Anonymising robotic data
A method is provided of anonymising data in a surgical robotic system. The surgical robotic system comprises a robot having a base and an arm extending from the base to an attachment for an instrument, the arm comprising a plurality of joints whereby the configuration of the arm can be altered. The method comprises receiving a data stream captured by the surgical robotic system, the data stream comprising data relating to a surgical procedure and comprising personally-identifiable data; determining one or more personally-identifiable feature in the received data stream; and generating, in dependence on the determined personally-identifiable feature and the received data stream, an anonymised data stream omitting the personally-identifiable data.
1. A method of anonymising data in a surgical robotic system, the surgical robotic system comprising a robot having a base and an arm extending from the base to an attachment for an instrument, the arm comprising a plurality of joints whereby the configuration of the arm can be altered, the method comprising:
receiving a data stream captured by the surgical robotic system, the data stream comprising data relating to a surgical procedure and comprising personally-identifiable data;
determining one or more personally-identifiable feature in the received data stream in dependence on a state of the surgical robotic system, wherein the personally-identifiable feature is determined in response to one or more of:
determining that an instrument is not attached to the arm,
a control input received from the control console, determining a change in a mode of the robotic system that modifies the operational coupling between a control console and at least one surgical instrument other than the endoscope, and a mode of the robotic system, the mode of the robotic system being one of a group of modes of the robotic system comprising a surgical mode in which robotic control of the instrument by an operator is enabled and a non-surgical mode in which robotic control of the instrument by the operator is not enabled; and
generating, in dependence on the determined personally-identifiable feature and the received data stream, an anonymised data stream omitting the personally-identifiable data, wherein generating the anonymised data stream comprises, in dependence on the determined personally-identifiable feature, one or more of:
removing a data portion from the received data stream, the removed data portion comprising personally-identifiable data, and
masking a data portion of the received data stream, the masked data portion comprising personally-identifiable data.
2. The method according to claim 1 , in which the data stream comprises one or more data channel from a group of data channels comprising:
video data received from an endoscope coupled to the surgical robotic system;
audio data recorded in respect of a surgical procedure;
telematics data corresponding to the surgical robotic system; and
state data comprising the state of at least a portion of the surgical robotic system.
3. The method according to claim 2 , in which the personally-identifiable feature is determined in dependence on the one or more data channel.
4. The method according to claim 2 , in which the personally-identifiable feature is determined in dependence on a first data channel of the group of data channels, and generating the anonymised data stream comprises modifying at least one of a second data channel of the group of data channels, and the first data channel.
5. The method according to claim 1 , in which masking the data portion comprises one or more of:
blurring the data portion; and
replacing values of data in the data portion with mask values.
6. The method according to claim 1 , in which the data portion comprises one or more partial frames of data and/or one or more frames of data.
7. The method according to claim 1 , in which generating the anonymised data stream comprises saving a subset of the received data stream; and/or the anonymised data stream is generated as the surgical procedure is being performed.
8. The method according to claim 1 , in which the personally-identifiable feature is further determined in dependence on one or more of:
an operational state of the robot other than an indication of whether or not the endoscope is inserted through the port,
an attachment state of an instrument used to capture at least a portion of the data stream, where the instrument is not an endoscope,
analysis of data captured by an endoscope to determine whether or not the endoscope is inserted through a port,
a configuration of the arm,
an operational state of an instrument attached to the arm, and
a configuration of an instrument attached to the arm.
9. The method according to claim 1 , in which the personally-identifiable feature is determined in dependence on one or more of:
an instrument sensor signal received from an instrument sensor, the instrument sensor being configured to detect an instrument passing through a port providing access to a surgical site;
whether the video data comprises a circle that grows or shrinks;
whether the video data comprises a port-identifying feature;
whether the video data comprises an image of a face;
a measure of image white balance;
a measure of image spectrum;
a measure of image gradient from centre to edge;
a procedure being performed;
a marker forming part of the data stream; and
a machine learning algorithm that has been trained using a plurality of known personally-identifiable features.
10. The method according to claim 1 , in which generating the anonymised data stream comprises modifying the data portion for a time period that is one of:
before the personally-identifiable feature is determined,
after the personally-identifiable feature is determined,
both before and after the personally-identifiable feature is determined, and
between two personally-identifiable features.
11. The method according to claim 1 , in which the method comprises:
generating the anonymised data stream in real time or substantially real time;
sending the anonymised data stream to a remote processor thereby enabling the remote processor to perform real time or substantially real time analysis of the anonymised data stream; and
receiving from the remote processor in real time or substantially real time the result of the analysis for assisting an operator of the surgical robotic system.
