Mixed reality image guidance for medical interventions
Disclosed are approaches that may provide image-guidance to interventionalists by providing true 3D visualization and quantitative feedback in real-time. A guidance system may allow a physician to manipulate a medical device and see a 3D rendering with quantitative feedback floating in mixed reality, next to standard monitors. Image tracking may detect and co-register the medical device's 3D position using, for example, bi-plane C-arm X-ray fluoroscopy and provide a 3D trajectory as quantitative feedback. Patterns in a fluoroscopic image may be used to accurately determine an object's z-position from a single angle projection.
1 . A method of providing real-time image guidance and/or navigation for a medical procedure, comprising:
receiving a first image data set of an anatomical target in relation to at least one fiducial marker in a selected coordinate system;
generating a 3D model of the anatomical target, in relation to the first image data set and the at least one fiducial marker, in the selected coordinate system;
receiving at least one single-plane fluoroscopic image comprising the at least one fiducial marker and a medical device;
generating a mask or a center coordinate of the at least one fiducial marker from the at least one single-plane fluoroscopic image;
generating a 3D model of the medical device from the at least one single-plane fluoroscopic image in the selected coordinate system by performing feature extraction on the medical device and the at least one fiducial marker, and using a trained machine learning model to infer, based on extracted features, a z-position of at least one portion of the medical device;
registering the 3D model of the medical device with the 3D model of the anatomical target in the selected coordinate system; and
generating a real-time output image comprising the 3D model of the medical device registered with a geometry of the anatomical target in the selected coordinate system.
2 . The method of claim 1 , wherein the first image data set comprises preoperative images, images acquired during the medical procedure, or images acquired both preoperatively and during the medical procedure.
3 . The method of claim 1 , wherein the at least one fiducial marker is visible in both an act of receiving the first image data set of the anatomical target and the act of receiving the at least one single-plane fluoroscopic image comprising the medical device.
4 . The method of claim 1 , wherein the first image data comprises at least one of fluoroscopy data, echocardiography data, computed tomography (CT) data, or magnetic resonance (MR) data.
5 . The method of claim 1 , wherein the at least one fiducial marker comprises at least one first fiducial marker internal to a subject and at least one second fiducial marker external to the subject.
6 . The method of claim 1 , wherein the feature extraction comprises edge detection of a dimension of at least a portion of the medical device or a dimension of a fiducial marker borne by the medical device.
7 . The method of claim 1 , wherein the feature extraction comprises edge detection of a rotational position of at least a portion of the medical device or a rotational position of the fiducial marker borne by the medical device relative to one or more axes.
8 . The method of claim 1 , wherein the generating the 3D model of the medical device comprises determining at least one of a rotation angle, a translation, or a scaling factor based on a projection of the at least one fiducial marker in the at least one single-plane fluoroscopic image.
9 . The method of claim 1 , further comprising outputting a haptic output to a haptic device responsive to a predefined proximity of at least a portion of the 3D model of the medical device to a predetermined portion of the 3D model of the anatomical target in the selected coordinate system.
10 . The method of claim 1 , wherein the registering of the 3D model of the medical device with the 3D model of the anatomical target in the selected coordinate system comprises registering respiratory motion, cardiac motion, or both respiratory motion and cardiac motion, at least in part by processing image data, as between the first image data set and the received at least one single-plane fluoroscopic image, having a same phase of motion.
11 . The method of claim 1 , further comprising outputting the real-time output image to a display device, wherein the display device comprises a virtual reality (VR) headset, an augmented reality (AR) headset, a mixed reality (MR) headset, a wearable device comprising a heads-up-display, or an area display.
12 . The method of claim 1 , wherein the generating of the real-time output image yields quantitative feedback of z-axis position of at least a portion of the medical device relative to the anatomical target or another anatomical feature with an accuracy ranging from 0.1 mm to 5 mm and a precision of about 10 μm to 1 mm.
13 . The method of claim 12 , further comprising outputting the quantitative feedback to a device providing auditory, visual or tactile guidance for open-loop instruction to an interventionalist.
14 . The method of claim 12 , further comprising outputting the quantitative feedback as a closed-loop instruction to a robotic controller of the medical device.
15 . A system for image guidance comprising:
one or more processors; and
a non-transitory memory device storing processor executable instructions to cause the one or more processors, upon execution of the executable instructions, to:
receive a first image data set of an anatomical target in relation to at least one fiducial marker in a selected coordinate system;
generate a 3D model of the anatomical target, in relation to the first image data set and the at least one fiducial marker, in the selected coordinate system;
receive at least one single-plane fluoroscopic image comprising the at least one fiducial marker and a medical device;
generate a mask of the at least one fiducial marker from the at least one single-plane fluoroscopic image;
generate a 3D model of the medical device from the at least one single-plane fluoroscopic image in the selected coordinate system by performing feature extraction on the medical device and the at least one fiducial marker, and using a trained deep learning model to infer, based on extracted features, a z-position of at least one portion of the medical device;
register the 3D model of the medical device with the 3D model of the anatomical target in the selected coordinate system; and
generate a real-time output image comprising the 3D model of the medical device registered with a geometry of the anatomical target in the selected coordinate system.
16 . The system of claim 15 , wherein the first image data set comprises images acquired both preoperatively and during a medical procedure.
17 . The system of claim 15 , further comprising a haptic device, wherein the executable instructions stored on the non-transitory memory device cause the one or more processors, upon execution of the executable instructions, to output a haptic output to the haptic device responsive to a predefined proximity of at least a portion of the 3D model of the medical device to a predetermined portion of the 3D model of the anatomical target in the selected coordinate system.
18 . The system of claim 15 , further comprising a display device, the display device including a virtual reality (VR) headset, an augmented reality (AR) headset, a mixed reality (MR) headset, a wearable device comprising a heads-up-display, or an area display, wherein the executable instructions stored on the non-transitory memory device cause the one or more processors, upon execution of the executable instructions, to output the real-time output image to the display device.
19 . The system of claim 15 , further comprising:
a device providing auditory, visual and/or tactile guidance for open-loop instruction to an interventionalist,
wherein the executable instructions stored on the non-transitory memory device cause the one or more processors, upon execution of the executable instructions, to output quantitative feedback of z-axis position of at least a portion of the medical device relative to the anatomical target or another anatomical feature to the device to provide guidance for open-loop instruction to the interventionalist.
20 . The system of claim 15 , further comprising a robotic controller and a robot comprising an end effector to control the medical device,
wherein the executable instructions stored on the non-transitory memory device cause the one or more processors, upon execution of the executable instructions, to output quantitative feedback of z-axis position of at least a portion of the medical device relative to the anatomical target or another anatomical feature as a closed-loop instruction to the robotic controller to cause the robot to move the end effector controlling the medical device to cause a related movement of the medical device.