CN117830307A - Skeleton image recognition method and system based on artificial intelligence - Google Patents
Skeleton image recognition method and system based on artificial intelligence Download PDFInfo
- Publication number
- CN117830307A CN117830307A CN202410240699.2A CN202410240699A CN117830307A CN 117830307 A CN117830307 A CN 117830307A CN 202410240699 A CN202410240699 A CN 202410240699A CN 117830307 A CN117830307 A CN 117830307A
- Authority
- CN
- China
- Prior art keywords
- recognition
- identification
- modes
- data
- matters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 86
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 28
- 210000000988 bone and bone Anatomy 0.000 claims abstract description 50
- 238000009826 distribution Methods 0.000 claims abstract description 33
- 238000012163 sequencing technique Methods 0.000 claims abstract description 8
- 238000001514 detection method Methods 0.000 claims description 68
- 239000012634 fragment Substances 0.000 claims description 52
- 238000004590 computer program Methods 0.000 claims description 6
- 238000007405 data analysis Methods 0.000 claims description 6
- 238000004891 communication Methods 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 abstract description 4
- 238000003491 array Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
According to the artificial intelligence-based bone image recognition method and system, the matters in the initial matters are distributed based on the recognition data corresponding to the matters in the initial matters to obtain matters distribution, each matters in the matters distribution carry sequencing information, recognition result clusters corresponding to the recognition modes are determined from the first matters, sample image clusters corresponding to the recognition modes are determined from the second matters, recognition modes are performed based on the recognition result clusters corresponding to the recognition modes and sequencing information carried by the matters in the sample image clusters, data recognition results corresponding to the recognition modes are obtained, and the data recognition results represent whether the implementation of the recognition modes is different for tag data. Therefore, abnormal data in the skeleton image can be determined, the information of the patient can be judged, and in the follow-up treatment, the patient can be subjected to targeted treatment, so that the rapid rehabilitation of the patient is facilitated.
Description
Technical Field
The application relates to the technical field of image recognition, in particular to a bone image recognition method and system based on artificial intelligence.
Background
Along with the continuous development and progress of science and technology, the field that artificial intelligence relates to is more wide, and in the bone image recognition of artificial intelligence specific application to medical field, there may be some problems, because see the doctor not only need professional knowledge, still need more experiences can be more accurate to carry out the disease judgement to the patient, seriously lack the experience that the patient was cured at artificial intelligence. Therefore, artificial intelligence may have a problem of inaccurate bone image recognition in bone image recognition, and how to solve the problem of inaccurate bone image recognition is a technical problem that is currently difficult to solve.
Disclosure of Invention
In order to improve the technical problems in the related art, the application provides a skeleton image recognition method and system based on artificial intelligence.
In a first aspect, there is provided an artificial intelligence based bone image recognition method, comprising:
obtaining a plurality of preset initial item sets corresponding to identification modes; each item in the initial item set has tag data; the label data comprises a patient bone image label and identification data corresponding to the patient bone image label; the patient bone image labels corresponding to all matters in the initial matters set are consistent; the initial item set comprises a first item set and a second item set; each item in the first item set is an item for which one of the plurality of identification methods is implemented;
distributing the matters in the initial matters set by combining the identification data corresponding to each matters in the initial matters set to obtain matters distribution conditions; each item in the item distribution situation carries sequencing information; the precedence information indicates the positioning of items in the item distribution situation;
determining recognition result clusters corresponding to all recognition modes from the first event set, and determining sample image clusters corresponding to all recognition modes from the second event set;
carrying out recognition mode recognition by combining the recognition result clusters corresponding to the recognition modes and sequence information carried by matters in the sample image clusters to obtain data recognition results corresponding to the recognition modes; the data identification result indicates whether the implementation of the identification mode is different for the tag data.
In the application, the first item set includes a plurality of local item sets corresponding to the plurality of identification modes one by one; the obtaining the initial item set corresponding to the preset plurality of identification modes comprises the following steps:
determining the recognition mode detection condition of each recognition mode in the plurality of recognition modes;
executing corresponding identification modes for matters in the local matters set corresponding to the identification modes through the identification mode detection conditions of the identification modes;
and on the premise that the implementation of the plurality of identification modes is completed for the plurality of local item sets, obtaining initial item sets corresponding to the plurality of identification modes.
In this application, the determining the recognition method detection condition of each recognition method of the plurality of recognition methods, and executing, by using the recognition method detection condition of each recognition method, a corresponding recognition method for a transaction in the local transaction set corresponding to each recognition method includes:
determining an identification mode application area and an identification mode application fragment of each identification mode;
and implementing a corresponding recognition mode for the matters in the local matters set corresponding to each recognition mode through the recognition mode application area and the recognition mode application fragment of each recognition mode.
In this application, on the premise that the implementation of the plurality of recognition modes is completed for the plurality of local item sets, obtaining an initial item set corresponding to the plurality of recognition modes includes:
on the premise that the implementation of the plurality of identification modes is completed for the plurality of local event sets, the same detection area is determined by combining the identification mode application areas of the identification modes;
determining the same detection fragments by combining the identification mode application fragments of the identification modes;
and acquiring initial item sets corresponding to the plurality of identification modes through the same detection areas and the same detection fragments.
In this application, the determining the same detection area in combination with the identification mode application area of each identification mode and determining the same detection segment in combination with the identification mode application segment of each identification mode includes:
determining an application region overlapping set by combining the identification mode application regions of the identification modes;
determining the same detection area through the application area overlapping set;
determining an application fragment superposition set by combining the identification mode application fragments of each identification mode;
and determining the same detection fragments through the application fragment coincidence set.
In this application, said determining said same detection segment by said applying a segment coincidence set comprises:
if the application segment superposition set covers the malformed segment corresponding to the malformed data, determining a difference value segment through the application segment superposition set and the malformed segment;
and determining the same detection fragment through the difference fragment.
In this application, the collecting, by the same detection area and the same detection segment, the initial item set corresponding to the plurality of identification modes includes: collecting initial item sets corresponding to the plurality of identification modes through the same detection areas and the same detection fragments by using a clustered data feedback channel; the clustering data feedback channel comprises a data analysis channel or a data classification channel.
In this application, the determining, from the first event set, the recognition result cluster corresponding to each recognition mode includes: and determining the recognition result clusters corresponding to the recognition modes from the local item sets corresponding to the recognition modes.
In this application, the identifying means for identifying by combining the identifying result clusters corresponding to the identifying means and the sequence information carried by the matters in the sample image clusters to obtain the data identifying result corresponding to the identifying means includes:
identifying the identification result clusters corresponding to the identification modes and the sequence information carried by matters in the sample image clusters by utilizing an artificial intelligent thread to obtain first integrated data corresponding to the identification modes;
and determining a data identification result corresponding to each identification mode through the first integrated data corresponding to each identification mode and preset critical data.
In this application, the identifying means for identifying by combining the identifying result clusters corresponding to the identifying means and the sequence information carried by the matters in the sample image clusters to obtain the data identifying result corresponding to the identifying means includes:
identifying the sequence information carried by matters in the identification result clusters and the sample image clusters corresponding to the identification modes by utilizing a convolution thread to obtain second integrated data corresponding to the identification modes;
the second integrated data meets specified requirements; the second integration data comprises differences between the recognition result clusters and the sample image clusters and occurrence probability;
and determining the data identification results corresponding to the identification modes according to the differences corresponding to the identification modes and the occurrence probability.
In a second aspect, an artificial intelligence based bone image recognition system is provided comprising a processor and a memory in communication with each other, the processor being adapted to read a computer program from the memory and execute the computer program to implement the method as described above.
According to the artificial intelligence-based bone image recognition method and system, through obtaining the initial item set corresponding to the preset recognition modes, each item in the initial item set is provided with tag data, the tag data comprise a patient bone image tag and recognition data corresponding to the patient bone image tag, the patient bone image tag corresponding to each item in the initial item set is consistent, the initial item set comprises a first item set and a second item set, each item in the first item set is an item of which one recognition mode is implemented, the items in the initial item set are distributed based on the recognition data corresponding to each item in the initial item set, the item distribution situation is obtained, each item in the item distribution situation carries sequential order information, the sequential order information represents the positioning of the item in the item distribution situation, a recognition result cluster corresponding to each recognition mode is determined from the first item set, a sample image cluster corresponding to each recognition mode is determined from the second item set, recognition modes are recognized based on the recognition result cluster corresponding to each recognition mode and the sequential order information carried in the sample image cluster, and the data recognition result corresponding to each recognition mode is obtained, and whether the difference of the recognition modes of the data of the embodiment of the tag data exists or not. Therefore, abnormal data in the skeleton image can be determined, the information of the patient can be judged, and in the follow-up treatment, the patient can be subjected to targeted treatment, so that the rapid rehabilitation of the patient is facilitated.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments will be briefly described below, it being understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered limiting the scope, and that other related drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a bone image recognition method based on artificial intelligence according to an embodiment of the present application.
Detailed Description
In order to better understand the technical solutions described above, the following detailed description of the technical solutions of the present application is provided through the accompanying drawings and specific embodiments, and it should be understood that the specific features of the embodiments and embodiments of the present application are detailed descriptions of the technical solutions of the present application, and not limit the technical solutions of the present application, and the technical features of the embodiments and embodiments of the present application may be combined with each other without conflict.
Referring to fig. 1, an artificial intelligence based bone image recognition method is shown, which may include the following steps S201-S207.
S201, obtaining a plurality of preset initial item sets corresponding to identification modes; each item in the initial item set has tag data; the label data comprises a patient bone image label and identification data corresponding to the patient bone image label; the patient bone image labels corresponding to all matters in the initial matter set are consistent; the initial item set includes a first item set and a second item set; each item in the first set of items is an item for which one of several identification methods is implemented.
By way of example, the recognition means may be understood as the recognition means and the recognition path of the bone image, etc.
In this embodiment of the present application, each of the plurality of identification manners may be preset by a technician, and each data of the present application may include an experimental group and a sample group corresponding to the each identification manner. The sample group is a set of items for which the identification mode is not implemented, and the experiment group is a set of items for which the identification mode is implemented.
In some possible implementations, the initial set of events may include a first set of events for which an identification mode is implemented and a second set of events for which no identification mode is implemented. Further, the step of distinguishing the initial set of events into the first set of events and the second set of events may be performed prior to implementing the identification means. Further, the step of dividing the initial item set into the first item set and the second item set may be performed after the identification means is performed, and the step may be performed according to whether the identification means is performed on the items in the initial item set.
In order to ensure independence of the test and thus accuracy of the test, each item in the first event set can only be identified by one of a number of identification methods set in advance.
