Neural network powered microscopic system for cataract surgery

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Phacoemulsification with intraocular lens (IOL) implantation is a standard cataract treatment.
• Current surgical microscopes face challenges in complex clinical scenarios, impacting surgical precision.
• Accurate eye localization and orientation are critical for successful surgical outcomes.

Purpose of the Study:

• To develop a neural network-powered surgical microscopic system for enhanced cataract surgery guidance.
• To address the unmet clinical need for high-precision navigation in ophthalmic surgery.
• To improve the real-time positioning and alignment of IOLs during surgery.

Main Methods:

• An end-to-end navigation network using an attention-based encoder-decoder architecture for eye center localization.
• A combination of siamese network, correlation filter, and spatial transformation network for tracking eye rotation.
• Development of a computer-assisted surgical microscope with integrated imaging, display, control software, and algorithms.
• Collection and augmentation of 100 clinical surgery videos for training the neural network.

Main Results:

• The proposed system demonstrates a neural network-powered approach for precise surgical navigation.
• The system integrates advanced algorithms for real-time eye localization and rotation tracking.
• Evaluation on a human eye model indicates promising navigation performance for ophthalmic procedures.

Conclusions:

• The developed AI-powered surgical microscopic system offers a potential solution for improving precision in cataract surgery.
• This technology can assist surgeons by providing real-time guidance for critical steps like IOL implantation.
• Further validation and integration into clinical practice could significantly enhance patient outcomes in ophthalmic surgery.