Deep Learning Accelerated Image Reconstruction of Fluid-Attenuated Inversion Recovery Sequence in Brain Imaging: Reduction of Acquisition Time and Improvement of Image Quality

Arne Estler; Till-Karsten Hauser; Annerose Mengel; Merle Brunnée; Leonie Zerweck; Vivien Richter; Mario Zuena; Martin Schuhholz; Ulrike Ernemann; Georg Gohla

doi:10.1016/j.acra.2023.05.010

Deep Learning Accelerated Image Reconstruction of Fluid-Attenuated Inversion Recovery Sequence in Brain Imaging: Reduction of Acquisition Time and Improvement of Image Quality

Acad Radiol. 2024 Jan;31(1):180-186. doi: 10.1016/j.acra.2023.05.010. Epub 2023 Jun 4.

Authors

Affiliations

¹ Diagnostic and Interventional Neuroradiology, Department of Radiology, University Hospital Tuebingen, Hoppe-Seyler-Str. 3, 72076 Tübingen, Baden-Württemberg, Germany (A.E., T.-K.H., L.Z., V.R., M.Z., U.E., G.G.). Electronic address: Arne.estler@med.uni-tuebingen.de.
² Diagnostic and Interventional Neuroradiology, Department of Radiology, University Hospital Tuebingen, Hoppe-Seyler-Str. 3, 72076 Tübingen, Baden-Württemberg, Germany (A.E., T.-K.H., L.Z., V.R., M.Z., U.E., G.G.).
³ Department of Neurology & Stroke, Eberhard-Karls University of Tübingen, Tuebingen, Germany (A.M.).
⁴ Department of Neuroradiology, Neurological University Clinic, Heidelberg University Hospital, Heidelberg, Germany (M.B.).
⁵ Faculty of Medicine, University of Tuebingen, Tübingen, Germany (M.S.).

PMID: 37280126
DOI: 10.1016/j.acra.2023.05.010

Abstract

Rationale and objectives: Fluid-attenuated inversion recovery (FLAIR) imaging is playing an increasingly significant role in the detection of brain metastases with a concomitant increase in the number of magnetic resonance imaging (MRI) examinations. Therefore, the purpose of this study was to investigate the impact on image quality and diagnostic confidence of an innovative deep learning-based accelerated FLAIR (FLAIR_DLR) sequence of the brain compared to conventional (standard) FLAIR (FLAIR_S) imaging.

Materials and methods: Seventy consecutive patients with staging cerebral MRIs were retrospectively enrolled in this single-center study. The FLAIR_DLR was conducted using the same MRI acquisition parameters as the FLAIR_S sequence, except for a higher acceleration factor for parallel imaging (from 2 to 4), which resulted in a shorter acquisition time of 1:39 minute instead of 2:40 minutes (-38%). Two specialized neuroradiologists evaluated the imaging datasets using a Likert scale that ranged from 1 to 4, with 4 indicating the best score for the following parameters: sharpness, lesion demarcation, artifacts, overall image quality, and diagnostic confidence. Additionally, the image preference of the readers and the interreader agreement were assessed.

Results: The average age of the patients was 63 ± 11years. FLAIR_DLR exhibited significantly less image noise than FLAIR_S, with P-values of< .001 and< .05, respectively. The sharpness of the images and the ability to detect lesions were rated higher in FLAIR_DLR, with a median score of 4 compared to a median score of 3 in FLAIR_S (P-values of<.001 for both readers). In terms of overall image quality, FLAIR_DLR was rated superior to FLAIR_S, with a median score of 4 vs 3 (P-values of<.001 for both readers). Both readers preferred FLAIR_DLR in 68/70 cases.

Conclusion: The feasibility of deep learning FLAIR brain imaging was shown with additional 38% reduction in examination time compared to standard FLAIR imaging. Furthermore, this technique has shown improvement in image quality, noise reduction, and lesion demarcation.

Keywords: Brain imaging; Deep learning; FLAIR.

MeSH terms

Aged
Artifacts
Brain / diagnostic imaging
Brain / pathology
Brain Neoplasms* / pathology
Deep Learning*
Humans
Image Processing, Computer-Assisted / methods
Magnetic Resonance Imaging / methods
Middle Aged
Retrospective Studies