Objective To explore the efficacy of extracting biopsy needle pose in chest CT images based on point cloud processing.Methods Three-dimensional point clouds were generated through segmentation of chest CT images and surface reconstruction.Spatial point cloud clustering and geometric constraints were applied to filter regions contained the puncture needle in space.The principal direction of the needle was judged using principal component analysis,and a cylindrical model was constructed to enclose the needle data.Then random sample consensus algorithm was used for needle trajectory fitting to accurately extract the spatial position and orientation of the puncture needle.The efficacy of the above method was evaluated using a 3D-printed anatomical model based on common clinical combinations of puncture depths and angles.Results The anatomical model experiments showed a 100％success rate in puncture needle identification,with angular error of(1.013±0.424)° and positional error of(2.023±1.553)mm,indicating that this method had good accuracy and stability.Conclusion The puncture needle's position in chest CT images could be extracted with high precision based on point cloud processing.

Objective To explore the efficacy of extracting biopsy needle pose in chest CT images based on point cloud processing.Methods Three-dimensional point clouds were generated through segmentation of chest CT images and surface reconstruction.Spatial point cloud clustering and geometric constraints were applied to filter regions contained the puncture needle in space.The principal direction of the needle was judged using principal component analysis,and a cylindrical model was constructed to enclose the needle data.Then random sample consensus algorithm was used for needle trajectory fitting to accurately extract the spatial position and orientation of the puncture needle.The efficacy of the above method was evaluated using a 3D-printed anatomical model based on common clinical combinations of puncture depths and angles.Results The anatomical model experiments showed a 100％success rate in puncture needle identification,with angular error of(1.013±0.424)° and positional error of(2.023±1.553)mm,indicating that this method had good accuracy and stability.Conclusion The puncture needle's position in chest CT images could be extracted with high precision based on point cloud processing.