Research presented today suggests that an artificial intelligence tool called DeepGEM may drive advances in genomic testing, providing accurate, cost-effective and timely methods for predicting gene mutations in histopathology slices.
The research was presented by Professor Liang Wenhua from the State Key Laboratory of Respiratory Diseases of China and the National Respiratory Disease Clinical Research Center of the First Affiliated Hospital of Guangzhou Medical University at the IASLC 2024 World Lung Cancer Congress today.
Accurate detection of driver gene mutations is crucial for effective treatment planning and prognosis prediction of lung cancer. Traditional genomic testing relies on high-quality tissue samples and is often time-consuming and resource-intensive, limiting accessibility, especially in resource-poor environments. To address this gap, Professor Liang and his team used DeepGEM, which uses routinely obtained histology slides to predict genetic mutations, significantly improving the accessibility and efficiency of mutation screening.
Professor Liang and his colleagues analyzed data sets from 16 centers and 3,658 patients. The dataset includes paired pathology images and gene mutation data, supplemented by the public dataset of the Cancer Genome Atlas.
DeepGEM was initially trained and evaluated on an internal dataset of 1,717 patients, and was subsequently tested on an external dataset of 1,719 patients and a public dataset of 535 patients from an additional 15 centers. To assess prevalence, the model was also tested on a lymph node metastasis dataset consisting of 331 biopsies.
In the internal dataset, DeepGEM demonstrated performance with a median area under the curve of 0.938 for resection biopsies and a median area under the curve of 0.891 for aspiration biopsies. On the multi-center external dataset, the model achieved a median AUC of 0.859 in resection biopsies and a median AUC of 0.826 in aspiration biopsies.
The model was further verified on the TCGA dataset with an AUC of 0.874, demonstrating its effectiveness in different ethnic contexts. Importantly, DeepGEM’s ability to predict mutations in primary biopsies extends to lymph node metastases, demonstrating its potential in predicting outcomes for targeted therapies.
According to Professor Liang, DeepGEM also provides interpretable results, generates spatial gene mutation maps at the single cell level, and has been verified based on immunohistochemical results, emphasizing the accuracy and reliability of the model.
“Compared to previous studies, DeepGEM achieved robust and excellent prediction performance across genes, validating the largest multicenter dataset to date. DeepGEM’s rapid prediction capabilities allow faster treatment decisions, allowing patients with severe symptoms to receive targeted therapy more quickly,”Professor Liang reported.
“In addition, it also provides opportunities for polygene mutation testing and precise treatment in economically underdeveloped areas where genomic testing is unaffordable. This innovative approach has the potential to transform the clinical management of lung cancer patients, making advanced genomic insights more accessible and feasible.”
There are links in the video below this video. If you are interested, you can open it and have a look.
Thank you for watching this video. If you like it, please subscribe and like it. thank
Original text:https://medicalxpress.com/news/2024-09-artificial-intelligence-method-advance-gene.html
Provided by the International Association for the Study of Lung Cancer
Oil tubing: