Abstract With the advancement of treatment modalities in radiation therapy for cancer patients, outcomes have improved, but at cost increased plan complexity and planning time. The accurate prediction dose distributions would alleviate this issue by guiding clinical optimization to save time maintain high quality plans. We modified a convolutional deep network model, U-net (originally designed segmentation purposes), predicting from patient image contours target volume (PTV) organs risk...

The use of neural networks to directly predict three-dimensional dose distributions for automatic planning is becoming popular. However, the existing methods only patient anatomy as input and assume consistent beam configuration all patients in training database. purpose this work was develop a more general model that considers variable configurations addition achieve comprehensive with potentially easier clinical implementation, without need train specific models different settings.The...

In this paper, we present a fully automatic, fast and accurate deformable registration technique. This technique deals with free-form deformation. It minimizes an energy functional that combines both similarity smoothness measures. By using calculus of variations, the minimization problem was represented as set nonlinear elliptic partial differential equations (PDEs). A Gauss-Seidel finite difference scheme is used to iteratively solve PDE. The refined by multi-resolution approach. whole...

Accurate and automatic brain metastases target delineation is a key step for efficient effective stereotactic radiosurgery (SRS) treatment planning. In this work, we developed deep learning convolutional neural network (CNN) algorithm segmenting on contrast-enhanced T1-weighted magnetic resonance imaging (MRI) datasets. We integrated the CNN-based into an segmentation workflow validated both Multimodal Brain Tumor Image Segmentation challenge (BRATS) data clinical patients' data. Validation...

The looming potential of deformable alignment tools to play an integral role in adaptive radiotherapy suggests a need for objective assessment these complex algorithms. Previous studies this area are based on the ability reproduce analytically generated deformations applied sample image data, or use contours bifurcations as ground truth evaluation accuracy. In study, phantom was embedded with 48 small plastic markers, placed regions varying from high contrast roughly uniform regional...

The incorporation of daily images into the radiotherapy process leads to adaptive radiation therapy (ART), in which treatment is evaluated periodically and plan adaptively modified for remaining course radiotherapy. Deformable registration between planning image a key component ART. In this paper, we report our researches on deformable kVCT MVCT sets. method based fast intensity-based free-form technique. Considering noise contrast resolution differences MVCT, an 'edge-preserving smoothing'...

Patient motion, especially respiratory results in various artefacts such as blurring and streaks tomographic images. The interplay of the movement beam aperture variations organ anatomy during delivery can create 'hot' 'cold' spots throughout field intensity-modulated radiation therapy (IMRT). Detection correction patient motion is extremely important imaging IMRT. Tomographic projection data (sinogram) encode not only information, but also intra-scanning information. In this paper, we...

Inversion of deformation fields is applied frequently to map images, dose, and contours between the reference frame study frame. A prevailing approach that takes negative forward as inverse oversimplified can cause large errors for deformations or are composites several deformations. Other approaches, including Newton's method scatter data interpolation, either require first derivative very inefficient. Here we propose an iterative easy implement, converges quickly when it does, works a...

Delineating regions of interest (ROIs) on each phase four-dimensional (4D) computed tomography (CT) images is an essential step for 4D radiotherapy. The requirement manual phase-by-phase contouring prohibits the routine use This paper develops automatic re-contouring algorithm that combines techniques deformable registration and surface construction. ROIs are manually contoured slice-by-slice in reference image. A constructed based these contours using a triangulated construction technique....

Deep learning has started to revolutionize several different industries, and the applications of these methods in medicine are now becoming more commonplace. This study focuses on investigating feasibility tracking patients clinical staff wearing Bluetooth Low Energy (BLE) tags a radiation oncology clinic using artificial neural networks (ANNs) convolutional (CNNs). The performance was compared relative received signal strength indicator (RSSI) thresholding triangulation. By utilizing...

Partly due to the use of exhaustive-annotated data, deep networks have achieved impressive performance on medical image segmentation. Medical imaging data paired with noisy annotation are, however, ubiquitous, but little is known about effect learning based We studied in context mandible segmentation from CT images. First, 202 images head and neck cancer patients were collected our clinical database, where organs-at-risk annotated by one twelve planning dosimetrists. The mandibles roughly as...

Similar to conventional conformal radiotherapy, during lung tomotherapy, a motion margin has be set for respiratory motion. Consequently, large volume of normal tissue is irradiated by intensive radiation. To solve this problem, we have developed new mitigation method incorporating target into treatment optimization. In method, the delivery‐breathing correlation determined prior plan Beamlets are calculated using CT images at corresponding breathing phases from dynamic (four‐dimensional)...

Convolution/superposition (C/S) is regarded as the standard dose calculation method in most modern radiotherapy treatment planning systems. Different implementations of C/S could result significantly different distributions. This paper addresses two major implementation issues associated with collapsed cone C/S: one how to utilize tabulated kernels instead analytical parametrizations and other deal voxel size effects. Three methods that are presented this paper. These differ effective used:...

We present a novel framework that enables very large scale intensity-modulated radiation therapy (IMRT) planning in limited computation resources with improvements cost, plan quality and throughput. Current IMRT optimization uses voxel-based beamlet superposition (VBS) requires pre-calculation storage of amount data, resulting temporal spatial complexity. developed non-voxel-based broad-beam (NVBB) for capable direct treatment parameter (DTPO). In this framework, both objective function...

Abstract Deep learning (DL)-based auto-segmentation has the potential for accurate organ delineation in radiotherapy applications but requires large amounts of clean labeled data to train a robust model. However, annotating medical images is extremely time-consuming and clinical expertise, especially segmentation that demands voxel-wise labels. On other hand, without annotations are abundant highly accessible. To alleviate influence limited number labels, we propose weakly supervised DL...

We propose a novel BIRADS-SSDL network that integrates clinically-approved breast lesion characteristics (BIRADS features) into task-oriented semi-supervised deep learning (SSDL) for accurate diagnosis of ultrasound (US) images with small training dataset. Breast US are converted to BIRADS-oriented feature maps (BFMs) using distance-transformation coupled Gaussian filter. Then, the BFMs used as input an SSDL network, which performs unsupervised stacked convolutional auto-encoder (SCAE) image...

Collapsed-cone convolution/superposition (CCCS) dose calculation is the workhorse for IMRT calculation. The authors present a novel algorithm computing CCCS on modern graphic processing unit (GPU).The GPU includes TERMA that has no write-conflicts and linear computation complexity. uses either tabulated or exponential cumulative-cumulative kernels (CCKs) as reported in literature. have demonstrated use of can reduce complexity by order dimension achieve excellent accuracy. Special attentions...

The aim of this work is to develop a novel recursive ensemble OARs segmentation (REOS) framework for accurate organs-at-risk (OARs) automatic segmentation. REOS recursively segment individual by ensembling images features extracted from an organ localization module and contour detection module. Both modules are based on 3D U-Net architecture. trained rough localize region interest (ROI) that encompasses the to-be-delineated OAR, while OAR within identified ROI. In study, developed applied...