The integration of machine learning (ML) into lung cancer radiotherapy (RT) represents a major leap forward in the field of oncology. As the burden of lung cancer continues to increase globally, there is an urgent need for innovative approaches that can increase the precision, efficacy, and personalization of treatment. This Research Topic in Frontiers in Oncology, “Machine Learning in Lung Cancer Radiotherapy,” brings together cutting-edge studies that demonstrate the transformative potential of ML technologies. By leveraging advanced algorithms and large datasets, these studies aim to optimize treatment plans, improve prediction accuracy, and ultimately improve patient outcomes. This collection of papers not only highlights current advances but also lays the groundwork for future innovations in the integration of ML into lung cancer radiotherapy. Highlights include using deep learning to enhance adaptive radiotherapy and a literature analysis of ML in non-small cell lung cancer (NSCLC) radiotherapy. Additionally, research topics include studies demonstrating the efficiency of automated treatment planning with reinforcement learning, evaluating the interfraction stability of volumetric modulated radiation therapy (VMAT) or intensity modulated radiation therapy (IMAT) dose distributions for lung cancer, and investigating the predictive power of machine learning in assessing the risk of radiation pneumonitis.
Hooshangnejad et al. present a study on the implementation of a novel highly accelerated adaptive radiation therapy (DAART) approach for radiation treatment of lung cancer. As lung cancer remains the leading cause of cancer-related deaths and radiation therapy is an important treatment for medically inoperable early-stage NSCLC, Hooshangnejad et al.'s study addresses the important challenge of shortening the time from diagnosis to treatment initiation. The current median is 4 weeks, which can lead to restaging and loss of local control, but the DAART approach employing the innovative deepPERFECT system aims to significantly shorten this delay. Zhang et al. conducted a comprehensive bibliometric analysis to explore the progress, research trends, and hotspots in the application of ML to radiation treatment of NSCLC. As ML is increasingly integrated into radiation treatment of NSCLC, understanding these trends will be important to guide future research and development. provides valuable insights into the current state of ML applications in NSCLC RT and highlights potential hot areas for future research, helping researchers identify emerging trends and opportunities in the field. Moreover, Wang et al. present a novel integrated solution for automated planning of intensity-modulated radiation therapy (IMRT) in NSCLC cases. The study aims to increase the efficiency and consistency of treatment planning using advanced ML techniques. Wang et al. demonstrate the feasibility and potential of this integrated solution to streamline planning workflow and reduce the variability of plan quality across different regions and treatment centers, paving the way for further improvements and wider clinical implementation. Guberina et al. present a study aimed at evaluating the interfraction stability of dose distributions administered with expiratory-gated VMAT or IMAT for lung cancer. The study also aims to identify the main prognostic dosimetric and geometric factors influencing treatment efficacy. They show that the cumulative dose distributions throughout a treatment series are robust to interfraction CTV deformations when using expiratory gating and online image guidance. DMinutes It was identified as the most important parameter for predicting gEUD in single fractions. Other geometric parameters provided only limited additional predictive value. These findings highlight the importance of dosimetric information, especially the location and value of D.Minutes Inside CTVIfor optimizing image-guided radiation therapy. In addition, Ye et al. developed an optimal ML model for predicting the occurrence of radiation pneumonitis (RP) in lung cancer patients treated with VMAT. This study highlights the usefulness of lung equivalent uniform dose (lung EUD) as a predictive indicator of RP, aiming to improve prediction accuracy and treatment planning. In Ye et al.'s study, four prominent machine learning algorithms were used and it was demonstrated that lung EUD-based factors significantly improved the prediction performance of RP 2+. The results claim that a decision tree model with lung EUD-based predictors is the optimal tool for predicting RP in lung cancer patients treated with VMAT, which may replace traditional dosimetry parameters and simplify the complex neural network structure of the prediction model.
The collection of papers featured in this Research Topic represents a transformation in lung cancer care. Each study addresses key challenges in the field, from accelerating adaptive radiation therapy to predicting the occurrence of radiation pneumonitis. These advances represent major strides in optimizing treatment plans, increasing accuracy, and improving patient outcomes. Leveraging the power of ML algorithms, researchers have developed innovative solutions that streamline treatment workflows, reduce planning uncertainty, and enable personalized care for lung cancer patients. Furthermore, the integration of advanced dosimetry parameters and predictive models can provide clinicians with valuable insights into predictions of response to treatment and toxicity, ultimately leading to a more informed decision-making process.
The integration of ML is likely to play a pivotal role in shaping the future direction of radiation therapy for lung cancer. As demonstrated by the studies covered in this Research Topic, ML algorithms hold immense potential in optimizing treatment plans, predicting treatment outcomes, and personalizing patient care. Going forward, further research in this field is expected to focus on improving existing models, expanding datasets, and integrating multimodal data sources to improve prediction accuracy. Furthermore, efforts toward developing automated treatment planning systems and real-time adaptive strategies are expected to accelerate, aiming to streamline clinical workflow and reduce uncertainties during treatment delivery. Moreover, the integration of artificial intelligence and deep learning techniques represents a promising avenue to gain new insights into tumor biology, treatment response, and patient prognosis. Collaborative efforts between clinicians, physicists, and data scientists will be paramount in translating these technological advances into tangible clinical benefits. Moreover, it is essential to maintain a firm commitment to patient-centered care and ethical considerations to ensure that these transformative technologies are utilized responsibly to improve patient outcomes and quality of life.
Author contributions
JC: writing – original draft, writing – review and editing. TW: writing – review and editing.
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors declared their membership in the editorial board of Frontiers at the time of submission, which had no influence on the peer review process and the final decision.
Publisher's Note
All claims made in this article are those of the authors themselves and do not necessarily represent those of their institutions, publishers, editors, or reviewers. Any products evaluated in this article, or claims made by their manufacturers, are not endorsed or approved by the publishers.
keyword: Artificial Intelligence (AI), Lung Cancer, Radiation Therapy, Deep Learning – Artificial Intelligence, Machine Learning, Adaptive Radiation Therapy (ART), NSCLC, Big Data and Analytics
Quote: Chow JCL and Wang T (2024) Editorial: Machine learning in lung cancer radiotherapy. Front desk. Oncol. 14:1444543. Source: 10.3389/fonc.2024.1444543
received: June 5, 2024; approved: June 6, 2024;
release date: July 2, 2024.
Copyright © 2024 Chow and Wang. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY). Use, distribution, or reproduction in other forums is permitted, provided the original author and copyright holder are credited and the original publication in this journal is cited in accordance with scholarly practice. Any use, distribution, or reproduction not in accordance with these terms is not permitted.
*correspondence: James C. L. Chow, james.chow@uhn.ca
