Retrospective study of deep learning to reduce noise in non-contrast head CT images. (December 2021)

• Record Type: Journal Article
• Authors: Wong, Kelvin K.; Cummock, Jonathon S.; He, Yunjie; Ghosh, Rahul; Volpi, John J.; Wong, Stephen T.C.
• Is Part Of: Computerized medical imaging and graphics. Volume 94(2021)
• Publication Date: 2021-12
• Subjects: NCCT non-contrast CT -- SNR signal-to-noise ratio -- CNR contrast-to-noise ratio -- BM3D block-matching and 3D filtering -- GM grey matter -- WM white matter -- DG deep grey matter -- AI artificial intelligence -- RED-CNN residual encoder-decoder convolutional neural network -- SRED-GCNN skip residual encoder-decoder group convolutional neural network; Deep learning -- Acute ischemic stroke -- Non-contrast head CT -- CT denoising; Diagnostic imaging -- Periodicals; Imaging systems in medicine -- Periodicals; Diagnosis, Radioscopic -- Data processing -- Periodicals; Diagnostic Imaging -- Periodicals; Imagerie pour le diagnostic -- Périodiques; Diagnostic imaging; Periodicals; Electronic journals; Electronic journals; 616.0754
• DOI: 10.1016/j.compmedimag.2021.101996
• Languages: English
• ISSNs: 0895-6111

Abstract

Abstract: Purpose: Presented herein is a novel CT denoising method uses a skip residual encoder-decoder framework with group convolutions and a novel loss function to improve the subjective and objective image quality for improved disease detection in patients with acute ischemic stroke (AIS). Materials and methods: In this retrospective study, confirmed AIS patients with full-dose NCCT head scans were randomly selected from a stroke registry between 2016 and 2020. 325 patients (67 ± 15 years, 176 men) were included. 18 patients each with 4–7 NCCTs performed within 5-day timeframe (83 total scans) were used for model training; 307 patients each with 1–4 NCCTs performed within 5-day timeframe (380 total scans) were used for hold-out testing. In the training group, a mean CT was created from the patient's co-registered scans for each input CT to train a rotation-reflection equivariant U-Net with skip and residual connections, as well as a group convolutional neural network (SRED-GCNN) using a custom loss function to remove image noise. Denoising performance was compared to the standard Block-matching and 3D filtering (BM3D) method and RED-CNN quantitatively and visually. Signal-to-noise ratio (SNR) and contrast-to-noise (CNR) were measured in manually drawn regions-of-interest in grey matter (GM), white matter (WM) and deep grey matter (DG). Visual comparison and impact on spatial resolution were assessed through phantom images. Results: SRED-GCNN reduced the original CT image noise significantly better than BM3D, with SNR improvements in GM, WM, and DG by 2.47x, 2.83x, and 2.64x respectively and CNR improvements in DG/WM and GM/WM by 2.30x and 2.16x respectively. Compared to the proposed SRED-GCNN, RED-CNN reduces noise effectively though the results are visibly blurred. Scans denoised by the SRED-GCNN are shown to be visually clearer with preserved anatomy. Conclusion: The proposed SRED-GCNN model significantly reduces image noise and improves signal-to-noise and contrast-to-noise ratios in 380 unseen head NCCT cases. Highlights: The proposed SRED-GCNN deep learning model greatly reduced noise in full-dose non-contrast head CTs both quantitatively and visually. Signal-to-noise and contrast-to-noise ratios were boosted 2–3x over the original CT with preserved anatomy. SRED-GCNN significantly outperformed the gold-standard block-matching 3D in denoising performance. SRED-GCNN has equivalent performance in preserving high frequency features as block-matching 3D whereas RED-CNN has significant degradation.