In some possible embodiments, the first event set includes a number of local event sets corresponding to the number of identification means one-to-one, i.e. the number of identification means and the number of local event sets are identical, and each identification means has its corresponding local event set and the corresponding local event set is unique. Further, the step of dividing the first item set into a plurality of partial item sets corresponding to the plurality of identification modes one by one may be performed before the identification modes are implemented. Further, the step of dividing the first item set into a plurality of local item sets corresponding to the plurality of identification modes one by one may be performed in the identification modes, and then distinguishing the local item sets according to the patient bone image tags in the tag data of the identification modes. The embodiments of the present application do not limit the determining operation sequence related to the time points of determining the plurality of local event sets corresponding to the plurality of identification modes one by one.
In an alternative embodiment, a process of determining an initial item set provided in the embodiments of the present application is shown, and the details specifically described in step S201 include the following steps.
S2011, determining the identification mode detection condition of each of a plurality of identification modes.
In the embodiment of the present application, the identification mode detection condition of each of the plurality of identification modes may be determined, where the identification mode detection condition makes it necessary to satisfy the identification mode detection condition when implementing the corresponding identification mode.
Further, the recognition mode detection condition of each of the several recognition modes may include at least one of a recognition mode application area and a recognition mode application fragment. Thus, when the recognition mode detection condition of each recognition mode is determined, the recognition mode application area of each policy may be determined, or the recognition mode application fragment of each recognition mode may be determined, or the recognition mode application area and the recognition mode application fragment of each policy may be determined.
In the embodiment of the present application, the implementation of the identification mode 1, the identification mode 2 and the identification mode 3 are all directed to the same instruction, that is, a plurality of identification modes correspond to the same patient bone image tag, that is, the patient bone image tags corresponding to the respective items in the initial item set are consistent.
In the embodiment of the application, the content of each identification mode can be identified by utilizing the identification model in the artificial intelligence, so as to obtain the identification mode application area and/or the identification mode application fragment corresponding to each identification mode. And the key element matching method can also be utilized to match the key elements of the content of each identification mode, so as to obtain an identification mode application area and/or an identification mode application fragment corresponding to each identification mode.
S2013, according to the identification mode detection conditions of the identification modes, implementing the corresponding identification mode for the matters in the local matter set corresponding to the identification modes.
Further, the identification method application area may be used to apply the corresponding identification method to the item in the local item set corresponding to each identification method.
Further, the segments may be applied according to the recognition methods of the respective recognition methods, and the corresponding recognition methods may be implemented for the matters in the local matters set corresponding to the respective recognition methods.
Further, the corresponding identification method may be implemented for the items in the local item set corresponding to each identification method according to the identification method application area and the identification method application fragment of each identification method.
Based on the above specific explanation of the example, the operation of embodiment 1 may be performed on the items in the local item set corresponding to the identification mode 1 according to the identification mode application area (i.e., the first platform) corresponding to the identification mode 1 and the identification mode application fragment (i.e., the first period). An implementation 2 operation is performed on items in the local item set corresponding to the identification method 2 according to the identification method application area (i.e., the second platform) and the identification method application fragment (i.e., the second time period) corresponding to the identification method 2, and a C operation is performed on items in the local item set corresponding to the identification method 4 according to the identification method application area (i.e., the first platform) and the identification method application fragment (i.e., the third time period) corresponding to the identification method 3.
Further, the application areas of the recognition modes of different recognition modes can be consistent and can be different. The application fragments of the recognition modes of different recognition modes can be consistent and can be different. In general, the operation of the different recognition modes is different.
S2015, on the premise that implementation of the plurality of identification modes is completed for the plurality of local item sets, obtaining initial item sets corresponding to the plurality of identification modes.
Further, on the premise that implementation of a plurality of identification modes for a plurality of local item sets is completed, an initial item set corresponding to the plurality of identification modes can be obtained. Each item in the initial item set has tag data, and the tag data comprises a patient bone image tag and identification data corresponding to the patient bone image tag.
S202, distributing the matters in the initial matters set by combining the identification data corresponding to each matters in the initial matters set to obtain matters distribution; each item in the item distribution situation carries sequencing information; the precedence information indicates the positioning of items in the event distribution.
By way of example, distributing the items in the initial item set may be understood as ordering the items in the initial item set, and ordering the data in the initial item set according to a priority level, where the more serious information is set to be of a higher priority level.
In the embodiment of the present application, in order to save platform resources and time of an experimenter, the method may perform the distribution of the items in the initial item set based on the identification data corresponding to each item in the initial item set in step S203, so as to obtain an item distribution situation. In an alternative embodiment, the items in the initial item set are distributed based on the identification data corresponding to each item in the initial item set, and the premise of obtaining the item distribution condition is clustering of the data.
Based on the above, on the premise that the implementation of the plurality of recognition modes is completed for the plurality of local item sets, the data can be clustered before the initial item sets corresponding to the plurality of recognition modes are obtained. The method for determining the initial item set under the data clustering condition provided by the embodiment of the application specifically comprises the following steps.
S401, on the premise that implementation of a plurality of recognition modes is completed for a plurality of local event sets, the same detection area is determined based on the application area of the recognition mode of each recognition mode.
In the embodiment of the application, on the premise that implementation of a plurality of recognition modes is completed for a plurality of local event sets, an application region overlapping set can be determined based on the application regions of the recognition modes of the respective recognition modes, and the same detection region can be determined according to the application region overlapping set.
