Title: A CNN Regression Approach for Real-Time 2D/3D Registration Paper Link

Authors: Bob D. de Vos, Floris F. Berendsen, Max A. Viergever, Marius Staring, Ivana Išgum

Association: Image Sciences Institute, University Medical Center Utrecht, the Netherlands

Submission: September 2017

Image registration is a key component for medical image analysis to provide spatial correspondences.

…………………………………………………………………………………………………………………

»> Background of Deep Learning in Medical Image Analysis

[NEED A Reading Notes]

»> A Review: Robot-Assisted Endovascular Catheterization Technologies:

[NEED A Reading Notes]

…………………………………………………………………………………………………………………

»> Image Registration Basic Knowledge

[Basic Knowledge]

»> Image Registration Literature Review

[Literature Review]

»> Slice-To-Volume Medical Image Registration Background

[NEED A Reading Notes]

…………………………………………………………………………………………………………………

Contributions

(0) Unlike previous methods, the DIRNet is not trained with known registration transformations, but learns to register images by directly optimizing a similarity metric between the fixed and the moving image.

(1) propose a deep learning network for de- formable image registration (DIRNet), which consists of a convolutional neural network (ConvNet) regressor, a spatial transformer, and a resampler.

(2) generates the displacement vector field that enables the resampler to warp the moving image to the fixed image.

(3) end-to-end by unsupervised optimization of a similarity metric between input image pairs.

(4) A trained DIRNet can be applied to perform registration on unseen image pairs in one pass, thus non-iteratively. Evaluation

DIRNet

Performance

DIRNet had been evaluated with registration of images with handwritten digits and image slices from cine cardiac MRI scans.