Title: An Artificial Agent for Robust Image Registration (Thirty-First AAAI Conference on Artificial Intelligence (AAAI-17)) Paper Link
Authors: Rui Liao, Shun Miao, Pierre de Tournemire, Sasa Grbic, Ali Kamen, Tommaso Mansi, Dorin Comaniciu
Association: Technology Center, Medical Imaging Technologies Siemens Medical Solutions USA
Submission: 2017
…………………………………………………………………………………………………………………
»> Background of Deep Learning in Medical Image Analysis
»> A Review: Robot-Assisted Endovascular Catheterization Technologies:
…………………………………………………………………………………………………………………
»> Image Registration Basic Knowledge
»> Image Registration Literature Review
»> Slice-To-Volume Medical Image Registration Background
…………………………………………………………………………………………………………………
Motivation
Contributions
(0) decompose the registration task into a se- quence of (often easier) classification problems, i.e. find- ing the best action among a limited set of possible solutions to improve the alignment. Repeating this process can result in a converging solution.
(1) We train the intelligent agent in a greedy supervised fashion, which is a magni- tude more efficient with only a small fraction of the mem- ory footprint than in standard DRL setup, where the agent learns through repeated trial and errors (Mnih et al. 2015; Silver et al. 2016); 3)
(2) propose an effective data aug- mentation and sampling strategy so that the agent could be trained robustly using only a small number of labelled train- ing pairs available from patients;
(3) For a combined robust- ness and accuracy, we propose a hierarchical registration framework relying on the trained networks from the coarse image layer to register successively the more refined (higher- resolution) image layers
Methods
