Super Resolution Convolutional Neural Networks for Increasing Spatial Resolution of 1H Magnetic Resonance Spectroscopic Imaging

Sevim Cengiz, Maria Valdes Hernandez, Esin Ozturk-Isik

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Proton magnetic resonance spectroscopic imaging (1H-MRSI) provides noninvasive information regarding metabolic activity within the tissues. One of the main problems of MRSI is low spatial resolution due to clinical scan time limitations. Advanced post-processsing algorithms, like convolutional neural networks (CNN) might help with generation of super resolution MR spectroscopic images. In this study, the application of super resolution convolutional neural networks (SRCNN) for increasing the MRSI spatial resolution is presented. FLAIR, T1 weighted and T2 weighted MR images were used in training the SRCNN scheme. The spatial resolution of MRSI images were increased by using the model trained with the anatomical MR images. The results of the proposed technique were compared with bicubic resampling in terms of peak signal to noise ratio, structure similarity index, root mean square error, relative polar edge coherence, and visual information delity pixel. Our results indicated that SRCNN would contribute to reconstructing higher resolution MRSI.
Original languageEnglish
Title of host publicationMedical Image Understanding and Analysis
Subtitle of host publication21st Annual Conference, MIUA 2017, Edinburgh, UK, July 11–13, 2017, Proceedings.
Pages641-650
DOIs
Publication statusE-pub ahead of print - 22 Jun 2017

Publication series

NameCommunications in Computer and Information Science

Fingerprint Dive into the research topics of 'Super Resolution Convolutional Neural Networks for Increasing Spatial Resolution of 1H Magnetic Resonance Spectroscopic Imaging'. Together they form a unique fingerprint.

Cite this