A method for image recognition of medical records

Abstract

The present invention relates to a method for image recognition of medical archives, comprising the following steps: first, setting an image recognition tag on each page of a paper archive document of the medical archive; second, performing an image scan on each page of the paper archive document in the first step, obtaining image scan information, the image scan information including archive image information and tag image information, wherein the tag image information includes location information of the image to be scanned; and third, transmitting the image scan information in the second step to a processor for processing, wherein the processing steps include: step A, screening the archive image information based on the image location information, saving important image information, and deleting unprocessed image information; step B, performing distortion correction processing; and step C, performing image conversion on the important image information in step B.

Description

Technical Field

[0001] The present invention relates to an image recognition method, in particular to a method for performing image recognition on medical files. Background Art

[0002] Medical archives are scientific and technological archives generated by national and local medical and health departments during epidemic prevention and control, as well as health supervision and management activities. Based on their content, they can be divided into epidemic medical technical archives and health supervision and management archives. The former includes documents such as medical plans, medical accident appraisals, various Chinese medicine names, traditional medicine formulas, and medical records. The latter includes documents related to infectious disease and epidemic detection and prevention, environmental hygiene testing, food hygiene testing, occupational disease prevention and control, maternal and child health supervision, and border port health supervision and epidemic prevention.

[0003] Among the wide variety of medical records, the medical records generated in outpatient and clinical settings are the most important. Generally speaking, when patients go to the hospital for treatment, the order is roughly: registration, outpatient treatment, and hospitalization. Therefore, the medical records generally formed in the above process generally include: registration receipts, diagnosis receipts issued by doctors at each outpatient visit, each drug receipt, various examination receipts and drug receipts generated during hospitalization, etc.

[0004] The above-mentioned documents are of various types and in large quantities. With the development of big data technology, existing hospital systems urgently need to digitize various archived medical records and aggregate them to the cloud to facilitate future telemedicine. However, the existing method of converting medical records data is mostly based on manual input, which is inefficient and has a high error rate. This is the main disadvantage of the existing technology. Summary of the Invention

[0005] The technical solution adopted by the present invention is: a medical file image recognition method, which includes the following steps.

[0006] The first step is to set an image recognition tag on each page of the paper file of the medical file, and the paper file files of the same type of medical files are set with the same image recognition tag.

[0007] The second step is to perform image scanning on each page of the paper archive document in the first step one by one to obtain image scanning information, which includes archive image information and label image information, wherein the label image information includes image position information that needs to be scanned.

[0008] The archival image information includes a number of block image information, and the number of block image information is divided into a number of important image information and a number of unprocessed image information. The image position information corresponds to the number of important image information. The image position information contained in one image identification label corresponds to the number of important image information on a page of the paper archival document.

[0009] The third step is to transmit the image scan information in the second step to the processor for processing, and the processing steps include: Step A, screening the archival image information according to the image position information, saving the important image information, and deleting the unprocessed image information; Step B, performing distortion correction processing on the important image information obtained by screening; Step C, performing image conversion on the important image information in Step B, and storing the converted data.

[0010] The beneficial effects of the present invention are as follows: the image position information contained in one of the image recognition tags of the present invention corresponds to several block image information at specific positions on a page of the paper archive file. In practice, there are many types of medical records and their file formats are also various. The present invention first classifies the medical records through the image recognition tag. The positions of the important contents recorded on the paper archive files of the same category of medical records are the same. The present invention can lock the positions of the important contents on the paper archive file through the image recognition tag, that is, lock several specific important image information. The present invention can achieve targeted image processing of the important contents on the paper archive file by locking the specific important image information, and does not perform post-processing on the unlocked and unprocessed image information, thereby achieving the effect of improving the image processing speed and improving the image processing accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a schematic diagram of the paper archive document of the present invention.

[0012] Figure 2 It is a schematic block diagram of the principle of the present invention.

[0013] Figure 3 Schematic diagram of several marking units of the present invention.

[0014] Figure 4 It is a schematic diagram of the mother marking part and the child marking part of the present invention.

[0015] Figure 5 It is a specific schematic diagram of the mother marking part and the child marking part of the present invention.

[0016] Figure 6 Schematic cross-sectional view of the image recognition label of the present invention.

[0017] Figure 7This is an example of a specific implementation of defining image position information for the protruding marking strip and the protruding marking block of the present invention. DETAILED DESCRIPTION

[0018] like Figures 1 to 7 As shown, a method for image recognition of medical files includes the following steps.

[0019] In the first step, an image recognition label 100 is set on each page of the paper file 10 of the medical file. The paper file 10 of the same type of medical file is provided with the same image recognition label 100.

[0020] In specific implementation, there are many ways to divide the same type of medical records. For example, according to the medical department, the medical records of orthopedics are defined as the same type of medical records, the medical records of urology are defined as the same type of medical records, and so on. For example, according to the test function, the medical records of routine blood tests are defined as the same type of medical records, the medical records of routine urine tests are defined as the same type of medical records, and so on.

