【摘要】 骨折修复手术中，需要将螺钉置入远端孔以固定髓内钉。但由于远端铰锁孔位姿会随髓腔形状发生变化，因此难以准确瞄准远端铰锁孔。传统定位方法不能较好满足医生术中锁钉实际需求，费时费力且难以精确定位，目前，该手术的成功率只有30%，手术难点在于对精度要求高，误差需控制在2 mm以内。论文研究基于先验知识融合的定位方法，首先利用两张不同位置的X线片，深度学习回归判断远端铰锁孔轴线的投影；之后通过构造投影平面，根据双平面相交的方法，确定远端铰锁孔轴线的初始空间位姿；最后利用孔轮廓信息迭代修正轴线位姿。将该方法计算得到的轴线位姿与真实轴线位姿进行比较，实验结果表明，两轴线平均距离误差为0.82 mm，平均角度误差为0.75°。该方法能满足远端铰锁孔定位的实际需求，有效提高术中定位与规划效率。更多还原

【Abstract】 In the repair of human long bone fracture,the intramedullary nail needs to be inserted into the medullary cavity in order to fix the fracture area. The difficulty is that distal interlocking hole pose will change with the shape of medullary cavity. Therefore,it is difficult to target the distal interlocking hole with the matching aiming instrument. The traditional location method cannot well meet the actual needs of intraoperative locking nails,which is time-consuming,laborious and difficult to locate accurately,and the success rate is only 30%. High precision is required,and the error needs to be controlled within 2 mm. Firstly,two intraoperative X-ray images are taken at different angles with the C-arm machine,and the deep neural network is used to accurately predict the projection of the hole’s axis in the imaging plane. Then,the initial spatial pose of the hole’s axis is determined by constructing the projection plane and according to the intersection of the two planes. Finally,the nail’s contour information is used to iteratively correct the axis pose. The experiments are carried out in the simulated and clinical environments. Comparison is drawn between the calculated hole’s axis and the actual hole’s axis. Results show that the average distance error is 0.82 mm,and the average angle error is 0.75°. The method can meet the actual surgical needs,and improves the efficiency of locating and planning in the distal interlocking intramedullary nail surgery.更多还原