Methods for identifying, extracting and annotating graphic elements in medical images

By identifying and encoding graphic elements of medical images, and using 3D slicer tools to establish a computer-recognizable model, the problem of graphic elements recognition error in medical images is solved, and accurate and fast disease judgment is achieved, which is suitable for AI medical image recognition.

CN113592857BActive Publication Date: 2024-11-15桓由之
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202110978633.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-08-25
Publication Date
2024-11-15
Estimated Expiration
2041-08-25

AI Technical Summary

Technical Problem

The identification, extraction and labeling of graphic elements in the prior art are omissions and errors in the identification, extraction and labeling of graphic elements in medical images, and are affected by doctors' experience and subjective judgments, making it difficult to accurately judge diseases, especially when the differences between multiple diseases are subtle, resulting in diagnostic errors and delayed treatment. opportunity.

Method used

By splitting medical images and establishing graphic feature recognition frame groups, using 3D slicer tools for annotation and encoding, a computer-recognizable benchmark and comparison model is established, and automatic graphic feature extraction and labeling is realized to reduce the impact of manual reading.

Benefits of technology

It improves the accuracy and speed of medical image recognition results, reduces time consumption, realizes automatic judgment and information integrity of medical images, and is suitable for the recognition of AI medical images.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present invention proposes a method for identifying, extracting and annotating graphic elements in medical images. The method further identifies, extracts and annotates graphic elements in medical images, outputs and analyzes the information obtained, avoids outputting only a single graphic information of the medical image for the doctor to judge by himself, and solves the technical problem of identifying, extracting and annotating graphic elements in medical images. The present invention avoids the influence of personal factors or errors on manual reading of medical images, improves the output speed of medical images, can be applied to the identification, extraction and annotation of graphic elements in medical images by AI, uses the image features of basic lesions, assists in the output of medical image results, and obtains more accurate and comprehensive medical information. It is suitable for application as a method for identifying, extracting and annotating graphic elements in medical images.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to methods for identifying, extracting and annotating graphic elements in AI medical images. Background Art

[0002] Medical imaging refers to the technology and process of obtaining images of internal tissues of the human body or a part of the human body in a non-invasive manner for medical treatment or medical research. It includes the following two relatively independent research directions: medical imaging system and medical image processing. The former refers to the process of image formation, including the study of imaging mechanism, imaging equipment, imaging system analysis and other issues; the latter refers to further processing of the acquired images, the purpose of which is to restore the original unclear image, to highlight certain feature information in the image, or to classify the image pattern, etc.

[0003] Taking CT images as an example, CT images are composed of a certain number of pixels with different grayscales from black to white arranged in a matrix. These pixels reflect the X-ray absorption coefficient of the corresponding voxels. The pixel size and number of images obtained by different CT devices are different. The size can be 1.0 × 1.0 mm, 0.5 × 0.5 mm, etc.; the number can be 256 × 256, i.e. 65536, or 512 × 512, i.e. 262144. Obviously, the amount of information recorded in a single medical image is very large. There are usually more than one medical image for each patient and each part. If traditional methods are still used for annotation and manual interpretation, it is inevitable that there will be omissions and errors in the diagnosis results. In addition, the reading and interpretation of medical images are usually affected by the experience and subjective judgment of radiologists. When radiologists are inexperienced or overly insist on their own opinions, errors are likely to occur, thereby affecting the doctor's judgment of patient information.

[0004] Moreover, there are tens of thousands of types of diseases, but the corresponding number of human organs is limited. Some diseases are highly similar in the early stages or in their manifestations, with only individual parts differing, or some diseases overlap. If only feedback is obtained from a single part, accurate information cannot be obtained. Therefore, medical imaging assistance can be used as a basis for diagnosing the disease, and the accuracy of medical imaging information is even more necessary to help doctors assist in diagnosing the patient's condition.

[0005] For some diseases, it is impossible to accurately determine the type of disease by relying solely on the medical images of a certain organ or system. For example, symptoms such as coughing, sputum, night sweats, and fatigue are usually suspected to be respiratory diseases. In the early stages of the disease, they can be diagnosed as a variety of diseases such as colds, influenza, pneumonia, and tuberculosis. Different diseases have different treatment plans. If the best treatment opportunity is delayed, the condition may worsen. At this time, medical images will be used to assist in determining the patient's condition. The accuracy of the recognition and extraction of graphic elements in medical images is particularly important.

