Automated lung outline reconstruction in ventilation-perfusion scans using principal component analysis techniques

Area of Science:

• Medical Imaging
• Radiology
• Computer-Aided Diagnosis

Background:

• Pulmonary embolism (PE) diagnosis relies on accurate interpretation of ventilation-perfusion (V/Q) scans.
• Manual extraction of lung outlines from V/Q scans can be time-consuming and prone to variability.
• Standardized feature extraction is crucial for PIOPED-compliant PE diagnosis.

Purpose of the Study:

• To develop an automated software system for reconstructing lung outlines from V/Q scans.
• To enable accurate feature extraction for diagnosing pulmonary embolism.
• To evaluate the efficacy of Principal Component Analysis (PCA) algorithms for lung outline reconstruction.

Main Methods:

• Development of a software-based system using digitized V/Q scans and chest X-rays.
• Implementation of two PCA algorithms: Eigenlungs (adapted from Eigenfaces) and an artificial neural network.
• Utilizing MATLAB(TM) for simulation and evaluation of the reconstruction process.

Main Results:

• The PCA-based approach demonstrated significant viability in reconstructing lung outlines.
• Automated reconstruction facilitates the extraction of PIOPED-compliant features.
• The Eigenlungs and artificial neural network methods showed potential for this application.

Conclusions:

• Automated lung outline reconstruction from V/Q scans is a viable method for PE diagnosis.
• PCA algorithms offer a promising approach for enhancing diagnostic accuracy and efficiency.
• This software system has the potential to streamline the PE diagnostic workflow.