Lung Segmentation and Nodule Detection in Computed Tomography Scan using a Convolutional Neural Network Trained...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Lung cancer is a leading cause of cancer mortality worldwide.
• Early detection of pulmonary nodules via low-dose computed tomography (CT) is critical for patient outcomes.
• Manual screening of CT scans is time-consuming and prone to human error.

Purpose of the Study:

• To develop a computationally efficient framework for automated lung nodule detection.
• To improve the accuracy and reduce the time required for screening pulmonary nodules in CT scans.

Main Methods:

• A two-stage framework was proposed, utilizing a convolutional neural network (CNN).
• The first stage employed adversarial training with Turing test loss for lung region segmentation.
• The second stage involved classifying sampled patches from the segmented region for nodule detection.

Main Results:

• The proposed framework achieved a high Dice coefficient of 0.984±0.0007.
• Validation was performed using 10-fold cross-validation on the LUNA16 challenge dataset.

Conclusions:

• The developed computational framework offers an efficient and accurate solution for lung nodule detection.
• This automated approach has the potential to significantly aid in the early diagnosis of lung cancer.