Neural Implicit k-space with Trainable Periodic Activation Functions for Cardiac MR Imaging

Patrick T. Haft, Wenqi Huang, Gastao Cruz, Daniel Rueckert, Veronika A. Zimmer, Kerstin Hammernik

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

In MRI reconstruction, neural implicit k-space (NIK) representation maps spatial frequencies to k-space intensity values using an MLP with periodic activation functions. However, the choice of hyperparameters for periodic activation functions is challenging and influences training stability. In this work, we introduce and study the effectiveness of trainable (non-)periodic activation functions for NIK in the context of non-Cartesian Cardiac MRI. Evaluated on 42 radially sampled datasets from 6 subjects, NIKs with the proposed trainable activation functions outperform qualitatively and quantitatively other state-of-the-art reconstruction methods, including NIK with fixed periodic activation functions.

Original languageEnglish
Title of host publicationBildverarbeitung für die Medizin 2024 - Proceedings, German Conference on Medical Image Computing, 2024
EditorsAndreas Maier, Thomas M. Deserno, Heinz Handels, Klaus Maier-Hein, Christoph Palm, Thomas Tolxdorff
PublisherSpringer Science and Business Media Deutschland GmbH
Pages82-87
Number of pages6
ISBN (Print)9783658440367
DOIs
StatePublished - 2024
EventGerman Conference on Medical Image Computing, BVM 2024 - Erlangen, Germany
Duration: 10 Mar 202412 Mar 2024

Publication series

NameInformatik aktuell
ISSN (Print)1431-472X

Conference

ConferenceGerman Conference on Medical Image Computing, BVM 2024
Country/TerritoryGermany
CityErlangen
Period10/03/2412/03/24

Fingerprint

Dive into the research topics of 'Neural Implicit k-space with Trainable Periodic Activation Functions for Cardiac MR Imaging'. Together they form a unique fingerprint.

Cite this