Category Ranking
98%
Total Visits
921
Avg Visit Duration
2 minutes
Citations
20
Article Abstract
Purpose: Recently, it has been shown that automated treatment planning can be executed by direct fluence prediction from patient anatomy using convolutional neural networks. Proof of principle publications utilise a fixed dose prescription and fixed collimator (0°) and gantry angles. The goal of this work is to further develop these principles for the challenging lung cancer indication with variable dose prescriptions, collimator and gantry angles. First we investigate the impact of clinical applicable collimator angles and various input parameters. Then, the model is tested in a complete user independent planning workflow.
Methods: The dataset consists of 152 lung cancer patients, previously treated with IMRT. The patients are treated with either a left or a right beam setup and collimator angles and dose prescriptions adjusted to their tumour shape and stage. First we compare two CNNs with standard vs. personalised, clinical collimator angles. Next, four CNNs are trained with various combinations of CT and contour inputs. Finally, a complete user free treatment planning workflow is evaluated.
Results: The difference between the predicted and ground truth fluence maps for the fluence prediction CNN with all anatomical inputs in terms of the mean mean absolute error (MAE) is 4.17 × 10 for a fixed collimator angle and 5.46 × 10 for variable collimator angles. These differences vanish in terms of DVH metrics. Furthermore, the impact of anatomical inputs is small. The mean MAE is 5.88 × 10 if no anatomical information is given to the network. The DVH differences increase when a total user free planning workflow is examined.
Conclusions: Fluence prediction with personalised collimator angles performs as good as fluence prediction with a standard collimator angle of zero degrees. The impact of anatomical inputs is small. The combination of a dose prediction and fluence prediction CNN deteriorates the fluence predictions. More investigation is required.
Download full-text PDF |
Source |
|---|---|
| http://dx.doi.org/10.1016/j.ejmp.2022.05.008 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Unraveling Plasmon-Enhanced Reactive Oxygen Species Generation through Ultrafast Light.
J Phys Chem C Nanomater Interfaces
February 2025
The Photonics Center, Boston University, Boston, MA 02215, United States.
Reactive oxygen species (ROS) generation through gold nanorods (AuNRs) excited by 812 nm centered, 85 femtosecond (fs)-pulsed laser irradiation was investigated through a rhodamine B degradation assay. The initial rate of rhodamine B fluorescence intensity degradation is determined by the rate of ROS generation, but at later time points the laser irradiation induced deformation of AuNRs reduces the rate of rhodamine B degradation. For different AuNR preparations that all had a localized surface plasmon resonance (LSPR) mode at around 800 nm but differed in size, the initial rate of rhodamine B fluorescence intensity decrease follows a trend predicted by the simulated peak near-field intensities and absorption efficiencies except for the smallest AuNRs with dimensions of 30 by 7 nm.
View Article and Find Full Text PDFComplexity-based unsupervised machine learning for patient-specific VMAT quality assurance.
Med Phys
September 2025
Radiation Oncology, Department of Medical and Surgical Sciences (DIMEC), Alma Mater Studiorum, University of Bologna, Bologna, Italy.
Background: Patient-specific quality assurance (PSQA) is essential to guarantee the requested accuracy and safety of high-precision radiotherapy treatments. With the widespread adoption of modulated-intensity techniques, there is a growing need for increased operational efficiency. The potential of machine learning (ML) to accurately predict PSQA results has been investigated in recent years.
View Article and Find Full Text PDFConvolutional neural network-based prediction of gamma pass rate in volumetric modulated arc therapy using 3d fluence maps and complexity metrics.
Med Dosim
August 2025
Department of Radiation Oncology, Kyorin University, Tokyo, 181-8611, Japan.
This study aims to develop and validate a deep learning-based model to accurately predict gamma pass rate (GPR) in treatment plans for volumetric modulated arc therapy (VMAT). A dataset comprising 360 VMAT treatment plans was utilized to train and validate the deep learning models. The 3D fluence maps, along with complexity metrics of the treatment planning, were extracted from each treatment plan as input features.
View Article and Find Full Text PDFCalculation of effective pump dose in X-ray-pump/X-ray-probe experiments.
J Synchrotron Radiat
September 2025
Center for Free-Electron Laser Science CFEL, Deutsches Elektronen-Synchrotron DESY, Notkestr. 85, 22607 Hamburg, Germany.
In pump-probe experiments on solid materials performed within ultrafast X-ray science, the energy deposited by an X-ray pump pulse in the sample has a non-uniform spatial distribution. The following X-ray probe pulse then measures a volume-integrated average of contributions from the differently irradiated regions of the sample. Here we propose a scheme to calculate an effective fluence of the pump pulse such that the observable of interest calculated with the effective fluence is very close to the volume-integrated observable.
View Article and Find Full Text PDFIn nanolithography, optical diffraction from gratings etched into the scribe lanes of semiconductor devices is used for wafer alignment. As these gratings become increasingly smaller, achieving sufficiently strong diffraction signals requires higher light fluences, increasing the risk of optical damage. This study explores light-induced optical and structural changes in flat silicon and gratings etched in silicon when exposed to single femtosecond laser pulses.
View Article and Find Full Text PDF