Deep learning reconstruction for improving the visualization of acute brain infarct on computed tomography

Abstract

Purpose This study aimed to investigate the impact of deep learning reconstruction (DLR) on acute infarct depiction compared with hybrid iterative reconstruction (Hybrid IR). Methods This retrospective study included 29 (75.8 ± 13.2 years, 20 males) and 26 (64.4 ± 12.4 years, 18 males) patients with and without acute infarction, respectively. Unenhanced head CT images were reconstructed with DLR and Hybrid IR. In qualitative analyses, three readers evaluated the conspicuity of lesions based on five regions and image quality. A radiologist placed regions of interest on the lateral ventricle, putamen, and white matter in quantitative analyses, and the standard deviation of CT attenuation (i.e., quantitative image noise) was recorded. Results Conspicuity of acute infarct in DLR was superior to that in Hybrid IR, and a statistically significant difference was observed for two readers (p ≤ 0.038). Conspicuity of acute infarct with time from onset to CT imaging at < 24 h in DLR was significantly improved compared with Hybrid IR for all readers (p ≤ 0.020). Image noise in DLR was significantly reduced compared with Hybrid IR with both the qualitative and quantitative analyses (p < 0.001 for all). Conclusion DLR in head CT helped improve acute infarct depiction, especially those with time from onset to CT imaging at < 24 h.

Deep-Learning Reconstruction of High-Resolution CT Improves Interobserver Agreement for the Evaluation of Pulmonary Fibrosis

“…A higher interobserver agreement in the evaluation of ILD for DLR would have been resulted from an improved image quality. A better image noise reduction can be achieved compared to HIR in various regions such as the head, 19,27 vertebra, 21,23,24 and abdomen 20,22 as well as the chest. 25 Our study result corroborates with other studies, which revealed a significantly reduced quantitative image noise in DLR compared to HIR.…”

“…It was trained with paired data, one with low noise and high spatial resolution and another with noise contamination and learned the difference between noise and signal. 17,18 It has been reported that the appropriate algorithm of DLR can improve image quality of CT images in various body regions 19-24 including the lung, 25 we hypothesized that this reconstruction algorithm could also potentially improve the interobserver agreement in the evaluation of honeycombing in patients with ILD.…”

Deep learning reconstruction for high-resolution computed tomography images of the temporal bone: comparison with hybrid iterative reconstruction

“…In supervised training of DLR, low-quality data and high-quality images are respectively used as the input and reference data. A trained DLR is known to reduce image noise [ 19 ], thereby improving the quality of CT images in regions as varied as the head [ 20 , 21 ], chest [ 9 ], abdomen [ 10 , 19 , 22 , 23 ], and vertebrae [ 11 , 24 , 25 ]. We therefore hypothesized that DLR can also improve the quality of high-resolution CT of the temporal bone.…”