Dataset Information

Deep learning synthesis of cone-beam computed tomography from zero echo time magnetic resonance imaging.

ABSTRACT: Cone-beam computed tomography (CBCT) produces high-resolution of hard tissue even in small voxel size, but the process is associated with radiation exposure and poor soft tissue imaging. Thus, we synthesized a CBCT image from the magnetic resonance imaging (MRI), using deep learning and to assess its clinical accuracy. We collected patients who underwent both CBCT and MRI simultaneously in our institution (Seoul). MRI data were registered with CBCT data, and both data were prepared into 512 slices of axial, sagittal, and coronal sections. A deep learning-based synthesis model was trained and the output data were evaluated by comparing the original and synthetic CBCT (syCBCT). According to expert evaluation, syCBCT images showed better performance in terms of artifacts and noise criteria but had poor resolution compared to the original CBCT images. In syCBCT, hard tissue showed better clarity with significantly different MAE and SSIM. This study result would be a basis for replacing CBCT with non-radiation imaging that would be helpful for patients planning to undergo both MRI and CBCT.

SUBMITTER: Choi H

PROVIDER: S-EPMC10102229 | biostudies-literature | 2023 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Deep learning synthesis of cone-beam computed tomography from zero echo time magnetic resonance imaging.

Choi Hyeyeon H Yun Jong Pil JP Lee Ari A Han Sang-Sun SS Kim Sang Woo SW Lee Chena C

Scientific reports 20230413 1

Cone-beam computed tomography (CBCT) produces high-resolution of hard tissue even in small voxel size, but the process is associated with radiation exposure and poor soft tissue imaging. Thus, we synthesized a CBCT image from the magnetic resonance imaging (MRI), using deep learning and to assess its clinical accuracy. We collected patients who underwent both CBCT and MRI simultaneously in our institution (Seoul). MRI data were registered with CBCT data, and both data were prepared into 512 slic ...[more]

PMID: 37055501

Similar Datasets

Project description:BackgroundAccurate preoperative assessment of tumor size is important in developing a surgical plan for breast cancer. The purpose of this study was to evaluate the accuracy of cone-beam breast computed tomography (CBBCT) and magnetic resonance imaging (MRI) in the assessment of tumor size and to analyze the factors influencing the discordance.MethodsIn this retrospective study, patients with breast cancer who underwent preoperative contrast-enhanced CBBCT (CE-CBBCT) and dynamic contrast-enhanced MRI (DCE-MRI) and received a complete pathologic diagnosis from August 2020 to December 2021 were included, using the pathological result as the gold standard. Two radiologists assessed the CBBCT and MRI features and measured the tumor size with a 2-week washout period. Intraclass correlation coefficient (ICC) and Bland-Altman analyses were used to assess inter-observer reproducibility and agreement based on CBBCT, MRI and pathology. Univariate analyses of differences in clinical, pathological and CBBCT/MRI features between the concordant and discordant groups was performed using the t-test, Mann-Whitney U-test, Chi-squared test and Fisher's exact test. Multivariate analyses were used to identify factors associated with discordance of CBBCT/MRI with pathology.ResultsA total of 115 female breast cancer patients (115 lesions) were included. All patients had a single malignant tumor of the unilateral breast. The reproducibility and the agreement ranged from moderate to excellent (ICC =0.607-0.983). Receiver operating characteristic (ROC) analyses showed that the cut-off values of CBBCT-pathology and MRI-pathology discordance were 2.25 and 2.65 cm, respectively. CBBCT/MRI-pathology concordance was significantly associated with the extent of pathology, lesion type, presence of calcification, human epidermal growth factor receptor 2 (HER2) status and fatty infiltration (P<0.05). In lesions containing calcification, the difference of CBBCT-pathology was significantly smaller than MRI-pathology (P=0.021). Non-mass enhancement (NME) was the main predictor of CBBCT- or MRI-pathology discordance [odds ratio (OR) =3.293-6.469, P<0.05], and HER2 positivity was a predictor of CBBCT-pathology discordance (OR =3.514, P=0.019).ConclusionsCBBCT and MRI have comparable accuracy in measurement of tumor size, and CBBCT is advantageous in assessing the size of calcified lesions. NME and HER2 positivity are significant predictors of CBBCT-pathology discordance. This suggests that CBBCT might serve as an alternative imaging technique to assess tumor size when patients do not tolerate MRI.

Dataset Information

Deep learning synthesis of cone-beam computed tomography from zero echo time magnetic resonance imaging.

Publications

Deep learning synthesis of cone-beam computed tomography from zero echo time magnetic resonance imaging.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets