Your browser doesn't support javascript.
loading
Medical image registration via neural fields.
Sun, Shanlin; Han, Kun; You, Chenyu; Tang, Hao; Kong, Deying; Naushad, Junayed; Yan, Xiangyi; Ma, Haoyu; Khosravi, Pooya; Duncan, James S; Xie, Xiaohui.
Afiliación
  • Sun S; University of California, Irvine, Irvine, CA 92697, USA. Electronic address: shanlins@uci.edu.
  • Han K; University of California, Irvine, Irvine, CA 92697, USA. Electronic address: kunh7@uci.edu.
  • You C; Yale University, New Haven, CT 06520, USA. Electronic address: chenyu.you@yale.edu.
  • Tang H; University of California, Irvine, Irvine, CA 92697, USA. Electronic address: htang6@uci.edu.
  • Kong D; University of California, Irvine, Irvine, CA 92697, USA. Electronic address: deyingk@uci.edu.
  • Naushad J; University of California, Irvine, Irvine, CA 92697, USA. Electronic address: jnaushad@uci.edu.
  • Yan X; University of California, Irvine, Irvine, CA 92697, USA. Electronic address: xiangyy4@uci.edu.
  • Ma H; University of California, Irvine, Irvine, CA 92697, USA. Electronic address: haoyum3@uci.edu.
  • Khosravi P; University of California, Irvine, Irvine, CA 92697, USA. Electronic address: pooyak@hs.uci.edu.
  • Duncan JS; Yale University, New Haven, CT 06520, USA. Electronic address: james.duncan@yale.edu.
  • Xie X; University of California, Irvine, Irvine, CA 92697, USA. Electronic address: xhx@uci.edu.
Med Image Anal ; 97: 103249, 2024 Oct.
Article en En | MEDLINE | ID: mdl-38963972
ABSTRACT
Image registration is an essential step in many medical image analysis tasks. Traditional methods for image registration are primarily optimization-driven, finding the optimal deformations that maximize the similarity between two images. Recent learning-based methods, trained to directly predict transformations between two images, run much faster, but suffer from performance deficiencies due to domain shift. Here we present a new neural network based image registration framework, called NIR (Neural Image Registration), which is based on optimization but utilizes deep neural networks to model deformations between image pairs. NIR represents the transformation between two images with a continuous function implemented via neural fields, receiving a 3D coordinate as input and outputting the corresponding deformation vector. NIR provides two ways of generating deformation field directly output a displacement vector field for general deformable registration, or output a velocity vector field and integrate the velocity field to derive the deformation field for diffeomorphic image registration. The optimal registration is discovered by updating the parameters of the neural field via stochastic mini-batch gradient descent. We describe several design choices that facilitate model optimization, including coordinate encoding, sinusoidal activation, coordinate sampling, and intensity sampling. NIR is evaluated on two 3D MR brain scan datasets, demonstrating highly competitive performance in terms of both registration accuracy and regularity. Compared to traditional optimization-based methods, our approach achieves better results in shorter computation times. In addition, our methods exhibit performance on a cross-dataset registration task, compared to the pre-trained learning-based methods.
Asunto(s)
Palabras clave
Texto completo
Añadir a Mi BVS
Imprimir
XML
PubMed Links
Buscar en Google

Texto completo: 1 Colección: 01-internacional Base de datos: MEDLINE Asunto principal: Imagen por Resonancia Magnética / Redes Neurales de la Computación / Imagenología Tridimensional Límite: Humans Idioma: En Revista: Med Image Anal Asunto de la revista: DIAGNOSTICO POR IMAGEM Año: 2024 Tipo del documento: Article Pais de publicación: Países Bajos
Texto completo
Añadir a Mi BVS
Imprimir
XML
PubMed Links
Buscar en Google

Texto completo: 1 Colección: 01-internacional Base de datos: MEDLINE Asunto principal: Imagen por Resonancia Magnética / Redes Neurales de la Computación / Imagenología Tridimensional Límite: Humans Idioma: En Revista: Med Image Anal Asunto de la revista: DIAGNOSTICO POR IMAGEM Año: 2024 Tipo del documento: Article Pais de publicación: Países Bajos