Direct extraction of boundaries from computed tomography scans

Area of Science:

• Medical Imaging
• Image Processing
• Computational Science

Background:

• Traditional X-ray image analysis often requires a full image reconstruction step before boundary extraction.
• Existing filtered backprojection (FBP) techniques are primarily designed for image reconstruction, not direct feature extraction.

Purpose of the Study:

• To develop a method for direct boundary extraction from X-ray raw data using FBP.
• To compute image differentials (gradient and Laplacian) without intermediate reconstruction.
• To enable direct image segmentation based on these computed differentials.

Main Methods:

• Preprocessing raw X-ray data to directly compute reconstructed gradient and Laplacian values.
• Utilizing an extension of FBP techniques adapted for gradient and Laplacian computation, particularly for noisy data.
• Applying the computed differential operators for direct slice segmentation.

Main Results:

• Demonstrated direct computation of image gradient and Laplacian values from raw X-ray data.
• Successfully segmented objects in X-ray images without prior full image reconstruction.
• Presented visual results of reconstructed gradients, Laplacians, and segmented objects.

Conclusions:

• The proposed FBP-based method allows for direct boundary extraction and segmentation of X-ray images.
• This approach bypasses the need for conventional image reconstruction, potentially improving efficiency.
• The technique is robust and adaptable, even for noisy datasets, offering a new tool in medical image analysis.