Medical imaging characteristic detection, workflows, and AI model management
Systems and methods for processing electronic imaging data obtained from medical imaging procedures, with use of trained artificial intelligence (AI) models, are disclosed herein. In an example, a use of a medical evaluation workflow involving an AI model includes: obtaining image data and non-image data associated with a medical imaging study; using at least one AI model to analyze the image data, with the trained AI model being validated with a defined governance standard to identify a characteristic or particular type of characteristic; identifying the characteristic with the AI model; and communicating the identified characteristic to a location associated with evaluation of the medical imaging study.
1. A method of processing data in a medical evaluation workflow, performed with electronic operations executed with a processor of a computing device, the electronic operations comprising:
obtaining image data and non-image data associated with a medical imaging study for a human subject;
using at least one trained artificial intelligence (AI) model to analyze the image data, wherein the trained AI model is validated with a defined governance standard to identify at least one particular characteristic;
identifying the particular characteristic in the image data, based on output from the trained AI model, wherein the identified particular characteristic is not indicated by the non-image data; and
communicating the identified particular characteristic to a location associated with evaluation of the medical imaging study, wherein the identified particular characteristic controls at least one automated medical image data processing action at the location associated with evaluation.
2. The method of claim 1 , wherein the defined governance standard establishes a specificity and sensitivity to identify the particular characteristic within multiple medical imaging studies performed for multiple human subjects.
3. The method of claim 2 , wherein the trained AI model is validated based on a verified identification of the particular characteristic, using a comparison of results identified by natural language processing in study reports and results identified by the trained AI model, the study reports and results being performed on the multiple medical imaging studies performed for the multiple human subjects.
4. The method of claim 1 , wherein using the trained AI model includes use of multiple trained AI models, wherein output from a first trained AI model is used as an input to a second trained AI model.
5. The method of claim 1 , wherein using the trained AI model to analyze the image data comprises causing execution of at least a first AI model on a portion of the image data at an on-premise computing location, and causing execution of at least a second AI model on a portion of the image data at an off-premise computing location.
6. The method of claim 1 , the electronic operations further comprising:
selecting the trained AI model from among a plurality of trained AI models, based on at least one characteristic indicated in the non-image data.
7. The method of claim 1 , the electronic operations further comprising:
using at least one other trained AI model to analyze the image data and the non-image data, wherein a value of the particular characteristic identified by the trained AI model differs from a value of the particular characteristic identified by the other trained AI model.
8. The method of claim 1 , the electronic operations further comprising:
receiving report data provided from evaluation of the medical imaging study;
using at least one AI model, trained for analysis of natural language processing, to analyze the report data;
identifying the particular characteristic in the report data, based on output from the AI model trained for analysis of natural language processing; and
determining at least one evaluation action based on the identifying of the particular characteristic from the report data and the identifying of the particular characteristic from the image data.
9. The method of claim 1 , wherein the identified particular characteristic relates to at least one of:
priority of the medical imaging study;
characteristics of a medical imaging procedure used to capture the image data;
at least one anatomical feature represented in the image data; or
at least one medical condition represented in the image data.
10. The method of claim 1 , wherein communicating the identified particular characteristic to a location associated with evaluation causes at least one prioritization action, including at least one of:
adding or prioritizing of an assignment of the medical imaging study in a worklist;
providing a notification of an emergent or critical finding to a source of the medical imaging study; or
prioritizing results and reporting for the medical imaging study.
11. The method of claim 1 , wherein communicating the identified particular characteristic to a location associated with evaluation causes at least one validation action, including at least one of:
validation of data in the non-image data;
identification of discrepancies between characteristics detected in the image data and indicated in the non-image data;
adding the medical imaging study to a quality assurance workflow; or
adding the medical imaging study to a second read workflow.
12. The method of claim 1 , wherein communicating the identified particular characteristic to a location associated with evaluation causes at least one corrective action, including at least one of:
performing corrective actions to the non-image data; or
performing corrective actions to a report associated with the medical imaging study.
13. The method of claim 1 , wherein communicating the identified particular characteristic to a location associated with evaluation causes at least one display action, including at least one of:
providing an indication of the identified particular characteristic to an evaluator of the medical imaging study;
changing a display of the image data at a display of an evaluator of the medical imaging study; or
changing a report prepared at a display of an evaluator of the medical imaging study.
14. The method of claim 1 , wherein the image data is provided from a radiological imaging procedure, wherein the non-image data is provided from a radiological imaging order, and wherein the medical imaging study corresponds to a radiological read request for diagnostic evaluation of the image data by a medical professional evaluator, the radiological read request indicated by the radiological imaging order.
