Synthesizing 3D computed tomography from MRI or CBCT using 2.5D deep neural networks

TL;DR

This paper presents a novel 2.5D CNN-based method for synthesizing CT images from MRI and CBCT scans, demonstrating advantages over 3D CNNs in volumetric data processing for brain and pelvis imaging.

Contribution

Introduces a 2.5D CNN approach for synthetic CT generation from MRI and CBCT, highlighting its benefits over traditional 3D CNNs.

Findings

Effective MRI-to-sCT and CBCT-to-sCT synthesis for brain and pelvis.

2.5D CNNs outperform 3D CNNs in volumetric data tasks.

Potential for improved clinical imaging workflows.

Abstract

Deep learning techniques, particularly convolutional neural networks (CNNs), have gained traction for synthetic computed tomography (sCT) generation from Magnetic resonance imaging (MRI), Cone-beam computed tomography (CBCT) and PET. In this report, we introduce a method to syn-thesize CT from MRI or CBCT. Our method is based on multi-slice (2.5D) CNNs. 2.5D CNNs offer distinct advantages over 3D CNNs when dealing with volumetric data. In the experiments, we evaluate the performance of our method for two tasks, MRI-to-sCT and CBCT-to-sCT generation. Target organs for both tasks are brain and pelvis.