the accuracy of judgment, does not identify and delete redundant data, is not conducive to the rapid responsiveness of the system, and has certain limitations. Summary of the invention

[0004] The technical problem solved by the present invention is: the related technology does not establish and analyze the model based on historical detection data, lacks the scientific nature of auxiliary diagnosis, does not eliminate invalid data and eliminate errors in the acquisition part, which easily leads to wrong judgments and lacks the accuracy of judgments. It does not identify and delete redundant data, is not conducive to the rapid responsiveness of the system, and has certain limitations.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: an osteoporosis detection system based on nuclear magnetic resonance, comprising a pre-operation module, an image processing module, an analysis module and a precision module; The pre-operation module sets scanning parameters, and sets a positioning line and a positioning frame, identifies the subject's posture, performs a first operation according to the subject's posture, obtains a historical scan image, identifies a spinal part in the historical scan image, marks feature points in the spinal part according to a spatial order, obtains a spinal cord curve image according to the feature points and Hough transform, crops a corresponding historical intervertebral disc region image according to the spinal cord curve image, trains an image segmentation model according to the historical scan image and the historical intervertebral disc region image, and obtains an intervertebral disc image according to the image segmentation model; The image processing module obtains the image after the first operation is completed, preprocesses the image, inputs the preprocessed image into the image segmentation model to obtain an intervertebral disc region image, segments the preprocessed image according to the intervertebral disc region image to obtain an image of a region of interest, and numbers the image of the region of interest; The analysis module obtains relevant data of the region of interest, sets a relevant data threshold, determines a first osteoporosis grade of the region of interest according to the relevant data threshold, and sets a label of the first osteoporosis grade according to a corresponding number; The precise module calls the expert platform, inputs the image of the region of interest to the expert platform, obtains a second osteoporosis grade, compares the second osteoporosis grade with the first osteoporosis grade, and performs a second operation according to the comparison result.

[0006] As a preferred embodiment of the osteoporosis detection system based on nuclear magnetic resonance described in the present invention, the scanning parameters include radio frequency pulse interval time, echo time, layer thickness, layer spacing, matrix size and scanning time, and the scanning parameters are obtained according to historical detection experience.

[0007] As a preferred solution of the osteoporosis detection system based on nuclear magnetic resonance of the present invention, the setting logic of the positioning line and the positioning frame includes: Acquire a bottom surface image of the internal space of the device, draw a center line of the bottom surface of the internal space, the center line is represented by a line connecting the midpoints of the width of the ground, set the center line as a positioning line, draw a plane passing through the center line and perpendicular to the ground image, and record it as a first plane; Get the height inside the device, set the height of the bottom surface as the starting height, set the height inside the device w as the ending height, draw a plane that is parallel to the bottom surface, perpendicular to the first plane and does not pass through the first plane, record it as the second plane, and set the second plane as the positioning box.

[0008] As a preferred solution of the osteoporosis detection system based on nuclear magnetic resonance of the present invention, wherein: the first operation includes sending a posture correction signal and sending a start device signal, and the posture correction signal includes a translation correction signal and a rotation correction signal; When the device senses an obstruction, it uses machine vision to obtain the subject's image and identify the subject's posture. When the subject is lying flat, the positioning operation is started; In response to starting the positioning operation, starting a first detection, the detection logic of the first detection includes: Identify the edge line of the subject's image, divide the image inside the edge line parallel to the positioning line, calculate the number of pixels on both sides of the positioning line that belong to the edge line, calculate the difference in the number of pixels on both sides of the positioning line that belong to the edge line, record it as the first difference, set the first value as the first difference threshold, when the first difference is less than or equal to the first value, start the second detection, position the subject up and down according to the positioning frame, when the first difference is greater than or equal to the first value, send a translation correction signal, and the subject performs horizontal translation according to the translation posture correction signal until the first difference is less than the first value, and then stop sending the translation correction signal.

[0009] As a preferred solution of the osteoporosis detection system based on nuclear magnetic resonance of the present invention, the detection logic of the second detection includes: Identify the height of the subject, and obtain the subject's height corresponding to two points symmetrical to the positioning line and distributed on the positioning frame; When the two points on the positioning line are symmetrical and the corresponding heights of the subject distributed on the positioning frame are the same, a start scanning signal is sent; when the two points on the positioning line are symmetrical and the corresponding heights of the subject distributed on the positioning frame are different, a rotation correction signal is sent, and the subject flips the angle according to the rotation correction signal until the two points on the positioning line are symmetrical and the corresponding heights of the subject distributed on the positioning frame are the same, and then the rotation correction signal is stopped.