[0021] like Figures 1 to 2 As shown, in the second step, each page of the paper archive file 10 in the first step is image scanned one by one to obtain image scanning information 20, which includes archive image information 30 and label image information 40, wherein the label image information 40 includes image position information 41 that needs to be scanned.

[0022] The archival image information 30 includes a plurality of block image information 31, and the plurality of block image information 31 is divided into a plurality of important image information 32 and a plurality of unprocessed image information 33. The image position information 41 corresponds to the plurality of important image information 32. That is, the image position information 41 contained in one image identification label 100 corresponds to a plurality of block image information 31 at a specific position on a page of the paper archival document 10.

[0023] In practice, there are many types of medical records, and their file formats are also various. The present invention first classifies the medical records through the image recognition label 100. The positions of the important contents recorded on the paper file files 10 of the same category of medical records are the same. The present invention can lock the positions of the important contents on the paper file files 10 through the image recognition label 100, that is, lock a specific number of the important image information 32. The present invention can achieve targeted image processing of the important contents on the paper file files 10 by locking the specific number of the important image information 32, and does not perform post-processing on the unlocked and unprocessed image information 33, thereby achieving the effect of improving the image processing speed and improving the image processing accuracy.

[0024] The third step is to transmit the image scanning information 20 in the second step to the processor 50 for processing, and the processing steps include:

[0025] In step A, the archive image information 30 is screened according to the image position information 41 , the important image information 32 is saved, and the unprocessed image information 33 is deleted.

[0026] Step B: performing distortion correction processing on the important image information 32 obtained by screening.

[0027] In specific implementation, distortion correction processing mainly includes adjusting clarity, adjusting contrast, deepening the color of text strokes, etc.

[0028] Step C: performing image conversion on the important image information 32 in step B and storing the converted data.

[0029] During specific implementation, the specific content of image conversion may be converting the image into binary code and storing it, or converting the image into other computer codes and storing it.

[0030] In practice, the image can also be converted into a QR code for storage and output to facilitate later scanning and viewing.

[0031] During specific implementation, the processor 50 can be a PC, a processor, a server, a cloud processor, etc.

[0032] like Figure 3 As shown, the image recognition label 100 in the first step has an identification surface 110 , on which a plurality of marking units 120 are arranged. The plurality of marking units 120 are arranged on the identification surface 110 to display the image position information 41 of the label image information 40 .

[0033] During specific implementation, through the image scanning action in the second step, the content of the image position information 41 can be obtained by scanning a plurality of the marking units 120 , thereby screening out the important image information 32 on the paper archive file 10 .

[0034] During specific implementation, the marking unit 120 has a variety of preferred implementations, which are now described below.

[0035] like Figure 3 As shown in the first embodiment, the marking unit 120 includes a filled mark 210 and a non-filled mark 220 . A plurality of the filled marks 210 and a plurality of the non-filled marks 220 are arranged together to display the content of the image position information 41 .

[0036] Embodiment 2: The marking unit 120 is a Braille marking. In a specific implementation, when the image recognition label 100 with the Braille marking is set on the paper archive file 10, the image recognition label 100 can not only play the role of marking the important image information 32, but also facilitate the archive management personnel to classify the numerous paper archive files 10. During classification, the management personnel can press the Braille marking with their thumb to classify the paper archive files 10 through touch.

[0037] like Figures 4 to 7 As shown in the third embodiment, the marking units 120 are divided into a main marking portion 310 and a sub-marking portion 320 , wherein the main marking portion 310 is disposed above the sub-marking portion 320 .

[0038] The mother marking portion 310 includes several marking strips, and the downward extension lines 313 of each marking strip intersect at a convergence point 314. That is, the several marking strips are arranged in a fan shape, and the several marking strips are divided into protruding marking strips 311 and recessed marking strips 312. The several protruding marking strips 311 and the several recessed marking strips 312 are arranged radially to form the mother marking portion 310. A color layer is provided on the top of each protruding marking strip 311, and the color layer can be black ink.

[0039] The sub-marking part 320 includes several marking blocks, which are arranged in sequence below the main marking part 310. The first marking block among the several marking blocks is set at the confluence point 314 of the main marking part 310. During specific implementation, the several marking blocks can be arranged horizontally, vertically or obliquely. The several marking blocks are divided into protruding marking blocks 321 and recessed marking blocks 322. Several of the protruding marking blocks 321 and several of the recessed marking blocks 322 are arranged to form the sub-marking part 320. A color layer is set on the top of each of the protruding marking blocks 321, and the color layer can be black ink.

[0040] In practice, when scanning an image, the content of the image position information 41 can be specifically defined based on the protruding marking strip 311 and the color layer set on the top of the protruding marking block 321 .

[0041] In addition, the downward extension lines 313 of each marking bar intersect at a convergence point 314. At the same time, the structural design idea of ​​setting the first marking block among the several marking blocks at the convergence point 314 position of the parent marking part 310 makes it convenient for managers to perceive the specific content of the image recognition label 100 through touch and thus classify the paper archive file 10.

