Accurate and timely organs-at-risk (OARs) segmentation is key to efficient high-quality radiation therapy planning. The purpose of this work develop a deep learning-based method automatically segment multiple thoracic OARs on chest computed tomography (CT) for radiotherapy treatment planning.We propose an adversarial training strategy train neural networks the organs CT images. proposed design networks, called U-Net-generative network (U-Net-GAN), jointly trains set U-Nets as generators...

Automated synthetic computed tomography (sCT) generation based on magnetic resonance imaging (MRI) images would allow for MRI-only treatment planning in radiation therapy, eliminating the need CT simulation and simplifying patient workflow. In this work, authors propose a novel method of sCT dense cycle-consistent generative adversarial networks (cycle GAN), deep-learning model that trains two transformation mappings (MRI to MRI) simultaneously.The cycle GAN-based was developed generate...

Purpose The incorporation of cone‐beam computed tomography (CBCT) has allowed for enhanced image‐guided radiation therapy. While CBCT allows daily 3D imaging, images suffer from severe artifacts, limiting the clinical potential CBCT. In this work, a deep learning‐based method generating high quality corrected (CCBCT) is proposed. Methods proposed integrates residual block concept into cycle‐consistent adversarial network (cycle‐GAN) framework, called res‐cycle GAN, to learn mapping between...

Reliable automated segmentation of the prostate is indispensable for image-guided interventions. However, task challenging due to inhomogeneous intensity distributions, variation in anatomy, among other problems. Manual can be time-consuming and subject inter- intraobserver variation. We developed an deep learning-based method address this technical challenge.We propose a three-dimensional (3D) fully convolutional networks (FCN) with supervision group dilated convolution segment on magnetic...

Purpose Current clinical application of cone‐beam CT (CBCT) is limited to patient setup. Imaging artifacts and Hounsfield unit (HU) inaccuracy make the process CBCT‐based adaptive planning presently impractical. In this study, we developed a deep‐learning‐based approach improve CBCT image quality HU accuracy for potential extended use in CBCT‐guided pancreatic radiotherapy. Methods Thirty patients previously treated with pancreas SBRT were included. The acquired prior first fraction...

Abstract Objective . Artificial intelligence (AI) methods have gained popularity in medical imaging research. The size and scope of the training image datasets needed for successful AI model deployment does not always desired scale. In this paper, we introduce a synthesis framework aimed at addressing challenge limited models. Approach proposed 2D is based on diffusion using Swin-transformer-based network. This consists forward Gaussian noise process reverse transformer-based denoising....

Abstract Background Daily or weekly cone‐beam computed tomography (CBCT) scans are commonly used for accurate patient positioning during the image‐guided radiotherapy (IGRT) process, making it an ideal option adaptive (ART) replanning. However, presence of severe artifacts and inaccurate Hounsfield unit (HU) values prevent its use quantitative applications such as organ segmentation dose calculation. To enable clinical practice online ART, is crucial to obtain CBCT with a quality comparable...

Abstract Background and purpose Magnetic resonance imaging (MRI)‐based synthetic computed tomography (sCT) simplifies radiation therapy treatment planning by eliminating the need for CT simulation error‐prone image registration, ultimately reducing patient dose setup uncertainty. In this work, we propose a MRI‐to‐CT transformer‐based improved denoising diffusion probabilistic model (MC‐IDDPM) to translate MRI into high‐quality sCT facilitate planning. Methods MC‐IDDPM implements processes...

Automatic segmentation of colon polyps can significantly reduce the misdiagnosis cancer and improve physician annotation efficiency. While many methods have been proposed for polyp segmentation, training large-scale networks with limited colonoscopy data remains a challenge. Recently, Segment Anything Model (SAM) has recently gained much attention in both natural image medical segmentation. SAM demonstrates superior performance several vision benchmarks shows great potential In this study,...

Abstract Objective . High-resolution magnetic resonance imaging (MRI) can enhance lesion diagnosis, prognosis, and delineation. However, gradient power hardware limitations prohibit recording thin slices or sub-1 mm resolution. Furthermore, long scan time is not clinically acceptable. Conventional high-resolution images generated using statistical analytical methods include the limitation of capturing complex, high-dimensional image data with intricate patterns structures. This study aims to...

Purpose: Xerostomia (dry mouth), secondary to irradiation of the parotid glands, is one most common side effects head‐and‐neck cancer radiotherapy. Diagnostic tools able accurately and efficiently measure gland injury have yet be introduced into clinic. This study's purpose investigate sonographic textural features as potential imaging signatures for quantitative assessment parotid‐gland after Methods: The authors investigated a series obtained from gray level co‐occurrence matrix (GLCM) –...

Deriving accurate structural maps for attenuation correction (AC) of whole-body positron emission tomography (PET) remains challenging. Common problems include truncation, inter-scan motion, and erroneous transformation voxel-intensities to PET µ-map values (e.g. modality artifacts, implanted devices, or contrast agents). This work presents a deep learning-based (DL-AC) method generate corrected (AC PET) from non-attenuation (NAC images imaging, without the use information. 3D patch-based...

Transrectal ultrasound (TRUS) is a versatile and real-time imaging modality that commonly used in image-guided prostate cancer interventions (e.g., biopsy brachytherapy). Accurate segmentation of the key to needle placement, brachytherapy treatment planning, motion management. Manual during these time-consuming subject inter- intraobserver variation. To address drawbacks, we aimed develop deep learning-based method which integrates supervision into three-dimensional (3D) patch-based V-Net...

Attenuation correction (AC) of PET/MRI faces challenges including inter-scan motion, image artifacts such as truncation and distortion, erroneous transformation structural voxel-intensities to PET mu-map values. We propose a deep-learning-based method derive synthetic CT (sCT) images from non-attenuation corrected (NAC PET) for AC on whole-body imaging.

Purpose Automatic breast ultrasound (ABUS) imaging has become an essential tool in cancer diagnosis since it provides complementary information to other modalities. Lesion segmentation on ABUS is a prerequisite step of computer‐aided (CAD). This work aims develop deep learning‐based method for tumor using three‐dimensional (3D) automatically. Methods For ABUS, we developed Mask scoring region‐based convolutional neural network (R‐CNN) that consists five subnetworks, is, backbone, regional...

Lowering either the administered activity or scan time is desirable in PET imaging as it decreases patient's radiation burden improves patient comfort and reduces motion artifacts. But reducing these parameters lowers overall photon counts increases noise, adversely impacting image contrast quantification. To address this low count statistics problem, we propose a cycle-consistent generative adversarial network (Cycle GAN) model to estimate diagnostic quality images using data. Cycle GAN...

Purpose To develop an accurate and fast deformable image registration (DIR) method for four‐dimensional computed tomography (4D‐CT) lung images. Deep learning‐based methods have the potential to quickly predict deformation vector field (DVF) in a few forward predictions. We developed unsupervised deep learning 4D‐CT DIR with excellent performances terms of accuracies, robustness, computational speed. Methods A method, namely LungRegNet, was proposed using learning. LungRegNet consists two...