Related Experiment Videos
Spatial domain image filtering in computed tomography: feasibility study in pulmonary embolism
Joachim E Wildberger1, Andreas H Mahnken, Thomas Flohr
1Department of Diagnostic Radiology, University Hospital, University of Technology (RWTH), Pauwelsstrasse 30, 52074 Aachen, Germany. wildberg@rad.rwth-aachen.de
European Radiology
|March 29, 2003
Summary
Spatial domain filtering is a feasible alternative to additional image reconstructions in chest CT, offering comparable diagnostic accuracy for pulmonary embolism detection. This method reduces processing time and storage costs while maintaining image quality and clinical utility.
Area of Science:
- Radiology and Medical Imaging
- Computational Imaging
Background:
- Conventional chest CT protocols often require multiple image reconstructions with different kernels to achieve desired image characteristics.
- Spatial domain filtering offers a potential method to generate different image sets from a single reconstruction, potentially streamlining workflows.
Purpose of the Study:
- To evaluate the clinical feasibility of spatial domain filtering as an alternative to additional image reconstructions in chest CT.
- To compare the diagnostic accuracy and image quality of spatial domain filtered images with conventionally reconstructed images for pulmonary embolism detection.
Main Methods:
- Forty adult patients with suspected pulmonary embolism underwent multi-slice CT.
- Thin collimated source images were used to generate two sets of 5 mm effective slice thickness (S(eff)) images with a 5 mm reconstruction increment (RI) using lung (B50) and soft tissue (B30) kernels.
- Additionally, B50 images were spatially filtered to simulate B30 images. Diagnostic accuracy was assessed for both filtered and reconstructed images against thin axial slices (S(eff) 1.25 mm, RI 0.8 mm; B30) as the gold standard. Subjective image quality and quantitative measurements were also performed.
Main Results:
- Spatial domain filtering produced images largely equivalent to B30 reconstructions.
- Diagnostic accuracy for detecting central, segmental, and subsegmental pulmonary embolism was comparable between spatial domain filtered images and conventionally reconstructed B30 images.
- Subjective image quality ratings were similar for both methods (1.30 for reconstructed vs. 1.35 for filtered), with a weighted kappa coefficient of 0.6117, indicating good agreement.
Conclusions:
- Spatial domain filtering is clinically feasible and provides diagnostic accuracy comparable to conventional soft tissue reconstructions for pulmonary embolism detection.
- This technique avoids the need for additional reconstructions, leading to reduced processing time and storage costs.
- Thin effective slice thicknesses and overlapping reconstruction increments remain crucial for detailed CT analysis of pulmonary embolism at segmental and subsegmental levels, regardless of the processing method.