Anisotropy-based robust focus measure for non-mydriatic retinal imaging

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Non-mydriatic retinal imaging aids ophthalmic disease diagnosis and monitoring.
• Pupil dilation is not required, making it suitable for mass screening.
• Manual focusing in retinal cameras is prone to errors, especially with inexperienced users.

Purpose of the Study:

• To develop a novel and robust focus measure for non-mydriatic retinal imaging.
• To improve the reliability of retinal image acquisition for ophthalmic diagnostics.

Main Methods:

• A new focus measure was developed based on image anisotropy.
• Image anisotropy was calculated using the directional variance of the normalized discrete cosine transform.
• The proposed method was validated through simulations and experimental tests.

Main Results:

• The proposed focus measure demonstrated effectiveness in achieving accurate image focus.
• The method showed robustness in compensating for refractive errors.
• Simulations and experiments confirmed the proposed focus measure's performance.

Conclusions:

• The developed focus measure enhances the accuracy and reliability of non-mydriatic retinal imaging.
• This automated focusing technique is valuable for screening programs and clinical diagnostics.
• The method offers a significant improvement over manual focusing in retinal photography.