Granted Patent B2

US 12,412,656 · App. 18/057,868 · Granted Sep 9, 2025

Inventors: Marcus Pfister (Bubenreuth, DE); Katharina Breininger (Erlangen, DE)

Assignee: SIEMENS HEALTHINEERS AG

G16H40/40A61B18/00A61B90/37A61B34/25

Abstract

Claims (79)

1. A computer-implemented method for provision of a control signal for carrying out a current or next working step in a process, the computer-implemented method comprising:

provisioning a plurality of data streams, wherein

the process is divided into a plurality of consecutive process sections,

the plurality of data streams depend on a process section of the process, and

each of the plurality of data streams is uniquely assigned an individual classifier from among a plurality of individual classifiers;

provisioning a list including a plurality of possible working steps;

applying the plurality of individual classifiers to the plurality of data streams, wherein

each individual classifier, based on the data stream assigned for each corresponding working step, determines a probability that the corresponding working step is being carried out as the current or next working step;

determining the current or next working step as a function of the determined probabilities; and

provisioning the control signal for carrying out the current or next working step.

2. The computer-implemented method as claimed in claim 1 , wherein

a data stream among the plurality of data streams is provided is an overview data stream,

the overview data stream is provided during the process,

the overview data stream is provided by a camera filming the process or via fluoroscopy, and

the computer-implemented method includes determining the process section as a function of the overview data stream.

3. The computer-implemented method as claimed in claim 2 , wherein

data streams provided in a process section are known, and

the computer-implemented method includes provisioning error information, which specifies that the process section determined is incorrect when an unexpected data stream is provided in the process section determined.

4. The computer-implemented method as claimed in claim 3 , further comprising:

correcting the process section determined as a function of the plurality of data streams.

5. The computer-implemented method as claimed in claim 2 , wherein the plurality of individual classifiers are trained individually at least one of centrally or in an institution in which the process is carried out, and wherein locations of the training of the plurality of individual classifiers are different for different ones of the plurality of individual classifiers.

6. The computer-implemented method as claimed in claim 2 , wherein the control signal is configured to control at least one of a medical system or an indication for operating personnel.

7. The computer-implemented method as claimed in claim 2 , wherein

each of the plurality of individual classifiers is assigned a confidence value and a threshold value,

the determining of the current or next working step includes determining a working step for which, with a corresponding individual classifier with a greatest confidence value, a greatest probability has been determined,

in response to the greatest probability being greater than the threshold value for the corresponding individual classifier, the working step is the current or next working step, and

in response to the greatest probability being less than or equal to the threshold value of the corresponding individual classifier, the determining of the current or next working step is repeated iteratively for an individual classifier with a next smallest confidence value until such time as a probability of the working step is determined to exceed the threshold value assigned to the corresponding individual classifier.

8. The computer-implemented method as claimed in claim 2 , wherein

each of the plurality of individual classifiers is assigned a confidence value,

the determining of the current or next working step includes

multiplying the probability determined for each working step by the confidence value assigned to a corresponding individual classifier, wherein

for each working step, a weighted probability is determined as a function of the corresponding individual classifier,

adding the weighted probabilities of different ones of individual classifiers for each working step, wherein

for each working step, a summed weighted probability is determined, and

the working step, for which the summed weighted probability is the greatest, is determined as the current or next working step.

9. The computer-implemented method as claimed in claim 1 , wherein the plurality of data streams are carried out as a function of an institution in which the process is carried out.

10. The computer-implemented method as claimed in claim 9 , wherein

in each process section, an individual classifier among the plurality of individual classifiers is a base classifier,

a data stream to which the base classifier is assigned is provided in each institution in at least a process section in which the base classifier is provided, and

other data streams of the plurality of data streams expected in the process are provided depending on the institution.

11. The computer-implemented method as claimed in claim 1 , wherein the plurality of individual classifiers are trained individually at least one of centrally or in an institution in which the process is carried out, and wherein locations of the training of the plurality of individual classifiers are different for different ones of the plurality of individual classifiers.