Specifically, the identification mode application area corresponding to the identification mode 1 is a first platform, the identification mode application area corresponding to the identification mode 2 is a second platform, the identification mode application area corresponding to the identification mode 3 is a first platform, the superposition set of the first platform and the second platform can be determined to be the application area superposition set, and then the superposition set of the first platform and the second platform can be used as the same detection area. If the superposition set of the first platform and the second platform is the full platform, the full platform can be used as the same detection area, namely, the identification mode application area of each identification mode is the full platform.
Based on the above-described recognition, the recognition mode application area of each recognition mode can be redefined as the same detection area obtained by the coincidence set.
S403, determining the same detection fragments based on the identification mode application fragments of the identification modes.
In the embodiment of the application, the application fragment overlapping set can be determined based on the identification mode application fragments of the identification modes, and the same detection fragment can be determined according to the application fragment overlapping set.
S405, collecting a plurality of initial item sets corresponding to the identification modes according to the same detection area and the same detection fragment.
In order to further ensure the clustering of the data and further improve the accuracy of the data identification result, a plurality of initial item sets corresponding to the identification modes can be acquired through a clustered data feedback channel according to the same detection area and the same detection fragment, wherein the clustered data feedback channel comprises a data analysis channel or a data classification channel.
Further, the data analysis channel clustering refers to that the collection of related data of each item in the initial item set is based on the same data analysis. Further, the data classification channel clustering refers to that the collection of related data of each item in the initial item set is based on the same data classification. The actual choice in the application may be based on the specific difficulty of engineering implementation, and the application is not limited.
Thus, the embodiment of the application may collect a plurality of initial item sets corresponding to the identification modes, that is, may collect a first item set of the implemented identification mode and a second item set of the identification mode not implemented, that is, collect a plurality of local item sets and second item sets corresponding to the identification modes one by one.
S203, distributing the matters in the initial matters set based on the identification data corresponding to each matters in the initial matters set to obtain matters distribution; each item in the item distribution situation carries sequencing information; the precedence information indicates the positioning of the items in the event distribution scenario.
By way of example, distributing the items in the initial item set may be understood as ordering the items in the initial item set, and ordering the data in the initial item set according to a priority level, where the more serious information is set to be of a higher priority level.
Further, the numerical processing may be performed according to the positioning of each item in the initial item set in the item distribution case.
S205, determining recognition result clusters corresponding to the recognition modes from the first event set, and determining sample image clusters corresponding to the recognition modes from the second event set.
By way of example, a sample image may be understood as historical data.
In the embodiment of the present application, the recognition result clusters corresponding to the respective recognition modes may be determined from the local item set corresponding to the respective recognition modes, and the sample image clusters corresponding to the respective recognition modes may be determined from the second item set.
For example, a recognition result cluster corresponding to recognition method 1 may be determined from the partial item set corresponding to recognition method 1, and a sample image cluster corresponding to recognition method 1 may be determined from the second item set. The recognition result cluster corresponding to recognition method 2 may be determined from the partial item set corresponding to recognition method 2, and the sample image cluster corresponding to recognition method 2 may be determined from the second item set. The recognition result cluster corresponding to recognition pattern 3 may be determined from the partial item set corresponding to recognition pattern 3, and the sample image cluster corresponding to recognition pattern 3 may be determined from the second item set.
Further, all items in the partial item set corresponding to the identification method 1 may constitute an identification result cluster corresponding to the identification method 1, or partial items in the partial item set corresponding to the identification method 1 may constitute an identification result cluster corresponding to the identification method 1. The recognition result clusters corresponding to the recognition mode 2 and the recognition mode 3 can be identified in the same manner, and are not described in detail.
Further, the matters covered by the sample image clusters corresponding to the recognition method 1, the recognition method 2, and the recognition method 3 may not be repeated at all, or there may be repeated matters.
Therefore, the identification result clusters and the sample image clusters corresponding to the identification modes can be obtained through one-time global distribution, compared with a plurality of distributions corresponding to a plurality of identification modes, the platform resources and the time of an experimenter can be saved, and all matters in the identification result clusters and the sample image clusters corresponding to the identification modes cover own sequencing information.
S207, identifying the identification modes based on the identification result clusters corresponding to the identification modes and the sequence information carried by the matters in the sample image clusters, and obtaining data identification results corresponding to the identification modes; the data identification result indicates whether the identification mode is different from the tag data.
In an alternative embodiment, the artificial intelligence thread may be used to identify the identification result clusters corresponding to the respective identification modes and the sequence information carried by the matters in the sample image clusters, so as to obtain first integrated data corresponding to the respective identification modes, and determine the data identification result corresponding to the respective identification modes according to the first integrated data corresponding to the respective identification modes and preset critical data.
