| CPC A61B 34/20 (2016.02) [A61B 5/0215 (2013.01); A61B 5/4842 (2013.01); A61B 34/10 (2016.02); G06T 7/11 (2017.01); A61B 2034/104 (2016.02); A61B 2034/105 (2016.02); G06T 2207/20084 (2013.01); G06T 2207/30104 (2013.01); G06T 2207/30172 (2013.01)] | 27 Claims |
|
1. A method for providing automatic guidance of a percutaneous coronary intervention (PCI) procedure in real-time, comprising:
receiving, at a computing node, a sequential plurality of images of a blood vessel of a subject from a catheter imaging system from at least a first and second view, the blood vessel having a lumen, a lumen surface, and a wall;
providing the sequential plurality of images to a first neural network, the first neural network trained on a plurality of angiography images to output attention maps upon receiving input images;
receiving, from the first neural network, an attention map for each of the sequential plurality of images;
determining an average signal intensity for each attention map;
determining a peak average signal intensity for each of the first and second views based on the average signal intensity for each attention map;
selecting a first image of the sequential plurality of images having the first view that represents an end-diastolic cardiac phase based on the peak average signal intensity for the first view;
selecting a second image of the sequential plurality of images having the second view that represents the end-diastolic cardiac phase based on the peak average signal intensity for the second view;
identifying one or more landmark pairs in the first and the second images;
building, at the computing node, a 3D model of the blood vessel based on the first image, the second image, and the one or more landmark pairs therebetween;
segmenting, by a second neural network at the computing node, one or more materials between the lumen and the wall of the blood vessel appearing in the first image and the second image;
reconstructing, at the computing node, the lumen surface based on the first and second image;
determining, at the computing node, material properties of the blood vessel, the material properties including one or more of (a) a wall thickness, (b) a plaque thickness, (c) a lumen area, (d) a plaque eccentricity and (e) a plaque constituent;
assigning, at the computing node, one or more material properties to the reconstructed lumen surface;
generating, at the computing node, a recommendation to guide an interventional procedure in real-time based on the 3D reconstructed vessel lumen surface and segmented materials, the recommendation comprising at least a PCI plan;
providing a display of an angiogram of the subject and the PCI plan superimposed upon the angiogram in real-time; and
performing on the subject, using the 3D reconstructed vessel lumen surface and segmented materials:
balloon catheter pre-dilation,
a percutaneous coronary intervention catheter, and
balloon catheter post-dilation.
|