Fully automatic prostate segmentation from transrectal ultrasound images based on radial bas-relief initialization...

Area of Science:

• Medical Imaging
• Computational Anatomy
• Biomedical Engineering

Background:

• Prostate segmentation in transrectal ultrasound (TRUS) images is crucial for prostate cancer diagnosis and treatment planning.
• Accurate segmentation is challenging due to image quality and anatomical variations.

Purpose of the Study:

• To develop a fully automatic, slice-based method for segmenting the prostate in TRUS images.
• To improve the accuracy and efficiency of prostate segmentation compared to existing methods.

Main Methods:

• A novel method combining radial bas-relief (RBR) and false edge removal determined the initial prostate contour.
• A 2D slice-based propagation approach utilized a level-set evolution model driven by dyadic wavelet transform-generated energy fields.
• An automated method selected the optimal initial slice for propagation based on contour accuracy assessment.

Main Results:

• The algorithm achieved an average mean absolute difference (MAD) of 0.79±0.26mm compared to manual segmentations.
• Accurate segmentation was demonstrated across mid-gland, base, and apex regions of the prostate.
• The method's performance is comparable to previously reported semi-automatic segmentation techniques.

Conclusions:

• The developed fully automatic slice-based segmentation method effectively segments the prostate in TRUS images.
• The approach provides accurate results comparable to manual and semi-automatic methods.
• This technique holds promise for enhancing prostate cancer diagnosis and treatment planning.