Lung Nodule Detection via Deep Reinforcement Learning.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Lung cancer is a leading cause of cancer death globally.
• Low-dose computed tomography (CT) screening is recommended for high-risk individuals.
• Interpreting the large volume of CT scans presents a challenge for radiologists.

Purpose of the Study:

• To develop and validate a deep reinforcement learning model for early lung nodule detection in thoracic CT images.
• To address the challenge of interpreting large volumes of screening CT scans using computer-aided detection (CAD).

Main Methods:

• A deep learning algorithm inspired by AlphaGo was developed.
• The model processes raw CT images as states to classify nodule presence.
• The LIDC/IDRI database from the LUNA challenge was used for training and testing.

Main Results:

• The model achieved 99.1% overall accuracy during training.
• Testing results showed an overall accuracy of 64.4% (sensitivity 58.9%, specificity 55.3%).
• Early results indicate potential for reducing false positives in lung nodule CT screening.

Conclusions:

• The developed deep learning model shows promise for early lung nodule detection.
• The algorithm may help reduce unnecessary follow-up tests and healthcare costs associated with lung cancer screening.
• Further validation is needed to address the observed performance gap between training and testing.