Deepnoid Presents Research Abstract on 'Lung Cancer Screening' at the American Roentgen Ray Society Meeting

Deepnoid, a first-generation domestic medical artificial intelligence company, announced on the 18th that it presented a research abstract on the "Deep learning-based automated Lung-RADS classification algorithm for lung cancer screening using low-dose CT (LDCT)" at the American College of Radiology (ACR) 2024.

The algorithm was designed to reduce the time required for lung image analysis for lung cancer diagnosis while minimizing classification variability. The classification accuracy for Lung-RADS Scores 4A and 4B, which indicate a high probability of lung cancer among detected lung nodules, showed performance rates of 81.41% and 96.38%, respectively.

Lung-RADS is a system that grades the probability of lung nodules being lung cancer. It classifies nodules from 1 to 4, with scores 2 to 3 considered benign and score 4 considered malignant.

The algorithm will be applied to the real-time lung nodule detection AI solution ‘DEEP:LUNG (Deeplung) DL-LN-02’ and is scheduled for commercialization in the second half of this year. DEEP:LUNG is an AI-based solution that detects suspicious lung nodule areas from low-dose chest CT images to assist medical professionals in diagnosis. Through this research, the Lung-RADS scoring function was added. Compared to the existing model, the performance for lung nodule detection improved with sensitivity and specificity increased by 18% and 11%, respectively.

The Deepnoid research team conducting the study stated, “Low-dose chest computed tomography (LDCT) used for lung cancer screening requires interpretation of hundreds of slices, which takes a significant amount of time during reading,” and added, “This research provides an efficient approach for lung cancer diagnosis, generating high market expectations.”

The research on Lung-RADS classification was conducted in collaboration with Professor Ho-Seok Lee, a thoracic surgeon and project leader at Pusan National University Hospital, as part of ‘Doctor Answer 2.0.’ This project supports AI precision medical solutions throughout the entire medical cycle, including disease diagnosis and treatment. It is supported by the Ministry of Science and ICT and the National IT Industry Promotion Agency.