Method and system for anonymizing raw surgical procedure videos
This patent disclosure provides various verification techniques to ensure that anonymized surgical procedure videos are indeed free of any personally-identifiable information (PII). In a particular aspect, a process for verifying that an anonymized surgical procedure video is free of PII is disclosed. This process can begin by receiving a surgical video corresponding to a surgery. The process next removes personally-identifiable information (PII) from the surgical video to generate an anonymized surgical video. Next, the process selects a set of verification video segments from the anonymized surgical procedure video. The process subsequently determines whether each segment in the set of verification video segments is free of PII. If so, the process replaces the surgical video with the anonymized surgical video for storage. If not, the process performs additional PII removal steps on the anonymized surgical video to generate an updated anonymized surgical procedure video.
1. A method comprising:
receiving surgical video captured by an endoscope that is inserted inside a patient's body during a surgical procedure performed within an operating room (OR);
detecting an out-of-body (OOB) event in a video segment of the surgical video that occurs when the endoscope was taken out of a patient's body while the endoscope is still on by:
detecting, using an image processing algorithm, 1) a field of view of the endoscope that transitions from within the patient's body to outside the patient's body in a first set of video frames of the surgical video as beginning phase of the OOB event and 2) a subsequent field of view of the endoscope that transitions from outside the patient's body to inside the patient's body in a second set of video frames of the surgical video as an ending phase of the OOB event, and
determining that a time interval between the beginning phase and the ending phase is within a threshold period of time; and
responsive to detecting the OOB event, de-identifying personally-identifiable information (PII) embedded within the video segment.
2. The method of claim 1 , wherein the video segment comprises a plurality of video frames that make up the OOB event, wherein the PII that is de-identified is within a portion but not all of the plurality of video frames.
3. The method of claim 1 , wherein the PII comprises textual information or faces associated with a patient or a surgical staff of the surgical procedure, wherein de-identifying the PII comprises blurring or editing out the textual information or the faces from the video segment.
4. The method of claim 1 , wherein the surgical video comprises a plurality of video frames that comprise the first and second sets of video frames and the video segment comprises a sequence of one or more video frames of the plurality of video frames, wherein de-identifying the PII comprises blacking out each video frame of the sequence of one or more video frames that are between the first and second sets of video frames.
5. The method of claim 1 further comprising identifying the video segment of the surgical video based on the surgical procedure, wherein the detecting of the OOB event is performed responsive to identifying the video segment.
6. The method of claim 1 , wherein the image processing algorithm is a machine-learning (ML) based OOB-event detection model that is trained to detect OOB events based on a set of labeled video segments of a set of OOB events extracted from actual surgical procedure videos.
7. The method of claim 5 , wherein the surgical video comprises one or more de-identified video segments, wherein identifying the video segment comprises:
determining that the video segment has a high-probability of having the OOB event based on the surgical video; and
determining that the video segment is not one of the one or more de-identified video segments.
8. An apparatus comprising:
at least one processor; and
memory having stored therein instructions that when executed by the at least one processor causes the apparatus to:
receive surgical video captured by an endoscope that is inserted inside a patient's body during a surgical procedure performed within an operating room (OR);
detect an out-of-body (OOB) event in a video segment of the surgical video that occurs when the endoscope was taken out of a patient's body while the endoscope is still on by:
detecting, using an image processing algorithm, 1) a field of view of the endoscope that transitions from within the patient's body to outside the patient's body in a first set of video frames of the surgical video as beginning phase of the OOB event and 2) a subsequent field of view of the endoscope that transitions from outside the patient's body to inside the patient's body in a second set of video frames of the surgical video as an ending phase of the OOB event, and
determining that a time interval between the beginning phase and the ending phase is within a threshold period of time; and
responsive to detecting the OOB event, de-identify personally-identifiable information (PII) embedded within the video segment.
9. The apparatus of claim 8 , wherein the surgical video comprises a data file with raw video data previously captured by the endoscope during a past surgical procedure performed within the OR.
10. The apparatus of claim 8 , wherein the PII comprises textual information or faces associated with a patient or a surgical staff of the surgical procedure, wherein the instructions to de-identify the PII comprises instructions to blur or edit out the textual information or the faces from the video segment.
11. The apparatus of claim 8 , wherein the surgical video comprises a plurality of video frames that comprise the first and second sets of video frames and the video segment comprises a sequence of one or more video frames of the plurality of video frames, wherein the instructions to de-identify the PII comprises instructions to black out each video frame of the sequence of one or more video frames that are between the first and second sets of video frames.
12. The apparatus of claim 8 , wherein the memory comprises further instructions to identify the video segment of the surgical video based on the surgical procedure, wherein the detecting of the OOB event is performed responsive to an identification of the video segment.
13. The apparatus of claim 8 , wherein the image processing algorithm is a machine-learning (ML) based OOB-event detection model that is trained to detect OOB events based on a set of labeled video segments of a set of OOB events extracted from actual surgical procedure videos.
14. The apparatus of claim 12 , wherein the surgical video comprises one or more de-identified video segments, wherein the instructions to identify the video segment comprises instructions to:
determine that the video segment has a high-probability of having the OOB event based on the surgical video; and
determine that the video segment is not one of the one or more de-identified video segments.
15. A non-transitory machine-readable medium having instructions that when executed by at least one processor causes the at least one processor to:
receive surgical video captured by an endoscope that is inserted inside a patient's body during a surgical procedure performed within an operating room (OR);
detect an out-of-body (OOB) event in a video segment of the surgical video that occurs when the endoscope was taken out of a patient's body while the endoscope is still on by:
detecting, using an image processing algorithm, 1) a field of view of the endoscope that transitions from within the patient's body to outside the patient's body in a first set of video frames of the surgical video as beginning phase of the OOB event and 2) a subsequent field of view of the endoscope that transitions from outside the patient's body to inside the patient's body in a second set of video frames of the surgical video as an ending phase of the OOB event, and
determining that a time interval between the beginning phase and the ending phase is within a threshold period of time; and
responsive to detecting the OOB event, de-identify personally-identifiable information (PII) embedded within the video segment.
16. The non-transitory machine-readable medium of claim 15 , wherein the PII comprises textual information or faces associated with a patient or a surgical staff of the surgical procedure, wherein the instructions to de-identify the PII comprises instructions to blur or edit out the textual information or the faces from the video segment.
17. The non-transitory machine-readable medium of claim 15 , wherein the surgical video comprises a plurality of video frames that include the first and second sets of video frames and the video segment comprises a sequence of one or more video frames of the plurality of video frames, wherein the instructions to de-identify the PII comprises instructions to black out each video frame of the sequence of one or more video frames that are between the first and second sets of video frames.
18. The non-transitory machine-readable medium of claim 15 comprises further instructions to identify the video segment of the surgical video based on the surgical procedure, wherein the detecting of the OOB event is performed responsive to identifying the video segment.
19. The non-transitory machine-readable medium of claim 15 , wherein the image processing algorithm is a machine-learning (ML) based OOB-event detection model that is trained to detect OOB events based on a set of labeled video segments of a set of OOB events extracted from actual surgical procedure videos.
20. The non-transitory machine-readable medium of claim 18 , wherein the surgical video comprises one or more de-identified video segments, wherein the instructions to identify the video segment comprises instructions to:
determine that the video segment has a high-probability of having the OOB event based on the surgical video; and
determine that the video segment is not one of the one or more de-identified video segments.