Automatic Segmentation of Bladder Layers in Optical Coherence Tomography Images Using Graph Theory and Dynamic Programming

Segmentation of anatomical and pathological structures in bladder wall images is crucial for diagnosis and study of bladder diseases. Manual segmentation, which is commonly used in this field, is often a time-consuming and subjective process. Those methods that have proposed normally are not suitable for bladder wall SDOCT (Spectral Domain Optical Coherence Tomography) images, although they do have really good effect in ophthalmic images. This paper presents an automatic approach for segmenting normal adult pig's bladder wall layers whose structure is similar to people's bladder wall in SDOCT images using graph theory and dynamic programming. Several other algorithms are implemented to segment the same images in order to compare the effect. The results show that our method accurately segments three bladder wall layer boundaries in normal pig bladder and significantly reduces the processing time required for image segmentation and feature extraction.