[0006] Based on this, there is an urgent need to develop a method for identifying, extracting and annotating graphic elements in medical images that is suitable for AI, and to use the image features of basic lesions to assist in the output of medical imaging results and obtain more accurate and comprehensive medical information. Summary of the invention

[0007] In order to identify, extract and annotate graphic elements in medical images, the present invention proposes a method for identifying, extracting and annotating graphic elements in medical images. The method further identifies, extracts and annotates graphic elements in medical images, outputs the information obtained through analysis, avoids outputting only a single graphic information of the medical image for doctors to judge by themselves, and solves the technical problems of identifying, extracting and annotating graphic elements in medical images.

[0008] The solution adopted by the present invention to solve the technical problem is:

[0009] First, based on the medical images of different organs or systems of the patient that can be obtained, the medical images of each organ or system are split and a graphic element recognition frame group is established according to the position, distribution, number, edge, density, morphology, peripheral changes, enhanced organ function changes, angiographic changes, magnetic resonance imaging sequences, relevant examination results, pathological size, dynamic observation, general condition, symptoms and signs, specialist conditions, inspection, auscultation, palpation, laboratory test results and pathological results. Graphic element coding is performed according to the medical images of each organ or system, and a unique code for the medical images of each organ or system is established. The medical image codes are used to establish identification element groups according to the above grouping.

[0010] Secondly, medical images of normal organs or systems are input as the judgment benchmark, and the medical images of normal organs or systems are labeled and divided according to the identification element groups using the 3D slicer drawing tool. The entire medical image is processed into fragmented computer-recognizable information according to the identification element groups, and the medical images of normal organs or systems are grouped according to the identification element groups. A computer-recognizable benchmark model of graphic elements is established for the graphic element groups that have been processed by computer informationization.

[0011] Next, input the lesion medical images related to the lesions or disease manifestations of each organ or system, use the 3Dslicer drawing tool to mark and divide the medical images of the abnormal organs or systems according to the identification element groups, process the entire medical image into fragmented computer-recognizable information according to the identification element groups, group the medical images of the abnormal organs or systems according to the identification element groups, and establish a computer-recognizable comparison model for the graphic element groups that have been processed by computer informationization.

[0012] Finally, the medical images of the patients are input into the computer in the unit of data package NAME+ID. The computer automatically extracts, divides and annotates the graphic elements of the medical images of the patients according to the established graphic element groups, splits them into fragmented graphic elements, and records them as judgment groups. The judgment groups are matched and compared with the computer-recognizable benchmark model and the computer-recognizable comparison model respectively. When the graphic element code of the judgment group completely matches the computer-recognizable benchmark model, the output result is only the position code+Normal.

[0013] When the graphic element code of the judgment group matches the graphic element code in the computer-recognizable comparison model, the specific graphic code of the matching group is output, the abnormal graphic code is output to the judgment group package, and the abnormality is displayed and marked in the form of a red warning symbol on the operation interface.

[0014] Positive effect: Since the present invention splits the medical image into fragmented computer-recognizable information according to the graphic code, and establishes a computer-recognizable benchmark model and a computer-recognizable comparison model, by comparing the judgment group with the basic model and the comparison model, outputting the comparison result to the judgment group package, and marking it with a red warning symbol on the operation interface, the method processes the simple image information into fragmented computer-recognizable information, so that the computer can automatically judge whether the medical image is abnormal or not, and a single input can achieve comprehensive information comparison of the NAME+ID package, without repeated entry and comparison of different departments, avoiding the influence of personal factors or errors on manual reading of medical images, improving the output speed of medical image recognition results, saving a lot of time, and having the advantages of accurate medical image recognition results, complete information, and fast recognition speed, it can be applied to AI recognition of medical images. It is suitable for use as a method for identifying, extracting and annotating graphic elements in medical images. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be described clearly and completely below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] First, based on the medical images of different organs or systems of the patient that can be obtained, the medical images of each organ or system are split according to the position, distribution, number, edge, density, morphology, peripheral changes, enhanced organ function changes, angiographic changes, MRI sequences, related examination results, pathological size, dynamic observation, general condition, symptoms and signs, specialist conditions, inspection, auscultation, palpation, laboratory test results and pathological results, and graphic element recognition frame groups are established. Graphic element coding is performed according to the medical images of each organ or system, and a unique code for the medical images of each organ or system is established. The medical image codes are used to establish recognition element groups according to the above grouping.