12. The computer-implemented method as claimed in claim 11 , wherein at least one of the plurality of individual classifiers is pre-trained centrally and is continuously retrained in the institution.

13. The computer-implemented method as claimed in claim 11 , wherein a classification result provided by one of the plurality of individual classifiers is used for supervised training of another of the plurality of individual classifiers in the institution.

14. The computer-implemented method as claimed in claim 1 , wherein the control signal is configured to control at least one of a medical system or an indication for operating personnel.

15. The computer-implemented method as claimed in claim 1 , wherein the control signal comprises information about a length of the current or next working step.

16. The computer-implemented method as claimed in claim 1 , wherein

each of the plurality of individual classifiers is assigned a confidence value and a threshold value,

the determining of the current or next working step includes determining a working step for which, with a corresponding individual classifier with a greatest confidence value, a greatest probability has been determined,

in response to the greatest probability being greater than the threshold value for the corresponding individual classifier, the working step is the current or next working step, and

in response to the greatest probability being less than or equal to the threshold value of the corresponding individual classifier, the determining of the current or next working step is repeated iteratively for an individual classifier with a next smallest confidence value until such time as a probability of the working step is determined to exceed the threshold value assigned to the corresponding individual classifier.

17. The computer-implemented method as claimed in claim 16 , wherein the threshold value for each individual classifier depends on the process section.

18. The computer-implemented method as claimed in claim 16 , wherein the confidence values assigned to the plurality of individual classifiers depend on the process section.

19. The computer-implemented method as claimed in claim 1 , wherein

each of the plurality of individual classifiers is assigned a confidence value,

the determining of the current or next working step includes

multiplying the probability determined for each working step by the confidence value assigned to a corresponding individual classifier, wherein

for each working step, a weighted probability is determined as a function of the corresponding individual classifier,

adding the weighted probabilities of different ones of individual classifiers for each working step, wherein

for each working step, a summed weighted probability is determined, and

the working step, for which the summed weighted probability is the greatest, is determined as the current or next working step.

20. The computer-implemented method as claimed in claim 1 , wherein the determining of the current or next working step comprises:

applying a trained function to the probability determined for each working step and each individual classifier, to determine the current or next working step.

21. The computer-implemented method as claimed in claim 1 , further comprising:

determining a selection of individual classifiers from the plurality of individual classifiers for the process section as a function of the plurality of data streams, and wherein in applying the plurality of individual classifiers, the selection of individual classifiers is applied to a corresponding selection of data streams.

22. The computer-implemented method as claimed in claim 1 , wherein the plurality of data streams include fluoroscopy recordings, film recordings, ECG recordings, sound recordings, a user input, or system information.

23. A non-transitory computer program product including a computer program, which is loadable into a memory of a provision system, the computer program including program sections for carrying out the computer-implemented method as claimed in claim 1 when the program sections are executed by the provision system.

24. A non-transitory computer-readable storage medium on which program sections are stored, the program sections, when executed at a provision system, cause the provision system to perform the computer-implemented method as claimed in claim 1 .

25. A provision system for provision of a control signal for carrying out a current or next working step in a process, the provision system comprising:

an interface configured to

provision a plurality of data streams,

provision a list including a plurality of possible working steps, and

provision a control signal for carrying out the current or next working step, wherein

the plurality of data streams depend on a process section of the process,

each of the plurality of data streams is uniquely assigned an individual classifier from among

a plurality of individual classifiers; and

at least one processor configured to

apply the plurality of individual classifiers to the plurality of data streams, wherein

each individual classifier, based on the data stream assigned, determines a probability for each corresponding working step, the probability specifying a probability that the corresponding working step is the current or next step, and determine the current or next working step as a

function of the determined probabilities.

Assignments (4)

Priority Claims (1)

Continuity (1)

References Cited (9)