Download PDFOpen PDF in browser

A Deep Learning Approach for Single Shot C-Arm Pose Estimation

5 pagesPublished: September 25, 2020

Abstract

During a typical fluoroscopic guided surgery, it is common to acquire multiple x-ray images to correctly position the C-arm. This can be a long process resulting in an in- crease in operation time and ionizing radiation exposure. Our purpose in this study is to implement a machine learning system for predicting the position of the C-arm based on the intraoperative radiographs. The prediction is achieved by training a Deep Learning Network based on Digitally Reconstructed Radiographs. We first showed a high prediction accuracy (4.5 mm and 1.1o) when patient-specific training was implemented. Additionally, we demonstrated a similar range of accuracy by applying transfer-learning on the last lay- ers of the network while reducing the processing time by 83%. In conclusion, in this study, we propose a C-arm position prediction system based on machine learning that can po- tentially reduce the number of intraoperatively acquired X-rays in a common orthopaedic surgical procedure.

Keyphrases: c arm, convolutional neural network, deep learning, pelvis, pose estimation, transfer learning

In: Ferdinando Rodriguez Y Baena and Fabio Tatti (editors). CAOS 2020. The 20th Annual Meeting of the International Society for Computer Assisted Orthopaedic Surgery, vol 4, pages 69-73.

BibTeX entry
@inproceedings{CAOS2020:Deep_Learning_Approach_Single,
  author    = {Hooman Esfandiari and Sebastian Andreß and Maternus Herold and Wolfgang Böcker and Simon Weidert and Antony J Hodgson},
  title     = {A Deep Learning Approach for Single Shot C-Arm Pose Estimation},
  booktitle = {CAOS 2020. The 20th Annual Meeting of the International Society for Computer Assisted Orthopaedic Surgery},
  editor    = {Ferdinando Rodriguez Y Baena and Fabio Tatti},
  series    = {EPiC Series in Health Sciences},
  volume    = {4},
  publisher = {EasyChair},
  bibsource = {EasyChair, https://easychair.org},
  issn      = {2398-5305},
  url       = {/publications/paper/MmwT},
  doi       = {10.29007/6mkk},
  pages     = {69-73},
  year      = {2020}}
Download PDFOpen PDF in browser