US 12,390,279 B1
Method and system for simulating temporomandibular joint (TMJ) surgery based on artificial intelligence (AI) technology
Songsong Zhu, Chengdu (CN); Ruiye Bi, Chengdu (CN); Han Fang, Chengdu (CN); Yanjing Zhan, Chengdu (CN); Yao Liu, Chengdu (CN); Nan Jiang, Chengdu (CN); Peng Wang, Chengdu (CN); Haopeng Yu, Chengdu (CN); Xianni Yang, Chengdu (CN); and Haohan Li, Chengdu (CN)
Assigned to Sichuan University, Chengdu (CN)
Filed by Sichuan University, Chengdu (CN)
Filed on Apr. 11, 2025, as Appl. No. 19/176,358.
Claims priority of application No. 202411253738.9 (CN), filed on Sep. 9, 2024.
Int. Cl. A61B 34/10 (2016.01); A61B 17/56 (2006.01); G06T 7/00 (2017.01)
CPC A61B 34/10 (2016.02) [A61B 17/56 (2013.01); G06T 7/0012 (2013.01); A61B 2017/564 (2013.01); A61B 2034/105 (2016.02); G06T 2207/10004 (2013.01); G06T 2207/20112 (2013.01); G06T 2207/30008 (2013.01)] 2 Claims
OG exemplary drawing
 
1. A method for simulating temporomandibular joint (TMJ) surgery based on an artificial intelligence (AI) technology, comprising the following steps:
obtaining a plurality of TMJ images and corners corresponding to each corner in each TMJ image in other TMJ images;
constructing a coordinate system of each TMJ image; in the coordinate system, obtaining a motion synchronization degree of any two corners in each TMJ image based on a distance between coordinates of each of the any two corners in the TMJ image and coordinates of the corresponding corner in an adjacent TMJ image; and obtaining temporal bone corners and a temporal bone edge contour in each TMJ image based on the motion synchronization degree of and a distance between the any two corners in each TMJ image;
obtaining a plurality of mandible corners and a mandible motion direction based on a distance between the temporal bone corners in each TMJ image and coordinates of the corners in the TMJ image in the coordinate system;
obtaining a real mandible edge contour in each TMJ image based on an edge contour on which the mandible corners are located and the mandible motion direction in each TMJ image; and
constructing a TMJ anatomical model based on real mandible edge contours and temporal bone edge contours in all TMJ images;
wherein the step of obtaining the motion synchronization degree of the any two corners in each TMJ image based on the distance between the coordinates of each of the any two corners in the TMJ image and the coordinates of the corresponding corner in the adjacent TMJ image is implemented based on the following formula:

OG Complex Work Unit Math
wherein Bu,v represents a motion synchronization degree of a uth corner and a vth corner in each TMJ image; n represents a quantity of the TMJ images; Cu,i represents a distance between coordinates of the uth corner in an ith TMJ image and coordinates of a corresponding corner in an (i+1)th TMJ image in the coordinate system; Cv,i represents a distance between coordinates of the vth corner in the ith TMJ image and coordinates of a corresponding corner in the (i+1)th TMJ image in the coordinate system; and | | represents an absolute value function;
the step of obtaining the temporal bone corners and the temporal bone edge contour in each TMJ image based on the motion synchronization degree of and the distance between the any two corners in each TMJ image comprises the following steps:
obtaining a motion synchronization index between the any two corners in each TMJ image based on the motion synchronization degree of and the distance between the any two corners in each TMJ image; and
obtaining the temporal bone corners and the temporal bone edge contour in each TMJ image based on the motion synchronization index between the any two corners in each TMJ image;
the step of obtaining the motion synchronization index between the any two corners in each TMJ image based on the motion synchronization degree of and the distance between the any two corners in each TMJ image is implemented based on the following formula:

