Computer-assisted counting of retinal cells by automatic segmentation after TV denoising

Area of Science:

• Ophthalmology
• Neuroscience
• Cell Biology

Background:

• Accurate quantification of retinal cells is crucial for studying retinal development, aging, and degeneration.
• Manual cell counting is laborious and prone to errors.
• Existing automated methods face limitations due to biological and preparation variability.

Purpose of the Study:

• To develop an adaptive algorithm for automated, quantitative evaluation of retinal cell mosaics.
• To overcome limitations of manual counting and existing automated techniques.

Main Methods:

• Utilized digital micrographs of cone photoreceptor mosaics from mammalian retinal wholemounts.
• Employed Rudin-Osher-Fatemi total variation (TV) denoising for cell segmentation.
• Developed a batch processing module in MATLAB adaptable with three manually adjusted parameters.

Main Results:

• Achieved error ratios below 10% in 16 out of 18 tested samples.
• Demonstrated a mean error of less than 6% compared to manual cell counts.
• Successfully segmented photoreceptors from background, debris, and other cell types.

Conclusions:

• The developed module enables traceable, automated acquisition of retinal cell density data.
• Potential for further error reduction through additional parameters.
• The module is suitable for integration into advanced imaging and analysis environments, including 3D applications.