Improved accuracy of pulmonary embolism computer-aided detection using iterative reconstruction compared with...

Area of Science:

• Radiology
• Medical Imaging
• Computer-Aided Detection

Background:

• Pulmonary embolism (PE) is a critical diagnosis.
• Computer-aided detection (CAD) systems aim to improve PE detection accuracy.
• Image reconstruction techniques can impact CAD performance.

Purpose of the Study:

• To evaluate if iterative reconstruction (IR) enhances the performance of a pulmonary embolism computer-aided detection (PE CAD) prototype.
• To compare PE CAD performance using filtered back projection (FBP) versus various levels of hybrid IR.

Main Methods:

• CT angiographic images from 40 PE-positive and 26 control cases were analyzed.
• Images were reconstructed using filtered back projection (FBP) and six levels of hybrid iterative reconstruction (HIR).
• A prototype PE CAD system assessed performance on a per-embolus and per-study basis for each reconstruction type.

Main Results:

• Hybrid IR significantly reduced false-positive marks from PE CAD compared to FBP.
• Specificity improved with higher HIR levels, but sensitivity (both per-embolus and per-study) decreased progressively.
• Overall accuracy was highest at HIR level 1, with higher levels showing reduced performance.

Conclusions:

• Iterative reconstruction (IR) effectively reduces false positives in PE CAD.
• However, high levels of IR lead to decreased sensitivity and overall accuracy.
• Avoid using very high IR levels concurrently with PE CAD to maintain optimal detection performance.