Deep learning proton beam range estimation model for quality assurance based on two‐dimensional scintillated light distributions in simulations

Lee, Eunho; Cho, Byungchul; Kwak, Jungwon; Jeong, C.; Park, Min‐Jae; Kim, Sung‐woo; Song, Si Yeol; Goh, Youngmoon

doi:10.1002/mp.16646

Cited by 3 publications

(1 citation statement)

References 41 publications

(69 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To address these issues, automatic TAC fitting is the most suitable approach, and for automation, both mathematical and deep learning approaches are being considered [10][11][12] . In mathematical approaches, optimization algorithms, such as the least squares method, are used to find the best fit of functions, like the Gamma function, to data points.…”

Section: Introductionmentioning

confidence: 99%

Feasibility of Deep Convolution Neural Network-Based Automatic Time Activity Curve Fitting Method for Non-Invasive Cerebral Blood Flow Quantification

NAGAOKA,

YAMASHITA,

YABUSA

et al. 2023

Radiat. Saf. Manage.

View full text Add to dashboard Cite

In this study, we aimed to develop a deep convolutional neural network (DCNN)-based automatic timeactivity curve (TAC) fitting method for input function determination. This will be achieved through a comparison between the DCNN method, manual method, and mathematical fitting methods using the expectation maximization algorithm (EM-method) to uncover the potential of the DCNN approach.A U-Net architecture based on convolutional neural networks was used to determine the TAC fittings. The area under the curve (AUC) values of the TAC by the EM and DCNN methods were compared to those obtained using the manual method.The AUC values for the EM-manual method exhibited similarity within an error range of approximately ±20%. Conversely, the error range for DCNN-manual method was approximately ±10%, signifying a reduction in the error range to approximately 1/2.Our findings indicate that the DCNN method provides accuracy equivalent to those of manual methods and even slightly superior to that of the EM method.

show abstract

Section: Introductionmentioning

confidence: 99%

Feasibility of Deep Convolution Neural Network-Based Automatic Time Activity Curve Fitting Method for Non-Invasive Cerebral Blood Flow Quantification

NAGAOKA,

YAMASHITA,

YABUSA

et al. 2023

Radiat. Saf. Manage.

View full text Add to dashboard Cite

show abstract

Application of novel hybrid deep learning architectures combining Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN): construction duration estimates prediction considering preconstruction uncertainties

Demiss,

Elsaigh

2024

Eng. Res. Express

View full text Add to dashboard Cite

Construction duration estimation plays a pivotal role in project planning and management, yet it is often fraught with uncertainties that can lead to cost overruns and delays. To address these challenges, this review article proposes three advanced conceptual models leveraging hybrid deep learning architectures that combine Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) while considering construction delivery uncertainties. The first model introduces a Spatio-Temporal Attention CNN-RNN Hybrid Model with Probabilistic Uncertainty Modeling, which integrates attention mechanisms and probabilistic uncertainty modeling to provide accurate and probabilistic estimates of construction duration, offering insights into critical areas of uncertainty. The second model presents a Multi-Modal Graph CNN-RNN Hybrid Model with Bayesian Uncertainty Integration, which harnesses multi-modal data sources and graph representations to offer comprehensive estimates of construction duration while incorporating Bayesian uncertainty measures, facilitating informed decision-making and optimized resource allocation. Lastly, the third model introduces a Hierarchical Spatio-Temporal Transformer CNN-RNN Hybrid Model with Fuzzy Logic Uncertainty Handling, which addresses the inherent vagueness and imprecision in construction duration estimates by incorporating hierarchical spatio-temporal transformer architecture and fuzzy logic uncertainty handling, leading to more nuanced and adaptable project management practices. These advanced models represent significant advancements in addressing construction duration challenges, providing valuable insights and recommendations for future research and industry applications. Moreover, this review article critically examines the application of hybrid deep learning architectures, specifically the combination of CNNs RNNs, in predicting construction duration estimates at the preconstruction stage while considering uncertainties inherent in construction delivery systems.

show abstract

Recent advances in the clinical applications of machine learning in proton therapy

Wildman,

Wynne,

Kesarwala

et al. 2024

Preprint

View full text Add to dashboard Cite

The present systematic review is an effort to explore the different clinical applications and current implementations of machine/deep learning in proton therapy. It will assist as a reference for scientists, researchers, and other health professionals who are working in the field of proton radiation therapy and need up-to-date knowledge regarding recent technological advances. This review utilized Pubmed and Embase to search for and identify research studies of interest published between 2019 and 2024. This systematic literature review utilized PubMed and Embase to search for and identify studies pertinent to machine learning in proton therapy. The time period of 2019 to 2024 was chosen to capture the most recent signficant advances. An initial search on PubMed was made with the search strategy ″′proton therapy′, ′machine learning′, ′deep learning′″, with filters including only research articles from 2019 to 2024, returning 84 results. Next, ″(″proton therapy″) AND (″machine learning″ OR ″deep learning″)″ was searched on Embase, retrieving 546 results. When filtered between 2019 to 2024 and to only research articles, 250 results were retrieved on Embase. Reviews, editorials, technical notes, and articles in any language other than English were excluded from the broad search on both databases. Filtering by title, papers were chosen based on two inclusion factors: explicit application to, or mention of, proton therapy, and inclusion of a machine learning algorithm. Assessing by abstract, works irrelevant to specific aspects of the proton therapy workflow in the scope of the review were excluded. Upon assessing and evaluating full texts for quality, studies were excluded that lacked a clear explanation of model architecture. If multiple studies of the same architecture applied to the same workflow step were identified, chronologically only the most recent advancement in application was included. An additional 5 studies that met all inclusion criteria were identified from references of chosen papers. In total, 38 relevant studies have been summarized and incorporated into this review. This is the first systematic review to comprehensively cover all current and potential areas of application of machine learning to the proton therapy clinical workflow.

show abstract

Deep learning proton beam range estimation model for quality assurance based on two‐dimensional scintillated light distributions in simulations

Cited by 3 publications

References 41 publications

Feasibility of Deep Convolution Neural Network-Based Automatic Time Activity Curve Fitting Method for Non-Invasive Cerebral Blood Flow Quantification

Feasibility of Deep Convolution Neural Network-Based Automatic Time Activity Curve Fitting Method for Non-Invasive Cerebral Blood Flow Quantification

Application of novel hybrid deep learning architectures combining Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN): construction duration estimates prediction considering preconstruction uncertainties

Recent advances in the clinical applications of machine learning in proton therapy

Contact Info

Product

Resources

About