[0010] As a preferred solution of the osteoporosis detection system based on nuclear magnetic resonance of the present invention, the recognition logic of the spine part includes: Retrieve a human spine reference image, extract the shape feature of the human spine reference image, record it as the first feature, extract the shape feature of the historical scan image, record it as the second feature, calculate the similarity between the first feature and the second feature by using the cosine similarity formula, set the second value as the similarity threshold, and when the similarity is greater than or equal to the second value, set the corresponding part in the historical scan image as the spine part, otherwise, jump to the second feature of the next part; The cropping logic for the historical disc region images includes: Marking feature points in the spine portion according to a spatial order, wherein the spatial order is represented from top to bottom in a plan view; A spinal cord curve image is obtained according to the feature points and Hough transform, and the amplitude of the spinal cord curve image is obtained; The third value and the fourth value are set as amplitude thresholds, a portion of the spinal cord curve corresponding to an amplitude distribution between the third value and the fourth value is set as a historical intervertebral disc curve portion, a portion of the historical scan image corresponding to the portion of the historical intervertebral disc curve is set as a historical intervertebral disc region, and the corresponding historical intervertebral disc region image is cropped; An image segmentation model is trained according to historical scanned images and historical intervertebral disc region images, and an intervertebral disc image is obtained according to the image segmentation model.

[0011] As a preferred solution of the osteoporosis detection system based on nuclear magnetic resonance described in the present invention, the training logic of the image segmentation model includes: Using historical scanned images and historical intervertebral disc region images as sample sets, dividing the sample sets into training sets and test sets according to a first ratio, wherein the first ratio is obtained through training experience; Inputting the training set into the neural network for training to obtain an image segmentation model, and testing the image segmentation model according to the test set until the accuracy reaches a first accuracy, and then stopping the testing; The image segmentation model uses the scanned image as input, the intervertebral disc region image as output, the actual intervertebral disc region image as training target, and the distance between the geometric center of the intervertebral disc region image and the geometric center of the actual intervertebral disc region image as the accuracy standard.

[0012] As a preferred solution of the osteoporosis detection system based on nuclear magnetic resonance of the present invention, the preprocessing includes contrast enhancement processing and denoising processing, the preprocessed image is input into the image segmentation model to obtain the intervertebral disc region image, the intervertebral disc region image is deleted from the preprocessed image, and the greatest common factor of the boundary line of the deleted image is used as the side length of the segmentation frame, the segmentation frame is a square, and the preprocessed image is segmented from the upper left corner of the deleted image according to the segmentation frame to obtain the region of interest image; The region of interest images are numbered, and the numbers of the region of interest images are natural numbers.

[0013] As a preferred solution of the nuclear magnetic resonance-based osteoporosis detection system of the present invention, wherein: the analysis module obtains relevant data of the region of interest, and the relevant data includes bone marrow fat fraction and fat area ratio; The fifth value and the sixth value are set as relevant data thresholds, and the first osteoporosis grade of the region of interest is determined according to the relevant data thresholds. The first osteoporosis grade includes an automatic first grade, an automatic second grade, and an automatic third grade. The determination logic of the first osteoporosis grade is: When the bone marrow fat score is less than or equal to the fifth value, and the fat area ratio is less than or equal to the sixth value, the first osteoporosis grade is set to the automatic first grade; when the bone marrow fat score is greater than the fifth value, and the fat area ratio is less than or equal to the sixth value, the first osteoporosis grade is set to the automatic second grade; when the bone marrow fat score is greater than the fifth value, and the fat area ratio is greater than the sixth value, the first osteoporosis grade is set to the automatic third grade; The severity of osteoporosis represented by the automatic first grade, the automatic second grade and the automatic third grade is in an increasing trend.

[0014] As a preferred solution of the osteoporosis detection system based on nuclear magnetic resonance of the present invention, wherein: the precision module calls the expert platform, inputs the image of the region of interest to the expert platform, and obtains the second osteoporosis grade, and the second osteoporosis grade includes the artificial first grade, the artificial second grade and the artificial third grade; Comparing the second osteoporosis level with the first osteoporosis level to obtain a comparison result, and performing a second operation according to the comparison result, wherein the second operation includes adjusting a threshold of relevant data and sending a detection completion signal; When the first osteoporosis level is the same as the second osteoporosis level, the second operation is set to send a detection completion signal and output a corresponding number; when the first osteoporosis level is different from the second osteoporosis level, the second operation is set to adjust the threshold of the relevant data; When the second operation is to adjust the relevant data threshold, the seventh value and the eighth value are set as the changing variables of the fifth value and the sixth value respectively, the fifth value and the sixth value are continuously increased or decreased, and the corresponding first osteoporosis level is continuously obtained until the adjusted first osteoporosis level is the same as the second osteoporosis level, and then the adjustment of the relevant data threshold is stopped.

[0015] The beneficial effects of the present invention are as follows: through the collaborative work of the pre-operation module, the image processing module, the analysis module and the precision module, an automated process from scanning parameter setting to image segmentation and data analysis is realized, the accuracy and efficiency of the detection are improved, and personalized detection can be performed according to the subject's posture and historical scan images to ensure the accuracy of the detection results. By identifying the subject's posture and spinal part, the system can more accurately locate and analyze the intervertebral disc area. Through the image segmentation model and the analysis module, the system can obtain relevant data of the region of interest and set relevant data thresholds to determine the osteoporosis grade, providing a more comprehensive osteoporosis assessment. The precision module calls the expert platform, inputs the image of the region of interest, obtains the second osteoporosis grade, and compares the second osteoporosis grade with the first osteoporosis grade to ensure the accuracy and reliability of the detection results. The use of MRI technology for osteoporosis detection avoids radiation exposure, and is a non-invasive and safe detection method. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of the basic flow of an osteoporosis detection system based on nuclear magnetic resonance is provided for one embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0018] Example, see Figure 1 , as an embodiment of the present invention, provides an osteoporosis detection system based on nuclear magnetic resonance, including a pre-operation module, an image processing module, an analysis module and a precision module; The pre-operation module sets scanning parameters, and sets a positioning line and a positioning frame, identifies the subject's posture, performs a first operation according to the subject's posture, obtains a historical scan image, identifies a spinal part in the historical scan image, marks feature points in the spinal part according to a spatial order, obtains a spinal cord curve image according to the feature points and Hough transform, crops a corresponding historical intervertebral disc region image according to the spinal cord curve image, trains an image segmentation model according to the historical scan image and the historical intervertebral disc region image, and obtains an intervertebral disc image according to the image segmentation model; The image processing module obtains the image after the first operation is completed, preprocesses the image, inputs the preprocessed image into the image segmentation model to obtain an intervertebral disc region image, segments the preprocessed image according to the intervertebral disc region image to obtain an image of a region of interest, and numbers the image of the region of interest; The analysis module obtains relevant data of the region of interest, sets a relevant data threshold, determines a first osteoporosis grade of the region of interest according to the relevant data threshold, and sets a label of the first osteoporosis grade according to a corresponding number; The precise module calls the expert platform, inputs the image of the region of interest to the expert platform, obtains a second osteoporosis grade, compares the second osteoporosis grade with the first osteoporosis grade, and performs a second operation according to the comparison result.

[0019] The present invention realizes an automated process from scanning parameter setting to image segmentation and data analysis through the collaborative work of a pre-operation module, an image processing module, an analysis module and a precision module, thereby improving the accuracy and efficiency of detection, and can perform personalized detection according to the subject's posture and historical scan images to ensure the accuracy of the detection results. By identifying the subject's posture and spinal part, the system can more accurately locate and analyze the intervertebral disc area. Through the image segmentation model and the analysis module, the system can obtain relevant data of the region of interest and set relevant data thresholds to determine the osteoporosis grade, providing a more comprehensive osteoporosis assessment. The precision module calls the expert platform, inputs the image of the region of interest, obtains the second osteoporosis grade, and compares the second osteoporosis grade with the first osteoporosis grade to ensure the accuracy and reliability of the detection results. The use of MRI technology for osteoporosis detection avoids radiation exposure, and is a non-invasive and safe detection method.

[0020] The scanning parameters include radio frequency pulse interval time, echo time, layer thickness, layer spacing, matrix size and scanning time, and the scanning parameters are obtained according to historical detection experience.