Specifically, the artificial intelligent thread may be used to identify the sequence information carried by the items in the identification result cluster and the sample image cluster corresponding to each identification mode, the number of the items in the identification result cluster and the sample image cluster, and the identification data of the items in the identification result cluster and the sample image cluster, so as to obtain first integrated data corresponding to each identification mode
It may be understood that, in the embodiment of the present application, by obtaining an initial item set corresponding to a plurality of preset identification modes, each item in the initial item set has tag data, the tag data includes a patient bone image tag and identification data corresponding to the patient bone image tag, each item in the initial item set corresponds to the patient bone image tag, the initial item set includes a first item set and a second item set, each item in the first item set is an item in which one identification mode is implemented in the plurality of identification modes, the items in the initial item set are distributed based on the identification data corresponding to each item in the initial item set, an item distribution situation is obtained, each item in the item distribution situation carries precedence order information, the precedence order information indicates a location of an item in the item distribution situation, an identification result cluster corresponding to each identification mode is determined from the first item set, and a sample image cluster corresponding to each identification mode is determined from the second item set, identification modes are identified based on the identification result cluster corresponding to each identification mode and precedence information carried by the item in the sample image cluster, and the data identification result corresponding to each identification mode is obtained, and the data identification result indicates whether the implementation of the identification modes has a difference with respect to the tag data. Therefore, abnormal data in the skeleton image can be determined, the information of the patient can be judged, and in the follow-up treatment, the patient can be subjected to targeted treatment, so that the rapid rehabilitation of the patient is facilitated.
In a possible implementation embodiment, determining the recognition mode detection condition of each recognition mode in the plurality of recognition modes, and implementing a corresponding recognition mode for the matters in the local matters set corresponding to each recognition mode according to the recognition mode detection condition of each recognition mode, which may specifically include the following steps.
S11, determining the identification mode application areas and the identification mode application fragments of the identification modes.
S12, implementing the corresponding identification mode for the matters in the local matters set corresponding to the identification modes through the identification mode application area and the identification mode application fragment of the identification modes.
In a possible implementation embodiment, the collecting, through the same detection area and the same detection segment, the initial item set corresponding to the plurality of identification modes includes: collecting initial item sets corresponding to the plurality of identification modes through the same detection areas and the same detection fragments by using a clustered data feedback channel; the clustering data feedback channel comprises a data analysis channel or a data classification channel.
It can be understood that the recognition mode detection conditions of each recognition mode in the plurality of recognition modes are determined, and the recognition mode detection conditions of each recognition mode are used for improving the problem that the application area of the recognition mode and the application fragment of the recognition mode are inaccurate, so that the corresponding recognition mode can be implemented on the matters in the local matters set corresponding to each recognition mode more accurately.
On the premise that implementation of the plurality of identification modes is completed for the plurality of local item sets, an initial item set corresponding to the plurality of identification modes is obtained, and the initial item set comprises the following steps.
And S01, on the premise that the implementation of the plurality of identification modes is completed for the plurality of local item sets, the same detection area is determined by combining the identification mode application areas of the identification modes.
S02, determining the same detection fragments by combining the identification mode application fragments of the identification modes.
S03, collecting initial item sets corresponding to the plurality of identification modes through the same detection area and the same detection fragment.
It can be understood that on the premise that implementation of the plurality of identification modes is completed for the plurality of local item sets, the problem that the same detection area and the same detection fragment are inaccurate is solved, so that accuracy of obtaining the initial item sets corresponding to the plurality of identification modes is improved.
In a possible implementation embodiment, the identifying method identifies the sequence information carried by the matters in the identifying result cluster and the sample image cluster corresponding to the identifying methods to obtain the data identifying result corresponding to the identifying methods, which includes the following steps.
And D1, identifying the identification result clusters corresponding to the identification modes and the sequence information carried by the matters in the sample image clusters by utilizing an artificial intelligent thread to obtain first integrated data corresponding to the identification modes.
And D2, determining data identification results corresponding to the identification modes through the first integrated data corresponding to the identification modes and preset critical data.
It can be understood that when the recognition modes are combined with the sequential information carried by the matters in the recognition result clusters corresponding to the respective recognition modes and the sample image clusters for recognition, the problem of inaccurate recognition is solved, so that the data recognition results corresponding to the respective recognition modes can be obtained more accurately.
In a possible implementation embodiment, the recognition mode is performed by combining the recognition result clusters corresponding to the respective recognition modes and the sequence information carried by the matters in the sample image clusters, so as to obtain the data recognition result corresponding to the respective recognition modes, which includes the following description.
And H1, identifying the identification result clusters corresponding to the identification modes and the sequence information carried by matters in the sample image clusters by utilizing a convolution thread to obtain second integrated data corresponding to the identification modes.
H2, the second integrated data meets the specified requirements; the second integration data includes differences between the recognition result clusters and the sample image clusters and occurrence probabilities.
And H3, determining the data identification result corresponding to each identification mode according to the difference corresponding to each identification mode and the occurrence probability.
It can be understood that when the recognition modes are combined with the sequence information carried by the matters in the recognition result clusters and the sample image clusters corresponding to the respective recognition modes to recognize, the problem of inaccuracy of the recognition result clusters and the sample image clusters is solved, so that the data recognition results corresponding to the respective recognition modes can be obtained more accurately.
On the basis of the above, there is provided an artificial intelligence based bone image recognition apparatus, the apparatus comprising:
the item obtaining module is used for obtaining a plurality of initial item sets corresponding to the preset identification modes; each item in the initial item set has tag data; the label data comprises a patient bone image label and identification data corresponding to the patient bone image label; the patient bone image labels corresponding to all matters in the initial matters set are consistent; the initial item set comprises a first item set and a second item set; each item in the first item set is an item for which one of the plurality of identification methods is implemented;
the distribution situation obtaining module is used for distributing the items in the initial item set by combining the identification data corresponding to each item in the initial item set to obtain item distribution situations; each item in the item distribution situation carries sequencing information; the precedence information indicates the positioning of items in the item distribution situation;
the recognition mode determining module is used for determining recognition result clusters corresponding to all recognition modes from the first event set and determining sample image clusters corresponding to all recognition modes from the second event set;
the result recognition module is used for recognizing the recognition modes by combining the recognition result clusters corresponding to the recognition modes and the sequence information carried by the matters in the sample image clusters to obtain data recognition results corresponding to the recognition modes; the data identification result indicates whether the implementation of the identification mode is different for the tag data.