15. The method of claim 1 , wherein the trained AI model is a: machine learning model or convolutional neural network.
16. A non-transitory machine-readable storage medium, the machine-readable storage medium comprising instructions that, when executed by a processor of a computing device, causes the computing device to perform operations comprising:
obtaining image data and non-image data associated with a medical imaging study for a human subject;
using at least one trained artificial intelligence (AI) model to analyze the image data, wherein the trained AI model is validated with a defined governance standard to identify at least one particular characteristic;
identifying the particular characteristic in the image data, based on output from the trained AI model, wherein the identified particular characteristic is not indicated by the non-image data; and
communicating the identified particular characteristic to a location associated with evaluation of the medical imaging study, wherein the identified particular characteristic controls at least one automated medical image data processing action at the location associated with evaluation.
17. The machine-readable storage medium of claim 16 ,
wherein the defined governance standard establishes a specificity and sensitivity to identify the particular characteristic within multiple medical imaging studies performed for multiple human subjects,
wherein the trained AI model is validated based on a verified identification of the particular characteristic, using a comparison of results identified by natural language processing in study reports and results identified by the trained AI model, the study reports and results being performed on the multiple medical imaging studies performed for the multiple human subjects.
18. The machine-readable storage medium of claim 16 , the instructions further to cause the computing device to perform operations comprising:
receiving report data provided from evaluation of the medical imaging study;
using at least one AI model, trained for analysis of natural language processing, to analyze the report data;
identifying the particular characteristic in the report data, based on output from the AI model trained for analysis of natural language processing; and
determining at least one evaluation action based on the identifying of the particular characteristic from the report data and the identifying of the particular characteristic from the image data.
19. The machine-readable storage medium of claim 16 , wherein communicating the identified particular characteristic to a location associated with evaluation causes at least one prioritization action, including at least one of:
adding or prioritizing of an assignment of the medical imaging study in a worklist;
providing a notification of an emergent or critical finding to a source of the medical imaging study; or
prioritizing results and reporting for the medical imaging study.
20. The machine-readable storage medium of claim 16 , wherein communicating the identified particular characteristic to a location associated with evaluation causes at least one validation action, including at least one of:
validation of data in the non-image data;
identification of discrepancies between characteristics detected in the image data and indicated in the non-image data;
adding the medical imaging study to a quality assurance workflow; or
adding the medical imaging study to a second read workflow.
21. The machine-readable storage medium of claim 16 , wherein communicating the identified particular characteristic to a location associated with evaluation causes at least one corrective action, including at least one of:
performing corrective actions to the non-image data; or
performing corrective actions to a report associated with the medical imaging study.
22. The machine-readable storage medium of claim 16 , wherein communicating the identified particular characteristic to a location associated with evaluation causes at least one display action, including at least one of:
providing an indication of the identified particular characteristic to an evaluator of the medical imaging study;
changing a display of the image data at a display of an evaluator of the medical imaging study; or
changing a report prepared at a display of an evaluator of the medical imaging study.
23. A computing system, comprising:
a processor; and
a memory device comprising instructions stored thereon, which when executed by the processor, configure the processor to perform electronic operations with the computing system comprising:
obtaining image data and non-image data associated with a medical imaging study for a human subject;
using at least one trained artificial intelligence (AI) model to analyze the image data, wherein the trained AI model is validated with a defined governance standard to identify at least one particular characteristic;
identifying the particular characteristic in the image data, based on output from the trained AI model, wherein the identified particular characteristic is not indicated by the non-image data; and
communicating the identified particular characteristic to a location associated with evaluation of the medical imaging study, wherein the identified particular characteristic controls at least one automated medical image data processing action at the location associated with evaluation.
24. The computing system of claim 23 ,
wherein the defined governance standard establishes a specificity and sensitivity to identify the particular characteristic within multiple medical imaging studies performed for multiple human subjects,
wherein the trained AI model is validated based on a verified identification of the particular characteristic, using a comparison of results identified by natural language processing in study reports and results identified by the trained AI model, the study reports and results being performed on the multiple medical imaging studies performed for the multiple human subjects.
25. The computing system of claim 23 , the processor further to perform electronic operations comprising:
receiving report data provided from evaluation of the medical imaging study;
using at least one AI model, trained for analysis of natural language processing, to analyze the report data;
identifying the particular characteristic in the report data, based on output from the AI model trained for analysis of natural language processing; and
determining at least one evaluation action based on the identifying of the particular characteristic from the report data and the identifying of the particular characteristic from the image data.