[0021] In the specific implementation, fast spin echo sagittal T1-weighted imaging (TR 645ms), fast spin echo sagittal T2WI (TR 3500ms), fast spin echo sagittal FS-T2WI (TR 3000ms), fast spin echo sagittal T1-weighted imaging (TE 9.6ms), fast spin echo sagittal T2WI (TE 127ms), fast spin echo sagittal FS-T2WI (TE87ms) were performed. The slice thickness was set to 4mm, the slice spacing was set to 1mm, the matrix was set to 300×300, the T1WI sagittal scan time was 1 minute and 9 seconds, and the T2WI sagittal scan time was 1 minute and 19 seconds.

[0022] The setting logic of the positioning line and positioning frame settings includes: Acquire a bottom surface image of the internal space of the device, draw a center line of the bottom surface of the internal space, the center line is represented by a line connecting the midpoints of the width of the ground, set the center line as a positioning line, draw a plane passing through the center line and perpendicular to the ground image, and record it as a first plane; Get the height inside the device, set the height of the bottom surface as the starting height, set the height inside the device w as the ending height, draw a plane that is parallel to the bottom surface, perpendicular to the first plane and does not pass through the first plane, record it as the second plane, and set the second plane as the positioning box.

[0023] In specific implementation, by obtaining the bottom surface image of the internal space of the equipment and drawing the center line, it is ensured that the positioning line accurately reflects the midpoint of the ground width. This precise positioning helps to ensure the symmetry and consistency of the scan. The logic of setting the positioning line and the positioning frame simplifies the operation process, allowing technicians to quickly and accurately set the scanning parameters, reducing operation time and potential errors. By drawing a second plane parallel to the bottom surface and perpendicular to the first plane as the positioning frame, the accuracy and integrity of the scan area are ensured. This setting helps to improve scanning efficiency and ensure that all necessary areas are covered. By ensuring that the positioning frame does not pass through the first plane, it reduces the errors and artifacts that may occur during the scanning process, thereby improving the accuracy and reliability of the image.

[0024] The first operation includes sending a posture correction signal and sending a start device signal, wherein the posture correction signal includes a translation correction signal and a rotation correction signal; When the device senses an obstruction, it uses machine vision to obtain the subject's image and identify the subject's posture. When the subject is lying flat, the positioning operation is started; In response to starting the positioning operation, starting a first detection, the detection logic of the first detection includes: Identify the edge line of the subject's image, divide the image inside the edge line parallel to the positioning line, calculate the number of pixels on both sides of the positioning line that belong to the edge line, calculate the difference in the number of pixels on both sides of the positioning line that belong to the edge line, record it as the first difference, set the first value as the first difference threshold, when the first difference is less than or equal to the first value, start the second detection, position the subject up and down according to the positioning frame, when the first difference is greater than or equal to the first value, send a translation correction signal, and the subject performs horizontal translation according to the translation posture correction signal until the first difference is less than the first value, and then stop sending the translation correction signal.

[0025] In a specific implementation, by sending posture correction signals, including translation correction signals and rotation correction signals, the system can accurately correct the posture of the subject. This correction ensures the correct posture of the subject during the scanning process, thereby improving the accuracy and reliability of the scan. When the device senses an obstruction, the subject's image is acquired and the subject's posture is identified through machine vision. If the subject is lying flat, the positioning operation is started. This automated positioning operation reduces manual intervention and improves the efficiency and accuracy of the operation. The detection logic of the first detection includes identifying the edge line of the subject's image and dividing the image inside the edge line parallel to the positioning line. By calculating the number of pixels on both sides of the positioning line that belong to the edge line and their difference, the system can effectively determine whether further detection or correction is needed. This efficient detection logic ensures the speed and accuracy of the detection process.

[0026] The detection logic of the second detection includes: Identify the height of the subject, and obtain the subject's height corresponding to two points symmetrical to the positioning line and distributed on the positioning frame; When the two points on the positioning line are symmetrical and the corresponding heights of the subject distributed on the positioning frame are the same, a start scanning signal is sent; when the two points on the positioning line are symmetrical and the corresponding heights of the subject distributed on the positioning frame are different, a rotation correction signal is sent, and the subject flips the angle according to the rotation correction signal until the two points on the positioning line are symmetrical and the corresponding heights of the subject distributed on the positioning frame are the same, and then the rotation correction signal is stopped.