On the above basis, an artificial intelligence based bone image recognition system is shown comprising a processor and a memory in communication with each other, the processor being adapted to read a computer program from the memory and execute it to implement the method as described above.
On the basis of the above, there is also provided a computer readable storage medium on which a computer program stored which, when run, implements the above method.
In summary, based on the above-mentioned scheme, by obtaining an initial item set corresponding to a plurality of preset identification modes, each item in the initial item set has tag data, the tag data includes a patient bone image tag and identification data corresponding to the patient bone image tag, each item in the initial item set corresponds to the patient bone image tag, the initial item set includes a first item set and a second item set, each item in the first item set is an item to which one of the plurality of identification modes is implemented, the items in the initial item set are distributed based on the identification data corresponding to each item in the initial item set, an item distribution situation is obtained, each item in the item distribution situation carries precedence information, the precedence information indicates the location of an item in the item distribution situation, an identification result cluster corresponding to each identification mode is determined from the first item set, and a sample image cluster corresponding to each identification mode is determined from the second item set, identification modes are identified based on precedence information carried by the items in the sample image cluster, and the data identification results corresponding to each identification mode are obtained, and the data identification results indicating whether the implementation of the identification modes has a difference with respect to the tag data. Therefore, abnormal data in the skeleton image can be determined, the information of the patient can be judged, and in the follow-up treatment, the patient can be subjected to targeted treatment, so that the rapid rehabilitation of the patient is facilitated.
It should be appreciated that the systems and modules thereof shown above may be implemented in a variety of ways. For example, in some embodiments, the system and its modules may be implemented in hardware, software, or a combination of software and hardware. Wherein the hardware portion may be implemented using dedicated logic; the software portions may then be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or special purpose design hardware. Those skilled in the art will appreciate that the methods and systems described above may be implemented using computer executable instructions and/or embodied in processor control code, such as provided on a carrier medium such as a magnetic disk, CD or DVD-ROM, a programmable memory such as read only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The system and its modules of the present application may be implemented not only with hardware circuitry, such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., but also with software, such as executed by various types of processors, and with a combination of the above hardware circuitry and software (e.g., firmware).
It should be noted that, the advantages that may be generated by different embodiments may be different, and in different embodiments, the advantages that may be generated may be any one or a combination of several of the above, or any other possible advantages that may be obtained.
Claims (10)
1. A bone image recognition method based on artificial intelligence, the method comprising:
obtaining a plurality of preset initial item sets corresponding to identification modes; each item in the initial item set has tag data; the label data comprises a patient bone image label and identification data corresponding to the patient bone image label; the patient bone image labels corresponding to all matters in the initial matters set are consistent; the initial item set comprises a first item set and a second item set; each item in the first item set is an item for which one of the plurality of identification methods is implemented;
distributing the matters in the initial matters set by combining the identification data corresponding to each matters in the initial matters set to obtain matters distribution conditions; each item in the item distribution situation carries sequencing information; the precedence information indicates the positioning of items in the item distribution situation;
determining recognition result clusters corresponding to all recognition modes from the first event set, and determining sample image clusters corresponding to all recognition modes from the second event set;
carrying out recognition mode recognition by combining the recognition result clusters corresponding to the recognition modes and sequence information carried by matters in the sample image clusters to obtain data recognition results corresponding to the recognition modes; the data identification result indicates whether the implementation of the identification mode is different for the tag data.
2. The artificial intelligence based bone image recognition method of claim 1, wherein the first item set includes a plurality of local item sets corresponding to the plurality of recognition modes one to one; the obtaining the initial item set corresponding to the preset plurality of identification modes comprises the following steps:
determining the recognition mode detection condition of each recognition mode in the plurality of recognition modes;
executing corresponding identification modes for matters in the local matters set corresponding to the identification modes through the identification mode detection conditions of the identification modes;
and on the premise that the implementation of the plurality of identification modes is completed for the plurality of local item sets, obtaining initial item sets corresponding to the plurality of identification modes.
3. The artificial intelligence based bone image recognition method according to claim 2, wherein the determining the recognition method detection condition of each of the plurality of recognition methods, by the recognition method detection condition of each recognition method, implements a corresponding recognition method for a matter in the local matter set corresponding to each recognition method, includes:
determining an identification mode application area and an identification mode application fragment of each identification mode;
and implementing a corresponding recognition mode for the matters in the local matters set corresponding to each recognition mode through the recognition mode application area and the recognition mode application fragment of each recognition mode.
4. The artificial intelligence based bone image recognition method according to claim 3, wherein the obtaining the initial item set corresponding to the plurality of recognition modes on the premise that the implementation of the plurality of recognition modes is completed for the plurality of local item sets includes:
on the premise that the implementation of the plurality of identification modes is completed for the plurality of local event sets, the same detection area is determined by combining the identification mode application areas of the identification modes;
determining the same detection fragments by combining the identification mode application fragments of the identification modes;
and acquiring initial item sets corresponding to the plurality of identification modes through the same detection areas and the same detection fragments.