12. A data anonymiser system for a surgical robotic system for anonymising data from the surgical robotic system, the surgical robotic system comprising a robot having a base and an arm extending from the base to an attachment for an instrument, the arm comprising a plurality of joints whereby the configuration of the arm can be altered, the data anonymiser system comprising:
a receiver configured to receive a data stream captured by the surgical robotic system, the data stream comprising data relating to a surgical procedure and comprising personally-identifiable data;
a personally-identifiable feature detector configured to determine one or more personally-identifiable feature in the received data stream in dependence on a state of the surgical robotic system, wherein the personally-identifiable feature is determined in response to one or more of:
determining that an instrument is not attached to the arm,
a control input received from the control console,
determining a change in a mode of the robotic system that modifies the operational coupling between a control console and at least one surgical instrument other than the endoscope, and
a mode of the robotic system, the mode of the robotic system being one of a group of modes of the robotic system comprising a surgical mode in which robotic control of the instrument by an operator is enabled and a non-surgical mode in which robotic control of the instrument by the operator is not enabled; and
a data anonymiser configured to generate, in dependence on the determined personally-identifiable feature and the received data stream, an anonymised data stream omitting the personally-identifiable data, wherein the data anonymiser is configured to generate the anonymised data stream by one or more of:
removing a data portion from the received data stream, the removed data portion comprising personally-identifiable data, and
masking a data portion of the received data stream, the masked data portion comprising personally-identifiable data.
13. The data anonymiser system according to claim 12 , in which the data stream comprises one or more data channel from a group of data channels comprising:
video data received from an endoscope coupled to the surgical robotic system;
audio data recorded in respect of a surgical procedure;
telematics data corresponding to the surgical robotic system; and
state data comprising the state of at least a portion of the surgical robotic system.
14. The data anonymiser system according to claim 13 , in which the personally-identifiable feature detector is configured to determine the personally-identifiable feature in dependence on a first data channel of the group of data channels, and the data anonymiser is configured to generate the anonymised data stream by modifying at least one of a second data channel of the group of data channels, and the first data channel.
15. The data anonymiser system according to claim 12 , in which the data anonymiser is configured to generate the anonymised data stream by one or more of:
removing a data portion from the received data stream, the removed data portion comprising personally-identifiable data, and
masking a data portion of the received data stream, the masked data portion comprising personally-identifiable data.
16. The data anonymiser system according to claim 15 , in which the received data stream comprises two or more data channels from the group of data channels, and the data anonymiser is configured to generate the anonymised data stream by modifying the data portion in respect of at least two of the data channels separately.
17. A non-transitory computer readable storage medium having stored thereon computer readable instructions that, when executed at a computer system, cause the computer system to perform a method of anonymising data in a surgical robotic system, the surgical robotic system comprising a robot having a base and an arm extending from the base to an attachment for an instrument, the arm comprising a plurality of joints whereby the configuration of the arm can be altered, the method comprising:
receiving a data stream captured by the surgical robotic system, the data stream comprising data relating to a surgical procedure and comprising personally-identifiable data;
determining one or more personally-identifiable feature in the received data stream in dependence on a state of the surgical robotic system, wherein the personally-identifiable feature is determined in response to one or more of:
determining that an instrument is not attached to the arm,
a control input received from the control console,
determining a change in a mode of the robotic system that modifies the operational coupling between a control console and at least one surgical instrument other than the endoscope, and
a mode of the robotic system, the mode of the robotic system being one of a group of modes of the robotic system comprising a surgical mode in which robotic control of the instrument by an operator is enabled and a non-surgical mode in which robotic control of the instrument by the operator is not enabled; and
generating, in dependence on the determined personally-identifiable feature and the received data stream, an anonymised data stream omitting the personally-identifiable data, wherein generating the anonymised data stream comprises, in dependence on the determined personally-identifiable feature, one or more of:
removing a data portion from the received data stream, the removed data portion comprising personally-identifiable data, and
masking a data portion of the received data stream, the masked data portion comprising personally-identifiable data.
18. The method according to claim 1 , wherein the non-surgical mode comprises:
a port training mode for establishing the location of a virtual pivot point about which the instrument is restricted to pivot when inserted through the port;
an instrument adjust mode for adjusting the instrument towards a position in which robotic control of the instrument by the operator is enabled; and
an instrument change mode for changing an instrument.
19. The data anonymiser system according to claim 12 ,
wherein the non-surgical mode comprises:
a port training mode for establishing the location of a virtual pivot point about which the instrument is restricted to pivot when inserted through the port;
an instrument adjust mode for adjusting the instrument towards a position in which robotic control of the instrument by the operator is enabled; and
an instrument change mode for changing an instrument.