[0027] In a specific implementation, by identifying the height of the subject and obtaining two symmetrical points on the positioning line, the system can accurately determine the position of the subject in the positioning frame. This precise height recognition helps to ensure the accuracy of the scan. The detection logic ensures the symmetry of the subject by comparing the height of the subject corresponding to the two symmetrical points on the positioning line on the positioning frame. This symmetry correction is a key step to ensure the accuracy of the scan results. When the two symmetrical points on the positioning line and the corresponding subject heights distributed on the positioning frame are the same, the system will automatically send a scan start signal. This automated scan start reduces manual intervention and improves scanning efficiency. When the two symmetrical points on the positioning line and the corresponding subject heights distributed on the positioning frame are different, the system will send a rotation correction signal, and the subject will perform flip angle correction according to the signal. This dynamic rotation correction mechanism ensures continuous correction of the subject's posture until the optimal scanning state is achieved.

[0028] The recognition logic of the spine includes: Retrieve a human spine reference image, extract the shape feature of the human spine reference image, record it as the first feature, extract the shape feature of the historical scan image, record it as the second feature, calculate the similarity between the first feature and the second feature by using the cosine similarity formula, set the second value as the similarity threshold, and when the similarity is greater than or equal to the second value, set the corresponding part in the historical scan image as the spine part, otherwise, jump to the second feature of the next part; The cropping logic for the historical disc region images includes: Marking feature points in the spine portion according to a spatial order, wherein the spatial order is represented from top to bottom in a plan view; A spinal cord curve image is obtained according to the feature points and Hough transform, and the amplitude of the spinal cord curve image is obtained; The third value and the fourth value are set as amplitude thresholds, a portion of the spinal cord curve corresponding to an amplitude distribution between the third value and the fourth value is set as a historical intervertebral disc curve portion, a portion of the historical scan image corresponding to the portion of the historical intervertebral disc curve is set as a historical intervertebral disc region, and the corresponding historical intervertebral disc region image is cropped; An image segmentation model is trained according to historical scanned images and historical intervertebral disc region images, and an intervertebral disc image is obtained according to the image segmentation model.

[0029] In a specific implementation, by retrieving a human spine reference image and extracting shape features, the system can accurately identify the spine part in the historical scanned image. This feature-based recognition method improves the accuracy of spine recognition. The cosine similarity formula is used to calculate the similarity between the first feature and the second feature, and a similarity threshold is set. This method ensures that only when the similarity reaches a certain standard, the part in the image will be identified as the spine, thereby reducing the possibility of misidentification. The third value and the fourth value are set as amplitude thresholds to ensure that only the part corresponding to the spinal cord curve with an amplitude distribution between these two thresholds is set as the historical intervertebral disc curve part. This threshold setting helps to accurately cut out the intervertebral disc area. The image segmentation model is trained based on the historical scanned image and the historical intervertebral disc area image. This data-driven model training method improves the accuracy and efficiency of intervertebral disc image segmentation.

[0030] The training logic of the image segmentation model includes: Using historical scanned images and historical intervertebral disc region images as sample sets, dividing the sample sets into training sets and test sets according to a first ratio, wherein the first ratio is obtained through training experience; Inputting the training set into the neural network for training to obtain an image segmentation model, and testing the image segmentation model according to the test set until the accuracy reaches a first accuracy, and then stopping the testing; The image segmentation model uses the scanned image as input, the intervertebral disc region image as output, the actual intervertebral disc region image as training target, and the distance between the geometric center of the intervertebral disc region image and the geometric center of the actual intervertebral disc region image as the accuracy standard.

[0031] The preprocessing includes contrast enhancement processing and denoising processing, the preprocessed image is input into the image segmentation model to obtain the intervertebral disc region image, the intervertebral disc region image is deleted from the preprocessed image, and the greatest common factor of the boundary line of the deleted image is used as the side length of the segmentation frame, the segmentation frame is a square, and the preprocessed image is segmented from the upper left corner of the deleted image according to the segmentation frame to obtain the region of interest image; The region of interest images are numbered, and the numbers of the region of interest images are natural numbers.

[0032] In the specific implementation, by using historical scanned images and historical intervertebral disc area images as sample sets, the system can train the image segmentation model based on actual data. This data-driven method improves the accuracy and generalization ability of the model. The sample set is divided into a training set and a test set according to the first ratio obtained through training experience. This reasonable data set division ensures the independence of model training and testing, avoids overfitting, and uses neural networks for model training. The powerful nonlinear mapping ability of neural networks enables the image segmentation model to capture complex image features and improve the accuracy of segmentation. The image segmentation model uses scanned images as input, intervertebral disc area images as output, and actual intervertebral disc area images as training targets. This precise output control ensures that the training objectives of the model are clear and improves the pertinence of the model.

[0033] The analysis module acquires relevant data of the region of interest, wherein the relevant data includes bone marrow fat fraction and fat area ratio; The fifth value and the sixth value are set as relevant data thresholds, and the first osteoporosis grade of the region of interest is determined according to the relevant data thresholds. The first osteoporosis grade includes an automatic first grade, an automatic second grade, and an automatic third grade. The determination logic of the first osteoporosis grade is: When the bone marrow fat score is less than or equal to the fifth value, and the fat area ratio is less than or equal to the sixth value, the first osteoporosis grade is set to the automatic first grade; when the bone marrow fat score is greater than the fifth value, and the fat area ratio is less than or equal to the sixth value, the first osteoporosis grade is set to the automatic second grade; when the bone marrow fat score is greater than the fifth value, and the fat area ratio is greater than the sixth value, the first osteoporosis grade is set to the automatic third grade; The severity of osteoporosis represented by the automatic first grade, the automatic second grade and the automatic third grade is in an increasing trend.

[0034] In a specific implementation, relevant data of the region of interest, such as the bone marrow fat fraction and the fat area ratio, can be obtained. These data are crucial for assessing the severity of osteoporosis. By setting the fifth value and the sixth value as the thresholds of the relevant data, the system can judge the osteoporosis grade of the region of interest based on these thresholds. This clear threshold setting helps to achieve automated grade judgment. According to the comparison of the bone marrow fat fraction and the fat area ratio with the threshold, the system automatically divides the osteoporosis grade into automatic first grade, automatic second grade and automatic third grade. This automated grade judgment reduces manual intervention and improves the efficiency and accuracy of judgment. The severity of osteoporosis represented by the automatic first grade, automatic second grade and automatic third grade is an increasing trend. This increasing representation method helps doctors and patients to clearly understand the severity of the disease.

[0035] The precise module calls the expert platform, inputs the region of interest image to the expert platform, and obtains the second osteoporosis grade, wherein the second osteoporosis grade includes an artificial first grade, an artificial second grade, and an artificial third grade; Comparing the second osteoporosis level with the first osteoporosis level to obtain a comparison result, and performing a second operation according to the comparison result, wherein the second operation includes adjusting a threshold of relevant data and sending a detection completion signal; When the first osteoporosis level is the same as the second osteoporosis level, the second operation is set to send a detection completion signal and output a corresponding number; when the first osteoporosis level is different from the second osteoporosis level, the second operation is set to adjust the threshold of the relevant data; When the second operation is to adjust the relevant data threshold, the seventh value and the eighth value are set as the changing variables of the fifth value and the sixth value respectively, the fifth value and the sixth value are continuously increased or decreased, and the corresponding first osteoporosis level is continuously obtained until the adjusted first osteoporosis level is the same as the second osteoporosis level, and then the adjustment of the relevant data threshold is stopped.

[0036] In the specific implementation, the precision module calls the expert platform, inputs the image of the region of interest, and obtains the second osteoporosis grade. This method of calling the expert platform helps to improve the accuracy and reliability of osteoporosis grade judgment. When the first osteoporosis grade is different from the second osteoporosis grade, the system will perform the operation of adjusting the threshold of the relevant data. This dynamic control mechanism helps to optimize the judgment criteria of the system, improve the adaptability and accuracy of the system, and the continuous threshold adjustment method helps to achieve accurate grade matching.

[0037] The present invention realizes an automated process from scanning parameter setting to image segmentation and data analysis through the collaborative work of a pre-operation module, an image processing module, an analysis module and a precision module, thereby improving the accuracy and efficiency of detection, and can perform personalized detection according to the subject's posture and historical scan images to ensure the accuracy of the detection results. By identifying the subject's posture and spinal part, the system can more accurately locate and analyze the intervertebral disc area. Through the image segmentation model and the analysis module, the system can obtain relevant data of the region of interest and set relevant data thresholds to determine the osteoporosis grade, providing a more comprehensive osteoporosis assessment. The precision module calls the expert platform, inputs the image of the region of interest, obtains the second osteoporosis grade, and compares the second osteoporosis grade with the first osteoporosis grade to ensure the accuracy and reliability of the detection results. The use of MRI technology for osteoporosis detection avoids radiation exposure, and is a non-invasive and safe detection method.