5. The artificial intelligence based bone image recognition method of claim 4, wherein the determining the same detection area in combination with the recognition mode application area of each recognition mode and determining the same detection fragment in combination with the recognition mode application fragment of each recognition mode includes:
determining an application region overlapping set by combining the identification mode application regions of the identification modes;
determining the same detection area through the application area overlapping set;
determining an application fragment superposition set by combining the identification mode application fragments of each identification mode;
and determining the same detection fragments through the application fragment coincidence set.
6. The artificial intelligence based bone image recognition method of claim 5, wherein the determining the same detection segment by the application segment coincidence set comprises:
if the application segment superposition set covers the malformed segment corresponding to the malformed data, determining a difference value segment through the application segment superposition set and the malformed segment;
and determining the same detection fragment through the difference fragment.
7. The artificial intelligence based bone image recognition method according to claim 4, wherein the acquiring the initial item set corresponding to the plurality of recognition modes through the same detection area and the same detection segment comprises: collecting initial item sets corresponding to the plurality of identification modes through the same detection areas and the same detection fragments by using a clustered data feedback channel; the clustering data feedback channel comprises a data analysis channel or a data classification channel.
8. The artificial intelligence based bone image recognition method of claim 2, wherein the determining the recognition result clusters corresponding to the respective recognition modes from the first event set includes: determining recognition result clusters corresponding to the recognition modes from the local item sets corresponding to the recognition modes;
the identifying method identifies the sequence information carried by the matters in the identification result clusters corresponding to the identifying methods and the sample image clusters to obtain the data identification results corresponding to the identifying methods, and the method comprises the following steps:
identifying the identification result clusters corresponding to the identification modes and the sequence information carried by matters in the sample image clusters by utilizing an artificial intelligent thread to obtain first integrated data corresponding to the identification modes;
and determining a data identification result corresponding to each identification mode through the first integrated data corresponding to each identification mode and preset critical data.
9. The artificial intelligence based bone image recognition method according to claim 8, wherein the identifying means for identifying the sequence information carried by the matters in the recognition result cluster and the sample image cluster corresponding to the respective identifying means to obtain the data recognition result corresponding to the respective identifying means includes:
identifying the sequence information carried by matters in the identification result clusters and the sample image clusters corresponding to the identification modes by utilizing a convolution thread to obtain second integrated data corresponding to the identification modes;
the second integrated data meets specified requirements; the second integration data comprises differences between the recognition result clusters and the sample image clusters and occurrence probability;
and determining the data identification results corresponding to the identification modes according to the differences corresponding to the identification modes and the occurrence probability.
10. An artificial intelligence based bone image recognition system comprising a processor and a memory in communication with each other, the processor being adapted to read a computer program from the memory and execute the computer program to implement the method of any one of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410240699.2A CN117830307B (en) | 2024-03-04 | 2024-03-04 | Skeleton image recognition method and system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410240699.2A CN117830307B (en) | 2024-03-04 | 2024-03-04 | Skeleton image recognition method and system based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117830307A true CN117830307A (en) | 2024-04-05 |
| CN117830307B CN117830307B (en) | 2024-05-14 |
Family
ID=90504391
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410240699.2A Active CN117830307B (en) | 2024-03-04 | 2024-03-04 | Skeleton image recognition method and system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117830307B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110277166A (en) * | 2019-06-28 | 2019-09-24 | 曾清福 | A kind of palace laparoscope assistant diagnosis system and method |
| CN113947709A (en) * | 2021-10-26 | 2022-01-18 | 八爪鱼人工智能科技(常熟)有限公司 | Image processing method and system based on artificial intelligence |
| US20220223271A1 (en) * | 2021-01-11 | 2022-07-14 | Emory University | Systems and Methods for Automatically Identifying and Tracking Medical Follow-Ups |
| JP2022137800A (en) * | 2021-03-09 | 2022-09-22 | 富士通株式会社 | Skeleton recognition method, skeleton recognition program, and physical exercise grading support system |
| CN115775631A (en) * | 2021-09-08 | 2023-03-10 | 深圳开立生物医疗科技股份有限公司 | Report comparison method, device and related equipment |
| CN115910300A (en) * | 2022-10-27 | 2023-04-04 | 周思宇 | Medical virtual platform system based on artificial intelligence and information processing method |
