Wavelet energy-guided level set-based active contour: a segmentation method to segment highly similar regions

Area of Science:

• Medical Imaging
• Image Segmentation
• Computational Biology

Background:

• Accurate segmentation of regions with intra-region intensity variations and similar adjacent distributions is challenging in medical imaging.
• Computed tomography (CT) imaging requires robust methods for identifying abnormalities like tumors.

Purpose of the Study:

• To introduce a novel segmentation approach, the wavelet energy-guided level set-based active contour (WELSAC) model.
• To address the limitations of existing methods in segmenting complex regions within CT images, particularly for tumor detection.

Main Methods:

• Adaptation of wavelet energy from wavelet transform to represent region information.
• Embedding wavelet energy into a level set model to formulate the WELSAC segmentation model.
• Evaluation using synthetic and CT images, with a focus on tumor cases.

Main Results:

• The WELSAC model successfully segmented regions of interest in CT images.
• Segmentation results closely corresponded with manual delineations by medical experts.
• Demonstrated effectiveness in cases with intra-region intensity variations and high similarity with adjacent regions.

Conclusions:

• The proposed WELSAC model provides an effective solution for segmenting challenging regions in CT images.
• The WELSAC model shows promise as a tool for accurate tumor detection and segmentation.