Smart X-Ray: Intelligent Nodule Detection

Objective: Small masses known as lung nodules, which appear as white, round dots, can be detected on lung X-rays or CT scans. These nodules are typically less than 3 cm in size, and timely detection is crucial in early lung cancer diagnosis. X-rays are often used as a quick and affordable imaging method in various clinical processes. With the help of artificial intelligence, these imaging results can be automatically processed and accurately labeled to identify any possible lung nodules.

Method: A solution method that involves three different CNN models, each with unique architectural features, has been developed. In pre-processing before training, X-rays are processed with Gamma correction and CLAHE methods. The first step involves lung segmentation and identifying nodules as a binary classification problem. If a nodule is found, the next step involves searching for nodule segmentation.

Innovative Features

• Works on Turkey-specific lung nodule diagnostic dataset
• Artificial intelligence service capable of automatic lung segmentation is available.
• Includes deep learning and lung x-ray classification and segmentation service.
• It can be adapted to different usage scenarios.
• Capable of reading and processing DICOM-formatted X-rays