Computational Decision Support for PE Diagnosis based on Ventilation Perfusion Ratio.

Area of Science:

• Medical Imaging
• Radiology
• Nuclear Medicine

Background:

• Pulmonary embolism (PE) diagnosis relies on interpreting ventilation/perfusion (V/Q) scans.
• Accurate quantification of lung lobe V/Q is crucial for effective PE assessment.
• Existing methods may lack the precision needed for robust clinical decision support.

Purpose of the Study:

• To evaluate computer-aided, semi-automated 3D lung lobe quantification for PE diagnosis support.
• To assess the utility of ventilation/perfusion ratio (VQR) and volume/perfusion ratio (VPR) in clinical practice.
• To determine if 3D quantification aids decision-making in PE diagnosis.

Main Methods:

• 100 patients underwent ventilation/perfusion single photon emission computed tomography (V/Q-SPECT/CT).
• Two 3D lung lobe quantification software tools (Q.Lung and LLQ) were used to assess VQR and VPR.
• Receiver operating characteristic (ROC) analysis was employed to develop a PE detection algorithm, with results compared to expert assessment using Youden index and Area Under the Curve (AUC).

Main Results:

• Both 3D software tools demonstrated good comparability in VPR and VQR analysis.
• The Youden index ranged from 0.2 to 0.5, indicating moderate agreement.
• Mean AUC values for the software tools ranged from 0.64 to 0.66, suggesting feasibility for PE detection.

Conclusions:

• 3D software tools are feasible for numerical pulmonary embolism detection.
• The use of numerical algorithms based on ROC analysis can support clinical decision-making in PE diagnosis.
• Semi-automated 3D lung lobe quantification shows promise in enhancing PE diagnostic accuracy.