| CN116072255A (en) * | 2023-04-07 | 2023-05-05 | 北京大学第三医院（北京大学第三临床医学院） | Artificial intelligence-based cervical vertebra degenerative disease CT (computed tomography) examination structured report system |
| CN116805522A (en) * | 2023-05-10 | 2023-09-26 | 平安科技（深圳）有限公司 | Diagnostic report output method, device, terminal and storage medium |
| US20230334663A1 (en) * | 2022-04-19 | 2023-10-19 | Synthesis Health Inc. | Development of medical imaging ai analysis algorithms leveraging image segmentation |
| CN117524402A (en) * | 2023-11-30 | 2024-02-06 | 山东派瑞医疗科技有限公司 | Method for analyzing endoscope image and automatically generating diagnostic report |
| CN117594183A (en) * | 2023-12-12 | 2024-02-23 | 中国科学院深圳先进技术研究院 | Radiological report generation method based on inverse fact data enhancement |
-
2024
- 2024-03-04 CN CN202410240699.2A patent/CN117830307B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110277166A (en) * | 2019-06-28 | 2019-09-24 | 曾清福 | A kind of palace laparoscope assistant diagnosis system and method |
| US20220223271A1 (en) * | 2021-01-11 | 2022-07-14 | Emory University | Systems and Methods for Automatically Identifying and Tracking Medical Follow-Ups |
| JP2022137800A (en) * | 2021-03-09 | 2022-09-22 | 富士通株式会社 | Skeleton recognition method, skeleton recognition program, and physical exercise grading support system |
| CN115775631A (en) * | 2021-09-08 | 2023-03-10 | 深圳开立生物医疗科技股份有限公司 | Report comparison method, device and related equipment |
| CN113947709A (en) * | 2021-10-26 | 2022-01-18 | 八爪鱼人工智能科技(常熟)有限公司 | Image processing method and system based on artificial intelligence |
| US20230334663A1 (en) * | 2022-04-19 | 2023-10-19 | Synthesis Health Inc. | Development of medical imaging ai analysis algorithms leveraging image segmentation |
| CN115910300A (en) * | 2022-10-27 | 2023-04-04 | 周思宇 | Medical virtual platform system based on artificial intelligence and information processing method |
| CN116072255A (en) * | 2023-04-07 | 2023-05-05 | 北京大学第三医院（北京大学第三临床医学院） | Artificial intelligence-based cervical vertebra degenerative disease CT (computed tomography) examination structured report system |
| CN116805522A (en) * | 2023-05-10 | 2023-09-26 | 平安科技（深圳）有限公司 | Diagnostic report output method, device, terminal and storage medium |
| CN117524402A (en) * | 2023-11-30 | 2024-02-06 | 山东派瑞医疗科技有限公司 | Method for analyzing endoscope image and automatically generating diagnostic report |
| CN117594183A (en) * | 2023-12-12 | 2024-02-23 | 中国科学院深圳先进技术研究院 | Radiological report generation method based on inverse fact data enhancement |
Non-Patent Citations (4)
| Title |
|---|
| KIHYUN YOU 等: "CXR-CLIP:Toward Large Scale Chest X-ray Language-Image Pre-training", 《MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION-MICCAI 2023》, 1 October 2023 (2023-10-01), pages 101 - 111, XP047671390, DOI: 10.1007/978-3-031-43895-0_10 * |
| ZIHAO ZHAO 等: "CLIP in medical imaging:A comprehensive survey", 《ARXIV》, 26 December 2023 (2023-12-26), pages 1 - 22 * |
| 邢素霞 等: "基于记忆驱动的多模态医学影像报告自动生成研究", 《生物医学工程杂志》, vol. 41, no. 1, 29 February 2024 (2024-02-29), pages 60 - 69 * |
| 鲁晓;张勇;李昕;丁博渊;邱逦;罗燕;: "智能语音识别技术在超声审核工作站的研究与实践", 《数字技术与应用》, vol. 38, no. 09, 25 September 2020 (2020-09-25), pages 56 - 58 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117830307B (en) | 2024-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112435753A (en) | Automatic verification of medical data | |
| CN115240189A (en) | Base classification method, gene sequencer, and computer-readable storage medium | |
| US20240119596A1 (en) | Method and systems for predicting medical conditions and forecasting rate of infection of medical conditions via artificial intellidence models using graph stream processors | |
| CN114663223A (en) | Credit risk assessment method, device and related equipment based on artificial intelligence | |
| CN112800248A (en) | Similar case retrieval method, similar case retrieval device, computer equipment and storage medium | |
| CN117830307B (en) | Skeleton image recognition method and system based on artificial intelligence | |
| Gurcan et al. | Computerized pathological image analysis for neuroblastoma prognosis | |
| CN113065607A (en) | Image detection method, image detection device, electronic device, and medium | |
| CN110716778B (en) | Application compatibility testing method, device and system | |
| EP3570173A1 (en) | Equivalence checking device and equivalence checking program | |
| GB2594498A (en) | Instruction scheduling | |
| CN113705687B (en) | Image instance labeling method based on artificial intelligence and related equipment | |
| CN113609948B (en) | Method, device and equipment for detecting video time sequence action | |
| CN113570616B (en) | Image processing method, image processing device, electronic equipment and computer readable storage medium | |
| CN112507703B (en) | Medical entity identification method, device, medium and electronic equipment | |
| CN117476110B (en) | Multi-scale biomarker discovery system based on artificial intelligence | |
| CN115062495B (en) | Method and device for analyzing insulating property of material, electronic equipment and storage medium | |
| CN112286724B (en) | Data recovery processing method based on block chain and cloud computing center | |
| Nogueira et al. | Automatic analysis of Leishmania infected microscopy images via Gaussian mixture models | |
| US20220180155A1 (en) | System and method that can learn to recognize and predict errors in the computational detection of cells within tissues based on their geometric and morphological properties and perform tissue, cell and subcellular-scale dynamics analysis within a single user interface | |
| CN117290257B (en) | Software lifecycle standardization management method and system based on plug-in call | |
| CN116635942A (en) | Device control for biological sample analysis | |
| CN117114103A (en) | Corpus reconstruction method and device | |
| CN111462913A (en) | Automatic segmentation method and device for disease diagnosis in case document | |
| CN113157926A (en) | Data processing method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |