A logarithmic opinion pool based STAPLE algorithm for the fusion of segmentations with associated reliability weights

Area of Science:

• Medical imaging
• Biomedical engineering
• Computational anatomy

Background:

• Pelvic floor dysfunction is a common postpartum condition.
• Accurate segmentation of pelvic floor magnetic resonance images (MRI) aids diagnosis and treatment.
• Manual segmentation is time-consuming and prone to significant inter- and intra-rater variability.
• Existing template fusion algorithms struggle with template alignment and segmentation inaccuracies.

Purpose of the Study:

• To develop a novel algorithm for automatic segmentation of female pelvic floor MRI.
• To improve segmentation accuracy by addressing template registration and segmentation errors.
• To enhance decision fusion in template-based segmentation methods.

Main Methods:

• Developed a novel algorithm for automatic pelvic floor MRI segmentation.
• The algorithm estimates and compensates for imperfect template registration and template segmentation inaccuracies.
• Utilized a local image similarity measure to infer local reliability weights.
• Employed a novel logarithmic opinion pooling for decision fusion.

Main Results:

• The novel algorithm demonstrated superior performance compared to nine state-of-the-art segmentation methods.
• The algorithm effectively compensates for registration and segmentation errors in template fusion.
• Achieved the highest performance in automatic segmentation of female pelvic floor MRI.

Conclusions:

• The developed algorithm offers a significant advancement in automatic pelvic floor MRI segmentation.
• This method has the potential to improve the diagnosis and treatment of pelvic floor dysfunction.
• The novel approach overcomes limitations of previous template fusion techniques.