OG Complex Work Unit Math
wherein Eu,v represents a motion synchronization index between the uth corner and the vth corner in each TMJ image; Bu,v represents the motion synchronization degree of the uth corner and the vth corner in each TMJ image; n represents the quantity of the TMJ images; Li,u,v represents an Euclidean distance between the uth corner and the vth corner in the ith TMJ image; L represents an average Euclidean distance between the uth corner and the vth corner in all TMJ images; and | | represents the absolute value function;
the step of obtaining the temporal bone corners and the temporal bone edge contour in each TMJ image based on the motion synchronization index between the any two corners in each TMJ image comprises the following step:
in each TMJ image, recording the two corners corresponding to a minimum motion synchronization index as the temporal bone corners, and an edge contour on which the temporal bone corners are located as the temporal bone edge contour;
the step of obtaining the plurality of mandible corners and the mandible motion direction based on the distance between the temporal bone corners in each TMJ image and the coordinates of the corners in the TMJ image in the coordinate system comprises the following steps:
in each TMJ image, recording a center point of a line segment connecting two temporal bone corners as a reference point, performing clockwise rotation from a horizontal right direction to a horizontal left direction to construct a semicircle with the reference point as a circle center and y times a distance between the two temporal bone corners as a radius, and recording corners in the semicircle which are not temporal bone corners as suspected mandible corners, wherein y represents a preset coefficient;
in the coordinate system, forming a vector of a zth suspected mandible corner in the ith TMJ image based on a distance and a direction from coordinates of the zth suspected mandible corner in the ith TMJ image to coordinates of a corresponding corner in the (i+t)th TMJ image;
performing anomaly detection on all suspected mandible corners through a local outlier factor (LOF) algorithm based on a direction of a vector of each suspected mandible corner in the ith TMJ image, to obtain abnormal mandible corners; and recording all suspected mandible corners except the abnormal mandible corners as the mandible corners; and
recording a direction of a sum vector of the vectors of all mandible corners in the ith TMJ image as the mandible motion direction in the ith TMJ image;
the step of obtaining the real mandible edge contour in each TMJ image based on the edge contour on which the mandible corners are located and the mandible motion direction in each TMJ image comprises the following steps:
obtaining a similar mandibular edge index of each TMJ image based on the edge contour on which the mandible corners are located and the mandible motion direction in each TMJ image; and
obtaining the real mandible edge contour in each TMJ image based on the similar mandibular edge index of each TMJ image;
the step of obtaining the similar mandibular edge index of each TMJ image based on the edge contour on which the mandible corners are located and the mandible motion direction in each TMJ image comprises the following steps:
recording the edge contour on which the mandible corners are located in each TMJ image as a mandible edge contour;
obtaining a vector of a center of a minimum circumcircle of each mandible edge contour in each TMJ image in a manner of obtaining the vector of the zth suspected mandible corner in the ith TMJ image;
in each TMJ image, recording the vector of the center of the minimum circumcircle of each mandible edge contour as a motion vector of each mandible edge contour;
recording the ith to (i+t)th TMJ images as reference TMJ images corresponding to the ith TJM image;
recording a direction of a sum vector of the motion vectors of all mandible edge contours in all reference TMJ images as a mandible edge contour motion direction corresponding to the ith TMJ image; and
calculating a similar mandibular edge index of the ith TMJ image based on the following formula:

OG Complex Work Unit Math
wherein Pi represents the similar mandibular edge index of the ith TMJ image; θi′ represents a minimum included angle between the mandible motion direction and the mandible edge contour motion direction corresponding to the ith TMJ image; as represents an average curvature of all mandible edge contours in an sth reference TMJ image; as-1 represents an average curvature of all mandible edge contours in an (s-1)th reference TMJ image; | | represents the absolute value function; and M represents a quantity of the reference TMJ images; and
the step of obtaining the real mandible edge contour in each TMJ image based on the similar mandibular edge index of each TMJ image comprises the following step:
equally grouping all TMJ images into a plurality of TMJ image sequences, and using the mandible edge contour in the TMJ image corresponding to a smallest similar mandibular edge index in each TMJ image sequence as the real mandible edge contour in each TMJ image.