Texture analysis of differently reconstructed PET images

Area of Science:

• Medical Imaging
• Radiological Sciences
• Computational Imaging

Background:

• Positron emission tomography (PET) image quality is affected by limited projections and radioactive decay statistics.
• Diverse reconstruction algorithms can lead to distinct image textures.
• Understanding these textures is crucial for accurate image interpretation.

Purpose of the Study:

• To investigate the impact of different reconstruction algorithms on PET image texture.
• To quantify and compare texture characteristics across various methods.
• To assess the reliability and classification capabilities of texture analysis in PET.

Main Methods:

• Examined four PET image reconstruction methods: filtered backprojection and three algebraic approaches.
• Employed grey-level morphological operators for texture extraction.
• Quantified texture using directionality, fractal dimension, and lacunarity parameters.

Main Results:

• Increasing iteration steps in algebraic methods refined texture from coarse to fine granularities.
• All methods exhibited similar texture evolution pathways during iteration.
• Filtered backprojection textures were encompassed within the set produced by algebraic methods.
• Texture remained independent of count numbers above ten per pixel in the region of interest.
• Artefacts were confined to regions lacking activity.

Conclusions:

• PET image reconstruction methods yield comparable texture characteristics.
• Texture analysis provides reliable image assessment and enables tissue classification.
• The study confirms high reliability in reconstructed PET images across tested algorithms.