A Deep Learning Approach to MR-less Spatial Normalization for Tau PET Images
Paper i proceeding, 2019

The procedure of aligning a positron emission tomography (PET) image with a common coordinate system, spatial normalization, typically demands a corresponding structural magnetic resonance (MR) image. However, MR imaging is not always available or feasible for the subject, which calls for enabling spatial normalization without MR, MR-less spatial normalization. In this work, we propose a template-free approach to MR-less spatial normalization for [18F]flortaucipir tau PET images. We use a deep neural network that estimates an aligning transformation from the PET input image, and outputs the spatially normalized image as well as the parameterized transformation. In order to do so, the proposed network iteratively estimates a set of rigid and affine transformations by means of convolutional neural network regressors as well as spatial transformer layers. The network is trained and validated on 199 tau PET volumes with corresponding ground truth transformations, and tested on two different datasets. The proposed method shows competitive performance in terms of registration accuracy as well as speed, and compares favourably to previously published results.


Jennifer Alvén

Chalmers, Elektroteknik, Signalbehandling och medicinsk teknik

Kerstin Heurling

Göteborgs universitet

Ruben Smith

Lunds universitet

Olof Strandberg

Lunds universitet

Michael Schöll

Göteborgs universitet

Oskar Hansson

Lunds universitet

Fredrik Kahl

Chalmers, Elektroteknik, Signalbehandling och medicinsk teknik

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

03029743 (ISSN) 16113349 (eISSN)

Vol. 11765 355-363
9783030322441 (ISBN)

International Conference on Medical Image Computing and Computer-Assisted Intervention
Shenzhen, China,


Informations- och kommunikationsteknik

Livsvetenskaper och teknik (2010-2018)


Datorseende och robotik (autonoma system)

Medicinsk bildbehandling



Mer information

Senast uppdaterat