2022 World Conference on Lung Cancer (ePosters)-EP01.05-001. Radiomics to Increase the Effectiveness of Lung Cancer Screening Programs. Radiolung Preliminary Results

EP01.05-001. Radiomics to Increase the Effectiveness of Lung Cancer Screening Programs. Radiolung Preliminary Results

Pdf Summary

This study aimed to establish a radiomic signature for pulmonary nodules (PN) to distinguish between benign and malignant nodules. The researchers conducted a prospective observational study using CT images of PNs that were studied and resected according to usual clinical practice. The images were sent to the Computer Vision Center, where the nodules were segmented. Gray Level Co-occurrence Matrix radiomic based texture features were extracted, which correlated significantly with malignancy. These variables were then used to train a neural network optimized for the diagnosis of the nodule. The model was trained with benign and malignant PNs from the hospital and validated with additional PNs from the hospital and a public database. <br /><br />The results showed that 51 PNs were analyzed, with a mean size of 22.68mm. The pathological results revealed various types of lung cancer, including adenocarcinoma, squamous cell carcinoma, and non-small cell lung cancer. The diagnostic accuracy of the hybrid system was found to be 96.30%, with 100% sensitivity and 83.3% specificity. <br /><br />In conclusion, the application of a hybrid radiomic system achieved high diagnostic accuracy (96.3%) in detecting malignant nodules on chest CT. The researchers suggest that external validation in a lung cancer screening program is needed. The study was funded by ACMCiB, BRN, Fundació Ramon Pla, and the Lung Ambition Alliance.

Asset Subtitle

Antoni Rosell

Meta Tag

Speaker Antoni Rosell

Topic Early Detection and Screening - Pulmonary Nodule

Keywords

radiomic signature

pulmonary nodules

benign

malignant

observational study

CT images

Computer Vision Center

texture features

neural network

lung cancer