[0038] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program codes. Among them, the storage medium may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk. These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0039] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An osteoporosis detection system based on nuclear magnetic resonance, characterized in that: It includes pre-operation module, image processing module, analysis module and precision module; The pre-operation module sets scanning parameters, and sets a positioning line and a positioning frame, identifies the subject's posture, performs a first operation according to the subject's posture, obtains a historical scan image, identifies a spinal part in the historical scan image, marks feature points in the spinal part according to a spatial order, obtains a spinal cord curve image according to the feature points and Hough transform, crops a corresponding historical intervertebral disc region image according to the spinal cord curve image, trains an image segmentation model according to the historical scan image and the historical intervertebral disc region image, and obtains an intervertebral disc image according to the image segmentation model; The image processing module obtains the image after the first operation is completed, preprocesses the image, inputs the preprocessed image into the image segmentation model to obtain an intervertebral disc region image, segments the preprocessed image according to the intervertebral disc region image to obtain an image of a region of interest, and numbers the image of the region of interest; The analysis module obtains relevant data of the region of interest, sets a relevant data threshold, determines a first osteoporosis grade of the region of interest according to the relevant data threshold, and sets a label of the first osteoporosis grade according to a corresponding number; The precise module calls the expert platform, inputs the image of the region of interest to the expert platform, obtains a second osteoporosis grade, compares the second osteoporosis grade with the first osteoporosis grade, and performs a second operation according to the comparison result.

2. The osteoporosis detection system based on nuclear magnetic resonance as claimed in claim 1, characterized in that: The scanning parameters include radio frequency pulse interval time, echo time, layer thickness, layer spacing, matrix size and scanning time, and the scanning parameters are obtained according to historical detection experience.

3. The osteoporosis detection system based on nuclear magnetic resonance as claimed in claim 1, characterized in that: The setting logic of the positioning line and positioning frame settings includes: Acquire a bottom surface image of the internal space of the device, draw a center line of the bottom surface of the internal space, the center line is represented by a line connecting the midpoints of the width of the ground, set the center line as a positioning line, draw a plane passing through the center line and perpendicular to the ground image, and record it as a first plane; Get the height inside the device, set the height of the bottom surface as the starting height, set the height inside the device w as the ending height, draw a plane that is parallel to the bottom surface, perpendicular to the first plane and does not pass through the first plane, record it as the second plane, and set the second plane as the positioning box.

4. The osteoporosis detection system based on nuclear magnetic resonance as claimed in claim 1, characterized in that: The first operation includes sending a posture correction signal and sending a start device signal, wherein the posture correction signal includes a translation correction signal and a rotation correction signal; When the device senses an obstruction, it uses machine vision to obtain the subject's image and identify the subject's posture. When the subject is lying flat, the positioning operation is started; In response to starting the positioning operation, starting a first detection, the detection logic of the first detection includes: Identify the edge line of the subject's image, divide the image inside the edge line parallel to the positioning line, calculate the number of pixels on both sides of the positioning line that belong to the edge line, calculate the difference in the number of pixels on both sides of the positioning line that belong to the edge line, record it as the first difference, set the first value as the first difference threshold, when the first difference is less than or equal to the first value, start the second detection, position the subject up and down according to the positioning frame, when the first difference is greater than or equal to the first value, send a translation correction signal, and the subject performs horizontal translation according to the translation posture correction signal until the first difference is less than the first value, and then stop sending the translation correction signal.

5. The osteoporosis detection system based on nuclear magnetic resonance as claimed in claim 4, characterized in that: The detection logic of the second detection includes: Identify the height of the subject, and obtain the subject's height corresponding to two points symmetrical to the positioning line and distributed on the positioning frame; When the two points on the positioning line are symmetrical and the corresponding heights of the subject distributed on the positioning frame are the same, a start scanning signal is sent; when the two points on the positioning line are symmetrical and the corresponding heights of the subject distributed on the positioning frame are different, a rotation correction signal is sent, and the subject flips the angle according to the rotation correction signal until the two points on the positioning line are symmetrical and the corresponding heights of the subject distributed on the positioning frame are the same, and then the rotation correction signal is stopped.