[0042] That is to say, when in use, the manager's thumb first presses on several of the marking bars, and then slides down to naturally contact several of the marking blocks. In other words, for a skilled manager, through the sliding action of the thumb, he can simultaneously perceive the specific content contained in the parent marking part 310 and the child marking part 320, thereby realizing efficient classification of the paper archive file 10.

[0043] During specific implementation, in order to prevent the color layer on the top of the protruding marking strip 311 from falling off and affecting image recognition, the protruding marking strip 311 has at least one inclined side wall, and the color layer is arranged on the inclined side wall, that is, the cross-section of the protruding marking strip 311 can be trapezoidal.

[0044] During specific implementation, in order to avoid the situation where the color layer on the top of the protruding marking block 321 falls off and affects image recognition, and at the same time improve the contact feel of the protruding marking block 321, the protruding marking block 321 includes a bottom ring groove 323 and a top contact platform 324, wherein the bottom ring groove 323 is arranged around the bottom of the protruding marking block 321, the color layer is arranged on the bottom surface of the bottom ring groove 323, and the top contact platform 324 is arranged on the top of the protruding marking block 321. The top contact platform 324 can be provided with structures such as bumps and convex rings to improve the contact feel.

[0045] In a specific implementation, the identification surface 110 of the image recognition label 100 is an inclined concave surface, so that the marking unit 120 has a physical tilt angle, thereby avoiding reflection due to vertical scanning during image scanning, thereby affecting the scanning accuracy.

[0046] In a specific implementation, the image recognition label 100 can be a folder-type tool, and the identification surface 110 is set on the front of the clip. In practice, the identification surface 110 can also be processed on the surface of the document paper by forcibly pressing with a steel stamp to realize the function of the image recognition label 100.

[0047] like Figure 7 , which is an example of a specific implementation of how the protruding marking strip 311 and the protruding marking block 321 define the image position information 41 in practice.

Claims

1. A method for image recognition of medical files, characterized in that: The steps include: The first step is to set an image recognition tag on each page of the paper file of the medical file. The paper file of the same type of medical file is set with the same image recognition tag. The second step is to scan each page of the paper archive document in the first step one by one to obtain image scanning information, which includes archive image information and label image information, wherein the label image information includes the image position information to be scanned. The file image information includes a plurality of block image information, the plurality of block image information is divided into a plurality of important image information and a plurality of unprocessed image information, the image position information corresponds to the plurality of important image information, and the image position information included in one image identification tag corresponds to the plurality of important image information on a page of the paper file. The third step is to transmit the image scan information in the second step to the processor for processing, wherein the processing steps include: Step A: filter the archive image information according to the image location information, save the important image information, and delete the unprocessed image information. Step B: perform distortion correction on the important image information obtained by screening. Step C: Perform image conversion on the important image information in step B and store the converted data. The image recognition label in the first step has an identification surface, and a plurality of marking units are arranged on the identification surface. The plurality of marking units are arranged on the identification surface to display the image position information content of the label image information. Through the image scanning action in the second step, the image position information content can be obtained by scanning the plurality of marking units, thereby screening out the important image information on the paper archive document. The marking units are divided into a main marking part and a sub-marking part, wherein the main marking part is arranged above the sub-marking part. The mother label portion includes a plurality of label strips, the downward extension lines of each label strip intersect at a convergence point, and the plurality of label strips are divided into protruding label strips and recessed label strips. The plurality of protruding label strips and the plurality of recessed label strips are radially arranged to form the mother label portion, and a color layer is provided on the top of each protruding label strip. The sub-marking portion includes a plurality of marking blocks, which are sequentially arranged below the main marking portion. The first of the plurality of marking blocks is set at the confluence point of the main marking portion. The plurality of marking blocks are divided into protruding marking blocks and recessed marking blocks. The plurality of protruding marking blocks and the plurality of recessed marking blocks are arranged to form the sub-marking portion. A color layer is set on the top of each protruding marking block. The protruding marking strip has at least one inclined side wall, and the color layer is arranged on the inclined side wall. The protruding marking block includes a bottom ring groove and a top contact platform, wherein the bottom ring groove is arranged around the bottom of the protruding marking block, the color layer is arranged on the bottom surface of the bottom ring groove, and the top contact platform is arranged on the top of the protruding marking block.

2. The medical file image recognition method according to claim 1, wherein: The distortion correction process in step B of the third step includes adjusting the clarity, adjusting the contrast, and deepening the color of the text strokes.

3. The medical file image recognition method according to claim 1, wherein: The image conversion in step C of the third step is to convert the image into binary code and store it.

4. The medical file image recognition method according to claim 1, wherein: Several of the marking blocks are arranged horizontally.

5. The medical file image recognition method according to claim 1, wherein: Several of the marking blocks are arranged vertically.

Images