電腦輔助皮質骨軌跡螺釘植入：以粒子群最佳化演算法提升 螺釘植入準確度

Abstract

椎弓根螺釘（Pedicle Screws, PS）固定手術是脊椎手術中的主要技術。然而，對於骨質密度較低或骨質疏鬆的患者，傳統螺釘植入方式面臨了固定效果不足和鬆脫的風險。本研究關注近年發展的皮質骨軌道（Cortical Bone Trajectory, CBT）螺釘植入技術，此技術由Santoni研究團隊於2009年提出。CBT技術透過將螺釘路徑調整為矢狀面上的尾向頭及橫切面上的由內向外方向，能夠有效增加螺釘與皮質骨的接觸面積，進而強化螺釘的固定力和界面強度。相較於傳統螺釘植入方式，CBT螺釘具備更優異的固定能力、恢復期較短、減低術後併發症等優勢。 然而，儘管CBT優勢顯著，但其植入路徑之準確性是決定手術成敗的關鍵。由於CBT螺釘植入角度與傳統技術的手感差異大，難以單純透過術者的經驗精準掌握，若植入不當，可能危及脊髓、神經根及椎動脈等周圍重要結構。因此，術前精確規劃螺釘植入路徑成為確保手術安全性和固定效果的重要前提。 為了提升CBT螺釘植入的準確性、安全性和固定效果，本研究旨在開發一套電腦輔助的自動化術前規劃系統。我們利用醫學影像技術結合人工智慧，以粒子群最佳化（Particle Swarm Optimization, PSO）演算法作為核心工具，根據每位患者的CT影像、其獨特的脊椎解剖結構和臨床需求，量身制定皮質骨軌道螺植入之最佳位置。此系統的開發旨在協助臨床醫師進行更精準的術前決策，從而提升手術成功率並且降低併發症發生的風險。

Pedicle Screw (PS) fixation is a widely applied technique in spinal surgery. However, in patients with low bone mineral density or osteoporosis, conventional screw placement often faces challenges related to poor fixation strength and the risk of loosening. This study focuses on the more recently developed implant technique: Cortical Bone Trajectory (CBT), first proposed by Santoni's research team in 2009. The CBT technique modifies the screw trajectory to follow a caudal-to-cephalad path in the sagittal plane and a medial-to-lateral path in the axial plane. This innovation significantly increases the contact area between the screw thread and the dense cortical bone, thereby enhancing screw pullout strength and interface rigidity. Compared to traditional spinal fusion and fixation surgeries, CBT screws offer superior fixation capacity, minimal invasiveness, and a shorter recovery period. Despite its advantages, the success of CBT is critically dependent on the accuracy of the screw insertion trajectory. Since the tactile feedback (hand-feel) of CBT placement differs substantially from that of traditional PS placement, accurate spatial orientation becomes challenging. Incorrect placement poses severe risks to surrounding vital structures, including the spinal cord, nerve roots, and vertebral arteries. Therefore, precise pre-operative planning is an essential step to ensure optimal fixation and minimize the risk of complications. To improve the efficacy, accuracy, and safety of CBT screw placement, this research aims to develop a computer-aided, automated pre-operative planning system. We integrate CT images (Computed Tomography images) with artificial intelligence, utilizing the Particle Swarm Optimization (PSO) algorithm as the core computational tool. This system is designed to individually tailor the optimal position and trajectory for CBT pedicle screw placement based on each patient's unique anatomical structure and specific needs. The goal is to assist surgeons in precise pre-operative decision-making, enhance screw placement accuracy, and ultimately reduce the incidence of complications.