Decision module and method for image-based operational decision support
Techniques are described for providing image-based operational decision support. The technique includes using a decision engine configured to generate output data based on a determined number of clinical findings in medical images of a patient. The decision module is configured to generate decision data depending on position data defining an actual position of the patient in a predefined clinical guideline, which comprises a set of possible next steps subsequent to the actual position. The decision data comprises information indicating which next steps from the set of possible next steps of the clinical guideline should be performed. The disclosure further describes a related method, a clinical decision system, and a related medical imaging system.
1. A decision engine configured to provide image-based operational decision support, comprising:
a memory configured to store instructions; and
processing circuitry configured to execute the instructions stored in the memory to:
generate output data based on a predetermined number of clinical findings in medical images of a patient; and
perform, based upon a trained module that is trained in accordance with training data, post-processing of the output data to generate decision data depending on position data defining a current position of the patient in a predefined clinical guideline comprising a set of possible next steps subsequent to the current position within the predefined clinical guideline,
wherein the decision data comprises information indicating which steps from among the set of possible next steps of the predefined clinical guideline should be performed.
2. The decision engine according to claim 1 , wherein the decision engine is trained on a predetermined number of training clinical findings and position data of a reference patient to generate the decision data as output data, and
wherein the output data comprises information indicating which steps from among the set of possible next steps of the predefined clinical guideline should be performed depending on the position data of the reference patient.
3. The decision engine according to claim 1 , wherein the memory is configured to store further instructions that, when executed by the processing circuitry, cause the processing circuitry to function as the trained module to perform the post-processing of the output data, and to generate the decision data by selecting the output data depending on the position data and the next steps of the predefined clinical guideline with respect to the current position of the patient within the predefined clinical guideline.
4. The decision engine according to claim 3 , wherein the training data comprises multiple sets of output data of the decision engine, a number of training clinical guidelines, and position data of a reference patient, to generate the decision data, and
wherein a ground truth within the training data comprises information indicating which next step from among a set of possible next steps of a training clinical guideline subsequent to a current position of the reference patient as indicated by the position data of the reference patient.
5. The decision engine according to claim 1 , wherein the memory is configured to store further instructions that, when executed by the processing circuitry, cause the processing circuitry to function as a selection module to select a decision engine from among a number of trained decision engines based on the position data.
6. The decision engine according to claim 5 , wherein each of the number of trained decision engines is trained based on respective training data to generate respective decision data as respective output data, the respective training data used to train each one of the number of trained decision engines comprising (i) a determined number of clinical findings of a reference patient, and (ii) a ground truth comprising information indicating which next step from among a set of possible next steps of a training clinical guideline subsequent to a current position indicated by position data of a reference patient.
7. The decision engine according to claim 1 , wherein the processing circuitry is configured to execute the instructions stored in the memory to generate the decision data further depending on predefined additional patient data including an age of the patient, a weight of the patient, a size of the patient, a sex of the patient, health conditions of the patient, known co-morbidities of the patient, laboratory results of the patient, and/or prior imaging results of the patient.
8. The decision engine according to claim 1 , wherein the processing circuitry is configured to execute the instructions stored in the memory to additionally generate information in a form of the output data or the decision data about one or more of a modality to be used for a further imaging examination, an imaging protocol to be used, and/or scheduling of a further imaging examination.
9. The decision engine according to claim 1 , wherein the processing circuitry is configured to execute the instructions stored in the memory to additionally output a suggested set of imaging parameter values for recording an image of the patient in the course of a further imaging examination.
10. A computer-implemented method for image-based operational decision support, comprising:
generating initial imaging data from a preceding examination of a patient;
determining, via automated processing of the initial imaging data, a number of clinical findings in medical images of a patient by:
providing position data defining a current position of the patient within a predefined clinical guideline; and
performing, based upon a trained module that is trained in accordance with training data, post-processing of the determined number of clinical findings to generate decision, the position data, and a structure of the predefined clinical guideline comprising a set of possible next steps subsequent to the current position of the patient in the predefined clinical guideline, the decision data comprising information indicating which steps from among the set of possible next steps of the predefined clinical guideline should be performed; and
outputting the decision data.
11. The method according to claim 10 , further comprising:
determining, based on patient data, the position data providing the current position of the patient within the predefined clinical guideline.
12. A clinical decision system for image-based operational decision support, comprising:
a data interface configured to receive (i) initial imaging data from a preceding examination of a patient, and (ii) position data defining a current position of the patient in a predefined clinical guideline; and
processing circuitry configured to:
determine a number of clinical findings via automated processing of the initial imaging data; and
perform, based upon a trained module that is trained in accordance with training data, post-processing of the determined number of clinical findings to generate decision data, the position data, and a structure of the predefined clinical guideline,
wherein the data interface is further configured to output the decision data.
13. The clinical decision system of claim 12 , wherein the clinical decision system is part of a medical imaging system configured to record images of the patient in the form of imaging data.
14. The decision engine according to claim 1 , wherein the training data comprises a dataset that functions to restrict the decision data to next steps within the predefined clinical guideline.
15. The decision engine according to claim 1 , wherein a ground truth within the training data comprises information indicating which next step from among a set of possible next steps of a training clinical guideline subsequent to a current position of a reference patient as indicated by position data of the reference patient.
16. The decision engine according to claim 1 , wherein the memory is configured to store further instructions that, when executed by the processing circuitry, cause the processing circuitry to function as a selection module to select a decision engine that is trained on the training data to generate the decision data as the output data, and
wherein the training data comprises a determined number of clinical findings of a reference patient.
17. The decision engine according to claim 16 , wherein the training data further comprises a ground truth comprising information indicating which next step from among a set of possible next steps of a training clinical guideline subsequent to a current position indicated by position data of the reference patient.
18. The decision engine according to claim 1 , wherein the training data functions to restrict the decision data to next steps within the predefined clinical guideline, which is associated with a medical facility at which the medical images of the patient were acquired.
19. The decision engine according to claim 1 , wherein the memory is configured to store further instructions that, when executed by the processing circuitry, cause the processing circuitry to restrict the decision data such that the steps from among the set of possible next steps of the predefined clinical guideline are associated only with diagnostic goals achieved via respective imaging modalities.