Deep Learning Algorithm for Lung Adenocarcinoma Subtype Recognition Based on Multimodal Data
I am writing to request access to the National Lung Screening Trial (NLST) database for conducting research on the classification of lung adenocarcinoma subtype using multi-modal data. I am a clinical surgeon with a primary focus on AI-assisted lung cancer diagnosis, possessing extensive research experience, and firmly believe that access to the NLST database would greatly benefit my research.
In July of this year, our team published an article on AI-assisted lung cancer diagnosis in Npj Digital Medicine (link: https://www.nature.com/articles/s41746-023-00866-z). We have developed an online diagnostic platform for classifying lung adenocarcinoma. The data used for this study were sourced from CT imaging data of patients from three Chinese hospitals. We introduced a novel deep learning algorithm for tasks involving benign/malignant discrimination, pre-invasive/invasive discrimination, and further classification of the invasive adenocarcinoma. Building upon this foundation, we plan to pursue two specific goal in the proposed project:
We aim to include the CT imaging data from the NLST database to perform external model validation and optimize our existing models and algorithms.
We intend to develop more accurate algorithms for the three classification tasks from a multi-modal data perspective to enhance preoperative diagnostic accuracy. The NLST database encompasses a wealth of patient information, including pathology, follow-up data, and treatment records, providing a solid data foundation for our multi-modal research.
In conclusion, I kindly request your support and approval to grant me access to the NLST database. This access will not only significantly impact my research but also contribute to advancements in the field of lung cancer research.
1、We plan to conduct deep learning research based on multi-modal data to identify different lung cancer subtypes, thereby enhancing personalized treatment options for patients.
2、Utilizing the NLST's long-term follow-up data, we aim to develop more precise preoperative diagnostic methods to guide the formulation of follow-up plans and surgical approaches.
3、We aim to analyze treatment data from both the NLST database and our own dataset to investigate the impact of different treatment regimens on the survival of lung cancer patients, ultimately providing valuable insights to guide clinical practices.
Professor Jing Zhou from Renmin University of China,
Wei Feng, The Third Xiangya Hospital of Central South University
Zhang Zhang, Changsha Central Hospital
Siyuan Ai, Beijing LIANGXIANG Hospital