Pedro D Batista1, Inês P Machado2, Pedro Roios2, José Lavrador3, Maria B Cattoni4, Jorge Martins2, Herculano Carvalho4. 1. Department of Neurosurgery, Hospital de Santa Maria, CHLN, Lisbon, Portugal. Electronic address: pedroduartebatista@gmail.com. 2. IDMEC/LAETA, Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal. 3. Department of Neurosurgery, Hospital de Santa Maria, CHLN, Lisbon, Portugal; Department of Adult and Paediatric Neurosurgery, King's College Hospital, Foundation Trust, London, United Kingdom. 4. Department of Neurosurgery, Hospital de Santa Maria, CHLN, Lisbon, Portugal.
Abstract
OBJECTIVE: Neuronavigation procedures demand high precision and accuracy. Despite this need, there are still few studies analyzing errors in such procedures. The aim of this study was to use a custom-built cranial phantom to measure target position and orientation errors in different phases of a simulated neuronavigation procedure. METHODS: A cranial phantom with 10 target sites was designed and imaged with computed tomography and magnetic resonance. A segmentation of a cloud of points of the phantom (ground truth) was obtained using an optical tracking system and compared with the images (imaging phase). Targets and trajectories were then planned with neuronavigation software and compared with the ground truth (planning phase). The same plan was used to identify the points in real space after image-to-phantom registration and calculate the final error of the procedure by comparison with the ground truth (registration and execution phase). RESULTS: The mean errors after the imaging phase were 1.11 ± 0.42 mm and 3.23° ± 1.69° for position and orientation, respectively. After planning the mean errors were 1.10 ± 0.39 mm and 5.55° ± 2.91°. The global errors after the registration and mechanical execution were 3.93 ± 1.70 mm and 3.65° ± 1.29°. CONCLUSIONS: After a stepwise analysis, registration and mechanical execution were the main contributors to the global position error.
OBJECTIVE: Neuronavigation procedures demand high precision and accuracy. Despite this need, there are still few studies analyzing errors in such procedures. The aim of this study was to use a custom-built cranial phantom to measure target position and orientation errors in different phases of a simulated neuronavigation procedure. METHODS: A cranial phantom with 10 target sites was designed and imaged with computed tomography and magnetic resonance. A segmentation of a cloud of points of the phantom (ground truth) was obtained using an optical tracking system and compared with the images (imaging phase). Targets and trajectories were then planned with neuronavigation software and compared with the ground truth (planning phase). The same plan was used to identify the points in real space after image-to-phantom registration and calculate the final error of the procedure by comparison with the ground truth (registration and execution phase). RESULTS: The mean errors after the imaging phase were 1.11 ± 0.42 mm and 3.23° ± 1.69° for position and orientation, respectively. After planning the mean errors were 1.10 ± 0.39 mm and 5.55° ± 2.91°. The global errors after the registration and mechanical execution were 3.93 ± 1.70 mm and 3.65° ± 1.29°. CONCLUSIONS: After a stepwise analysis, registration and mechanical execution were the main contributors to the global position error.