6. The osteoporosis detection system based on nuclear magnetic resonance as claimed in claim 1, characterized in that: The recognition logic of the spine includes: Retrieve a human spine reference image, extract the shape feature of the human spine reference image, record it as the first feature, extract the shape feature of the historical scan image, record it as the second feature, calculate the similarity between the first feature and the second feature by using the cosine similarity formula, set the second value as the similarity threshold, and when the similarity is greater than or equal to the second value, set the corresponding part in the historical scan image as the spine part, otherwise, jump to the second feature of the next part; The cropping logic for the historical disc region images includes: Marking feature points in the spine portion according to a spatial order, wherein the spatial order is represented from top to bottom in a plan view; A spinal cord curve image is obtained according to the feature points and Hough transform, and the amplitude of the spinal cord curve image is obtained; The third value and the fourth value are set as amplitude thresholds, a portion of the spinal cord curve corresponding to an amplitude distribution between the third value and the fourth value is set as a historical intervertebral disc curve portion, a portion of the historical scan image corresponding to the portion of the historical intervertebral disc curve is set as a historical intervertebral disc region, and the corresponding historical intervertebral disc region image is cropped; An image segmentation model is trained according to historical scanned images and historical intervertebral disc region images, and an intervertebral disc image is obtained according to the image segmentation model.

7. The osteoporosis detection system based on nuclear magnetic resonance as claimed in claim 1, characterized in that: The training logic of the image segmentation model includes: Using historical scanned images and historical intervertebral disc region images as sample sets, dividing the sample sets into training sets and test sets according to a first ratio, wherein the first ratio is obtained through training experience; Inputting the training set into the neural network for training to obtain an image segmentation model, and testing the image segmentation model according to the test set until the accuracy reaches a first accuracy, and then stopping the testing; The image segmentation model uses the scanned image as input, the intervertebral disc region image as output, the actual intervertebral disc region image as training target, and the distance between the geometric center of the intervertebral disc region image and the geometric center of the actual intervertebral disc region image as the accuracy standard.

8. The osteoporosis detection system based on nuclear magnetic resonance as claimed in claim 1, characterized in that: The preprocessing includes contrast enhancement processing and denoising processing, the preprocessed image is input into the image segmentation model to obtain the intervertebral disc region image, the intervertebral disc region image is deleted from the preprocessed image, and the greatest common factor of the boundary line of the deleted image is used as the side length of the segmentation frame, the segmentation frame is a square, and the preprocessed image is segmented from the upper left corner of the deleted image according to the segmentation frame to obtain the region of interest image; The region of interest images are numbered, and the numbers of the region of interest images are natural numbers.

9. The osteoporosis detection system based on nuclear magnetic resonance as claimed in claim 1, characterized in that: The analysis module acquires relevant data of the region of interest, wherein the relevant data includes bone marrow fat fraction and fat area ratio; The fifth value and the sixth value are set as relevant data thresholds, and the first osteoporosis grade of the region of interest is determined according to the relevant data thresholds. The first osteoporosis grade includes an automatic first grade, an automatic second grade, and an automatic third grade. The determination logic of the first osteoporosis grade is: When the bone marrow fat score is less than or equal to the fifth value, and the fat area ratio is less than or equal to the sixth value, the first osteoporosis grade is set to the automatic first grade; when the bone marrow fat score is greater than the fifth value, and the fat area ratio is less than or equal to the sixth value, the first osteoporosis grade is set to the automatic second grade; when the bone marrow fat score is greater than the fifth value, and the fat area ratio is greater than the sixth value, the first osteoporosis grade is set to the automatic third grade; The severity of osteoporosis represented by the automatic first grade, the automatic second grade and the automatic third grade is in an increasing trend.

10. The osteoporosis detection system based on nuclear magnetic resonance as claimed in claim 1, characterized in that: The precise module calls the expert platform, inputs the region of interest image to the expert platform, and obtains the second osteoporosis grade, wherein the second osteoporosis grade includes an artificial first grade, an artificial second grade, and an artificial third grade; Comparing the second osteoporosis level with the first osteoporosis level to obtain a comparison result, and performing a second operation according to the comparison result, wherein the second operation includes adjusting a threshold of relevant data and sending a detection completion signal; When the first osteoporosis level is the same as the second osteoporosis level, the second operation is set to send a detection completion signal and output a corresponding number; when the first osteoporosis level is different from the second osteoporosis level, the second operation is set to adjust the threshold of the relevant data; When the second operation is to adjust the relevant data threshold, the seventh value and the eighth value are set as the changing variables of the fifth value and the sixth value respectively, the fifth value and the sixth value are continuously increased or decreased, and the corresponding first osteoporosis level is continuously obtained until the adjusted first osteoporosis level is the same as the second osteoporosis level, and then the adjustment of the relevant data threshold is stopped.

Images