The present invention relates to an apparatus for monitoring a medical procedure and to an associated method.

Carrying out an interventional procedure on a patient on the wrong side of their body, or else at the wrong site, may cause significant, even irreparable, damage to them.

As an example, an ultrasound-guided peripheral nerve block (PNB) can be performed on either the left or right side of the patient, depending on which side anaesthesia/analgesia is required. However, it is very difficult to tell which side of the patient the procedure is being carried out on from the ultrasound image alone: anatomy is mirrored down the midline and therefore looks the same (or similar) on both sides of the patient. Furthermore, the image displayed could reflect laterality from either side of the body depending on the orientation of the ultrasound probe. Because of this, manual checks have been developed to reduce the rate of occurrence of wrong-sided procedures (e.g., WHO surgical safety checklist and the 'Stop Before You Block' campaign from RA-UK) . Nevertheless, mistakes still happen: A 10-year surveillance study in the USA identified a wrongsite peripheral nerve block (PNB) rate of 1.28 per 10,000 blocks. In the UK, the Safe Anaesthesia Liaison Group identified 67 wrong-side blocks over a 15-month period.

Causes of "wrong-side" mistakes include failures to properly check/conf irm/mark the correct side/site, or an incorrect understanding of which side the procedure should be performed on . Another cause , which can occur even when appropriate checks are carried out , is confusion when the practitioner moves from the front to the back of the patient, or when the patient is turned face-up or face-down . Similarly, environmental factors can contribute , such as when equipment set-up or positioning di f fers from that which the practitioner is used to .

There is therefore a need for improved checking prior to, and during, the procedure to reduce the rate of occurrence of such an error .

Embodiments of the present invention aim to provide an apparatus , and an associated method, for monitoring a medical procedure in which the aforementioned problems are addressed .

The present invention is defined in the attached independent claims , to which reference should now be made . Further, preferred features may be found in the sub-claims appended thereto .

According to one aspect of the present invention, there is provided apparatus for monitoring a medical procedure , the apparatus comprising a camera and an image processing unit , wherein the camera is arranged in use to capture an image of at least a part of a patient and the image processing unit processes image data from the image to determine a configuration of the patient . The apparatus may comprise a monitor, such as a video display monitor, and a result of the processing may be arranged to be displayed on the monitor.

The camera may comprise a video camera and the image may be a video image.

The image processing unit preferably includes an artificial intelligence unit for determining the configuration of the patient .

The artificial intelligence unit may comprise a processor and a database. The artificial intelligence unit is preferably arranged in use to apply at least a machine learning process to the image data. In a preferred arrangement, the artificial intelligence unit determines one or more of position, orientation, condition of the patient and/or an anatomical site of the patient.

The video camera may be arranged in use to capture an image of at least a part of a clinician carrying out a procedure on the patient.

In a preferred arrangement, the artificial intelligence unit applies a deep-learning model to the video image data to estimate the pose of both the patient and the clinician carrying out the procedure. Suitable deep learning models for pose estimation area available off-the-shelf, e.g., OpenPose (Z . Cao, G. Hidalgo Martinez, T. Simon, S. Wei, Y.

A. Sheikh. OpenPose : Realtime Multi-Person 2D Pose

Estimation using Part Affinity Fields. IEEE Transactions on Pattern Analysis and Machine Intel ligence . 2019 ) . Other suitable algorithms to estimate pose may also be used .

In use , the apparatus may be arranged to receive an instruction from an operator regarding the side of the patient that is designated for the procedure . The apparatus is then preferably arranged in use to provide a warning signal as a result of the processing in the event that an incorrect position, orientation, condition of the patient and/or an incorrect anatomical site of the patient is determined by the arti ficial intelligence unit .

The apparatus may be arranged in use to provide a guidance via the monitor to a clinician when the clinician is positioning an ultrasound probe and/or a surgical tool during the medical procedure .

According to another aspect of the present invention, there is provided a method of monitoring a medical procedure , the method comprising capturing an image of at least a part of a patient and processing the image data to determine a configuration of the patient .

The method preferably comprises displaying a result of the processing on a monitor .

The method may further comprise providing guidance via the monitor to a clinician when positioning an ultrasound probe and/or a surgical tool during the medical procedure . In a further aspect , the invention provides a computer programme product on a computer readable medium, comprising instructions that , when executed by a computer, cause the computer to perform a method of monitoring a medical procedure , the method comprising capturing an image of at least a part of a patient and processing image data to determine a configuration of the patient .

The invention also comprises a program for caus ing a device to perform a method of monitoring a medical procedure , the method comprising capturing an image of at least a part of a patient and processing the image data to determine a configuration of the patient .

The invention may include any combination of the features or limitations referred to herein, except such a combination of features as are mutually exclus ive , or mutually inconsistent .

A preferred embodiment of the present invention will now be described, by way of example only, with reference to the accompanying diagrammatic drawings , in which :

Figure 1 shows , schematically, an apparatus for monitoring a medical procedure , in accordance with an embodiment of the present invention;

Figure 2 shows , schematically, a display monitor in a first configuration; and Figure 3 shows , schematically, the display monitor of Figure 2 in a second configuration .

Turning to Figure 1 , there is shown generally at 100 an apparatus for monitoring a medical procedure , which in this example is to be an axillary PNB . A patient P lies on an operating table (not shown) . The patient is to undergo the procedure . A video camera 110 is mounted near the patient and an ultrasound probe 120 is positioned beside the patient in advance of the procedure .

Signals from the video camera 110 are fed back to an image processing unit 130 which is located close by the patient . The image processing unit has an arti ficial intelligence (Al ) unit in the form of a deep learning pose estimation module 140 , and also directly to a monitor 150 . The module 140 also has an output to the monitor, which is positioned within sight of a practitioner (not shown) performing the procedure .

The practitioner can observe the monitor during the procedure to receive , for example , confirmation of the correct orientation of the patient for the selected procedure and side , and that the designated procedure is being carried out on the correct side/ site of the patient , as will now be described .

The static video camera 110 is mounted either in or near the operating field, e . g . , on a stand or attached to the ceiling . The video camera 110 captures both the patient P and the practitioner . At the start of the procedure , the practitioner indicates to the system the site and side at which the procedure should take place. This can be incorporated into recognised safety checklists that are widely employed within medical practice (e.g., WHO surgical safety checklist and 'Stop Before You Block' ) . As the procedure is carried out, the video camera 110 captures the scene. The AT module 140 then processes the video stream to determine one or more of: the location of the patient, the orientation of the patient, whether the patient is in a prone, supine or lateral state and the location of the procedure site with respect to the patient (i.e., left vs right and which anatomical region the current scanning/block pertains to) .

The video image data is processed by applying the deep learning pose estimation model. This extracts the pose of the patient and the clinician performing the procedure and analyses the relative poses of the two to determine the patient's location and orientation with respect to the clinician. The pose estimation data is also analysed to identify the site and side of the procedure. This is then compared with the known site and side of the procedure as selected by the clinician.

Based on this data, the AT infers whether the procedure is being carried out on the same site and side that the practitioner had originally indicated. If not, a warning is displayed on the monitor. The warning may take the form of a visual message or indicium, an audible signal or a combination thereof. For example, Figure 2 shows schematically the monitor 150 and patient P in an example in which the system detects that the procedure is being carried out on the correct side/site of the patient. In contrast, Figure 3 shows a situation in which the procedure is about to be carried out on the wrong side/site of the patient. In this case a warning is issued.

In addition, the system can be used to record a confirmatory still/video image that demonstrates the side of the patient on which the procedure was performed, confirm to the practitioner (based on ultrasound probe position) which PNB was being performed (again, a safety control to check that the correct block was being done) , and, potentially, provide guidance to the practitioner as to where they need to move the ultrasound transducer to carry out the PNB (this may use the video data in conjunction with a real-time video feed from the ultrasound machine) .

Although the procedure used as an example above is of a PNB, the apparatus and methods described herein are applicable to other interventional procedures, including surgery.

In some situations, it may be possible to determine the side/site of the block from the silhouette of the patient/practitioner when performing the procedure. However, in other situations it will be difficult to do this (e.g., ESP block with patient lateral) . It may be necessary to require the practitioner to perform a specific action prior to starting (e.g. specific hand gesture) , to demonstrate which side they are about to perform the procedure . In a further alternative embodiment (not shown) the apparatus may simply indicate what the video camera is looking at , without the requirement of a pre-configured side/ site and/or without a warning message . Whilst endeavouring in the foregoing speci fication to draw attention to those features of the invention believed to be of particular importance , it should be understood that the applicant claims protection in respect of any patentable feature or combination of features referred to herein, and/or shown in the drawings , whether or not particular emphasis has been placed thereon .