Design of a Graph-Based System for Similar Case Retrieval of Pulmonary Nodules

Area of Science:

• Medical Imaging Analysis
• Computational Radiology
• Artificial Intelligence in Healthcare

Background:

• Diagnosing lung cancer is challenging, necessitating computational tools for imaging interpretation.
• Content-Based Image Retrieval (CBIR) aids specialists by finding similar lung nodules in databases.
• Existing CBIR systems struggle with visualizing multidimensional medical images effectively.

Purpose of the Study:

• To design a graph-based system for retrieving similar temporal computed tomography (CT) scans of pulmonary nodules.
• To optimize the visualization of multidimensional images within a CBIR system for lung nodule analysis.
• To enhance decision support for specialists in lung cancer diagnosis.

Main Methods:

• Developed a graph-based system for temporal CT scan retrieval.
• Employed Temporal Image Registration to compare segmented pulmonary nodules.
• Utilized a rooted tree graph for visualizing retrieved cases.

Main Results:

• Successfully designed a system for retrieving similar temporal CT scans of pulmonary nodules.
• Implemented a rooted tree graph for effective visualization of multidimensional image data.
• Deployed the system in a web-based platform for enhanced usability and portability.

Conclusions:

• The proposed graph-based system effectively addresses the visualization challenges in multidimensional medical imaging for CBIR.
• This approach offers improved decision support for specialists in diagnosing lung cancer through efficient retrieval and visualization of temporal CT scans.
• The web-based deployment ensures accessibility and practical application in clinical settings.