[0017] Graphic recognition element group coding (taking lungs as an example)

[0018] 1. Location Sowh

[0019] 1) Lung LU: 11. Apical segment of right upper lobe; 12. Anterior segment of right upper lobe; 13. Posterior segment of right upper lobe; 14. Medial segment of right middle lobe; 15. Lateral segment of right middle lobe; 16. Dorsal segment of right lower lobe; 17. Medial basal segment of right lower lobe; 18. Anterior basal segment of right lower lobe; 19. Lateral basal segment of right lower lobe; 110. Posterior basal segment of right lower lobe; 111. Right lung; 112. Right upper lobe; 113. Right middle lobe; 114. Right lower lobe; 115. Posterior segment of left upper lobe; 116. Anterior segment of left upper lobe ; 117. Upper lingular segment of left upper lobe; 118. Lower lingular segment of left upper lobe; 119. Dorsal segment of left lower lobe; 120. Anterior inner basal segment of left lower lobe; 121. External basal segment of left lower lobe; 122. Posterior basal segment of left lower lobe; 123. Left lung; 124. Upper lobe of left lung; 125. Lower lobe of left lung; 126. Right hilum; 127. Left hilum; 128. Pulmonary lobule; 129. Interlobular septum; 130. Lobular center; 131. Subpleural; 132. Right oblique fissure; 133. Horizontal fissure; 34. Oblique fissure of left lung…

[0020] 2) Trachea: 21. Upper end of trachea; 22. Lower end of trachea; 23. Tracheal bifurcation;

[0021] 3) Bronchus; 31. Left main bronchus; 32. Right main bronchus; 33. Left apical posterior segmental bronchus; 34. Right anterior segmental bronchus; 35. Left superior lobe bronchus; 36. Left superior lingual segmental bronchus; 37. Left inferior lingual segmental bronchus; 38. Left inferior lobe bronchus; 39. Left anterior basal segmental bronchus; 310. Left lateral segmental bronchus; 311. Left posterior basal segmental bronchus; 312. Left apical (superior) segmental bronchus; 313. Right superior lobe bronchus; 314. right apical segmental bronchus; 315. right posterior segmental bronchus; 316. right anterior segmental bronchus; 317. right middle lobe bronchus; 318. right lateral segmental bronchus; 319. right medial segmental bronchus; 320. right apical (superior) segmental bronchus; 321. right lower lobe bronchus; 322. right posterior basal segmental bronchus; 323. right anterior basal segmental bronchus; 324. right medial (cardiac) basal segmental bronchus; 325. right lateral basal segmental bronchus…

[0022] 4) Pulmonary artery: 41. Left pulmonary artery trunk; 42. Left pulmonary artery trunk; 43. Right pulmonary artery trunk; 44. Right pulmonary artery trunk…

[0023] 5) Pulmonary vein Pul.Hyp: 51. Right superior pulmonary vein; 52. Right inferior pulmonary vein; 53. Left superior pulmonary vein; 54. Left inferior pulmonary vein…

[0024] 6) Intrapulmonary lymph nodes Lym.Pup: 51. apical segment of right upper lobe; 52. anterior segment of right upper lobe; 53. posterior segment of right upper lobe; 54. medial segment of right middle lobe; 55. lateral segment of right middle lobe; 56. dorsal segment of right lower lobe; 57. medial basal segment of right lower lobe; 58. anterior basal segment of right lower lobe; 59. lateral basal segment of right lower lobe; 510. posterior basal segment of right lower lobe; 511. right lung; 512. right upper lobe; 513. right middle lobe; 514. right lower lobe; 515. posterior segment of apex of left upper lobe; 516. left lung Anterior segment of upper lobe; 517. Upper lingular segment of left upper lobe; 518. Lower lingular segment of left upper lobe; 519. Dorsal segment of left lower lobe; 520. Anterior inner basal segment of left lower lobe; 521. Outer basal segment of left lower lobe; 522. Posterior basal segment of left lower lobe; 523. Left lung; 524. Upper lobe of left lung; 525. Lower lobe of left lung; 526. Right hilum; 527. Left hilum; 528. Pulmonary lobule; 529. Interlobular septum; 530. Lobular center; 531. Subpleural; 532. Right oblique fissure; 533. Horizontal fissure; 534. Oblique fissure of left lung…

[0025] 2. Distribution and number Dist / Num

[0026] 1) Distribution Dist: D1. Scattered; D2. Diffuse distribution; D3. Centripetal distribution; D4. Centrifugal distribution; D5. Blood-borne distribution; D6. Along the airway; D7. Lung edge (subpleural) distribution; D8. No distribution at the lung edge (subpleural); D9. Localized distribution…

[0027] 2) Number Num: DN1. Single shot; DN2. Multiple shots;

[0028] 3. Edge

[0029] 1) Edge: E1. Clear and sharp edge; E2. Clear edge; E3. Blurred edge; E4. Burr sign; E5. Rabbit ear sign; E6. Halo sign; E7. Reverse halo sign; E8. Drainage trachea; E9. Sinus tract and fistula; E10. Vascular bundle sign; E11. Unclear boundary with pleura; E12. Unclear boundary with large blood vessels...

[0030] 4. Density

[0031] 1) Density: Den1. High density; Den2. Slightly high density; Den3. Equal density; Den4. Low density; Den5. Lower density; Den6. Mixed density; Den7. Liquid-gas level inside the lesion; Den8. Uneven density inside the lesion; Den9. Density inside the lesion is separated; Den10. There is density and liquid-liquid level inside the lesion; Den11. Soft tissue density; Den12. Calcification; Den13. Plaque-like calcification; Den14. Clustered microcalcification; Den15. Sand-like calcification; Den16. Eggshell-like calcification; Den17. Ground glass density; Den18. Water-like density; Den19. Fat density; Den20. Gas density; Den21. Sclerosis...

[0032] 5. Form

[0033] 1) Form: Form 1. Linear; Form 2. Strip; Form 3. Round; Form 4. Quasi-round; Form 5. Irregular; Form 6. Dotted; Form 7. Patchy; Form 8. Clustered; Form 9. Lobulation; Form 10. Tree-bud; Form 11. Concave umbilical; Form 12. Inverse parabola; Form 13. Inverted S; Form 14. Mosaic; Form 15. Reticulated; Form 16. Honeycomb; Form 17. Beaded…

[0034] 6. Surrounding Changes SurC

[0035] 1) Surrounding change: SurC1. Atrophy; SurC2. Swelling; SurC3. Infiltration; SurC4. Invasion; SurC5. Pulling; SurC6. Pushing…

[0036] 7. Changes in organ function enhance OFCE

[0037] 1) Organ function changes and enhancement: OFCE1. Pulmonary artery phase; OFCE2. Pulmonary venous phase; OFCE3. Enhancement; OFCE4. Obvious enhancement; OFCE5. Weak enhancement; OFCE6. No enhancement; OFCE7. Fast in and fast out enhancement; OFCE8. Gradual enhancement; OFCE9. Continuous enhancement; OFCE10. Slow in and slow out enhancement; OFCE11. Vascular occlusion; OFCE12. Filling defect; OFCE13. Tumor blood vessels; OFCE14. Tumor staining; OFCE15. Tumor lake; OFCE16. Early venous appearance; OFCE17. Arteriovenous fistula; OFCE18. Vascular aberration; OFCE19. Systemic circulation blood supply; OFCE20. Pulmonary veins entering the systemic circulation; OFCE21. Rosette enhancement…

[0038] 8. Angiographic changes ANg

[0039] 1) Angiographic changes: Ang1. Pulmonary artery phase; Ang2. Pulmonary vein phase; Ang3. Enhancement; Ang4. Obvious enhancement; Ang5. Weak enhancement; Ang6. No enhancement; Ang7. Fast in and fast out enhancement; Ang8. Gradual enhancement; Ang9. Continuous enhancement; Ang10. Slow in and slow out enhancement; Ang11. Vascular occlusion; Ang12. Filling defect; Ang13. Tumor blood vessels; Ang14. Tumor staining; Ang15. Tumor lake; Ang16. Early venous appearance; Ang17. Arteriovenous fistula; Ang18. Vascular aberration; Ang19. Systemic circulation; Ang20. Pulmonary veins entering the systemic circulation; Ang21. Rosette enhancement…

[0040] 9. Nuclear Magnetic Resonance Sequence NMR

[0041] 1) NMR Sequence: NMR1.T1; NMR2.T2; NMR3.Proton density; NMR4.Fat suppression sequence; NMR5.Water suppression sequence; NMR6.Fat suppression and water suppression sequence; NMR7.In phase; NMR8.Anti-phase; NMR9.Lava sequence; NMR10.T2* sequence; NMR11.Heavy T2 sequence; NMR12.Magnetic susceptibility imaging; NMR13.Pop imaging; NMR14.DWI imaging; NMR15.ADC image…

[0042] 10. Related examination results ABo

[0043] 1) XX-ray; CT.CT; MRI.MRI; PET-CT.PET-CT; PET-MR.PET-MR; OCT.OCT; Endos7.Endoscopic optics; Multi8.Multimodality; Mole.ima9.Molecular imaging; Gene10.Gene information…

[0044] 11. Sick Size Ms

[0045] 1) Ms. measured size: x*y*zmm……

[0046] Where x = lesion length, y = lesion width, z = lesion height, x, y, z can all be equal to 0, and the count is to 2 decimal places;

[0047] 12. Dynamic Observation Dyno

[0048] 1) Dyno1. One side of the lung; Dyno2. Across lobes; Dyno3. One lobe; Dyno4. Segments; Dyno5. Across segments; Dyno6. Increased reexamination volume or area; Dyno7. Decreased reexamination volume or area; Dyno8. Reexamination volume or area unchanged; Dyno9. Increased reexamination density; Dyno10. Decreased reexamination density, Dyno11. Reexamination density unchanged; Dyno12. New additions in reexamination…

[0049] Dynamic observation automatically obtains the last medical image information of the NAME+ID package, compares the latest medical image graphic recognition elements with the last medical image graphic recognition elements, and judges that the volume or area of ​​the lesion increases when Dyno6=n>x (x=the size of the lesion corresponding to the last medical image graphic element, n is the size of the lesion corresponding to the latest medical image graphic element). When Dyno7=n<x, it means that the volume or area of ​​the lesion decreases. When Dyno7=-x, it is judged that the original lesion disappears. When Dyno8=x, it is judged that the volume or area of ​​the lesion remains unchanged.

[0050] 13. General Situation

[0051] 1) Inf1. Age; Inf2. Gender; Inf3. Occupational history; Inf4. Contact history; Inf5. Permanent residence area; Inf6. Birth history; Inf7. Onset time; Inf8. Silica dust history; Inf9. Other dust history; Inf10. Asbestos history; Inf11. Radiation exposure history; Inf12. History of epidemic areas; Inf13. History of endemic disease areas; Inf14. History of pastoral areas…

[0052] 14. Symptoms and signs

[0053] SYS1. Fever; SYS2. Bleeding of the skin and mucous membranes; SYS3. Edema; SYS4. Cough and expectoration; SYS5 Hemoptysis; SYS6. Chest pain; SYS7. Cyanosis; SYS8. Dyspnea; SYS9. Palpitations; SYS10. Nausea and vomiting; SYS11. Hematemesis; SYS12. Blood in the stool; SYS13. Abdominal pain; SYS14. Diarrhea; SYS15. Constipation; SYS16. Jaundice; SYS17. Back pain; SYS18. Joint pain; SYS19. Hematuria; SYS20. Frequent, urgent and painful urination; SYS21. Oliguria, anuria and polyuria; SYS22. Headache; SYS23. Dizziness; SYS24. Syncope; SYS25. Seizures and convulsions; SYS26. Impairment of consciousness; SYS27. Associated symptoms...

[0054] 15. Spec

[0055] Spec. Detailed description of specialist symptoms and signs:…….

[0056] 16. Inspection status CNIs

[0057] CNIns1. Diagnosis by observation of tongue; CNIns2. Diagnosis by observation of qi.

[0058] 17. Olfaction diagnosis CNS

[0059] CNSme.Hear the sound.

[0060] 18. Palpation of CNPa

[0061] CNPalm.Pal.

[0062] 19. Laboratory Test Results

[0063] CBC. Blood routine; RT urine routine…

[0064] 20. Pathological examination results Pat.ex

[0065] Pat.ex. pathology results.

[0066] After the coding is completed, the medical image of the normal organ or system is input according to the above-mentioned coded graphic element group, the medical image of the normal organ or system is input as a judgment benchmark, the medical image of the normal organ or system is annotated and divided according to the identification element group using a 3D slicer drawing tool, the entire medical image is processed into fragmented computer-recognizable information according to the identification element group, the medical image of the normal organ or system is grouped according to the identification element group, and a graphic element computer-recognizable benchmark model is established for the graphic element group after computer information processing;

[0067] Then, the lesion medical images related to the lesions or disease representations of each organ or system are input, and the medical images of the abnormal organs or systems are marked and divided according to the identification element groups by using the 3Dslicer drawing tool, and the entire medical image is processed into fragmented computer-recognizable information according to the identification element groups, and the medical images of the abnormal organs or systems are grouped according to the identification element groups, and a computer-recognizable comparison model is established for the graphic element groups after computer information processing;

[0068] Finally, the medical images of the patients are input into the computer in the unit of data package NAME+ID. The computer automatically extracts, divides and annotates the graphic elements of the medical images of the patients according to the established graphic element groups, splits them into fragmented graphic elements, and records them as judgment groups. The judgment groups are matched and compared with the computer-recognizable benchmark model and the computer-recognizable comparison model respectively. When the graphic element code of the judgment group completely matches the computer-recognizable benchmark model, the output result is only the position code+Nor (Normal), and no specific small code is output, such as: LU+Nor;

[0069] When the graphic element code of the judgment group matches the graphic element code in the computer-recognizable comparison model, the specific graphic code of the matching group is output, such as: 11.D2.DN1.Den12.E3.Den4.Form5.SurC4, and the abnormal graphic code is output to the judgment group package. The output graphic code is matched with the lesion and disease representation. For example, the above output result is: right upper lobe apex segment + diffuse distribution + blurred edge + low density + calcification + irregular shape + (towards the surrounding) infiltration. Since the computer-recognizable comparison model is the graphic element of the lesion or disease representation of each organ or system, when the graphic element code of the judgment group corresponds to the code in the computer-recognizable comparison model, the corresponding code can be judged as the corresponding lesion or disease representation. There is no need to repeatedly read the medical image judgment and naming. According to the comparison of the graphic code and the corresponding lesion or disease representation, it can be output as: pulmonary tuberculosis, and the abnormality is displayed. The mark is displayed in the form of a red warning symbol on the operation interface for interpretation, and the reader can quickly and accurately obtain information about the abnormal part.

[0070] The medical images described in the present invention include X-rays, CT scans, ultrasonic scans and magnetic resonance images of the lungs, liver, brain, chest, head and gallbladder. The above images can be similarly used for graphic element extraction, division and annotation according to the method described in the present invention. The methods used and the effects achieved are the same as those of the present invention, and other parts will not be repeated.

[0071] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for identifying, extracting and annotating graphic elements in medical images, characterized by: First, according to the medical images of different organs or systems of the patient that can be obtained, the medical images of each organ or system are split according to the position, distribution, number, edge, density, morphology, peripheral changes, enhanced organ function changes, angiographic changes, nuclear magnetic resonance sequences, related examination results, morbid size, dynamic observation, general condition, symptoms and signs, specialist conditions, inspection conditions, auscultation conditions, palpation, laboratory test results and pathological results, and graphic element recognition frame groups are established. Graphic element coding is performed according to the medical images of each organ or system, and unique coding of the medical images of each organ or system is established. The medical image coding is used to establish recognition element groups according to the above grouping; Secondly, input the medical images of normal organs or systems as the judgment benchmark, use the 3D slicer drawing tool to mark and divide the medical images of normal organs or systems according to the identification element groups, process the entire medical image into fragmented computer-recognizable information according to the identification element groups, group the medical images of normal organs or systems according to the identification element groups, and establish a computer-recognizable benchmark model of graphic elements for the graphic element groups that have been processed by computer informationization; Then, the lesion medical images related to the lesions or disease representations of each organ or system are input, and the 3D slicer drawing tool is used to mark and divide the medical images of the abnormal organs or systems according to the identification element groups, and the entire medical image is processed into fragmented computer-recognizable information according to the identification element groups, and the medical images of the abnormal organs or systems are grouped according to the identification element groups, and a computer-recognizable comparison model is established for the graphic element groups after computer information processing; Finally, the medical images of the patients are input into the computer in the unit of data package NAME+ID. The computer automatically extracts, divides and annotates the graphic elements of the medical images of the patients according to the established graphic element groups, splits them into fragmented graphic elements, and records them as judgment groups. The judgment groups are matched and compared with the computer-recognizable benchmark model and the computer-recognizable comparison model respectively. When the graphic element code of the judgment group completely matches the computer-recognizable benchmark model, the output result is only the position code+Nor. When the graphic element code of the judgment group matches the graphic element code in the computer-recognizable comparison model, the specific graphic code of the matching group is output, the abnormal graphic code is output to the judgment group package, and the abnormality is displayed and marked in the form of a red warning symbol on the operation interface.

Citation Information

Patent Citations

  • Medical image auxiliary diagnosis method and system combining image recognition and report editing

    CN109583440A

  • Medical image processing method

    CN110543894A

  • AI medical diagnosis image picture computer input method

    CN111739615A

  • Method and device for establishing human tumor three-dimensional virtual model

    CN113269868A

  • System and method for automatically detecting lesions in medical image by means of multi-model fusion

    WO2018120942A1