POSTOPERATIVE ASSESSMENT OF MULTIFOCAL AND EXTENDED DEPTH OF FOCUS (EDOF) IOL CENTRATION WITH NEURAL NETWORK

To evaluate the postoperative centration of multifocal and extended depth-of-focus (EDOF) intraocular lenses (IOLs) using a deep neural network.

Department of Ophthalmology, University Hospital of Alexandroupolis, Alexandroupolis, Greece.

Pre- and postoperative ocular biometry measurement was performed using an optical biometry system, which creates a surgery plan for image guided presbyopia correction by phacoemulsification with multifocal or EDOF IOL implantation. The collected data of both measurements was transferred to a digital marker microscope system to determine the pre- and postoperative visual axis. Furthermore, through a slit-lamp integrated anterior segment imaging system, an image of the post-operative IOL centration was taken after pharmacological mydriasis. Finally, the post-operative IOL centration images, as well as the images including the pre- and post-operative visual axis, was processed through a neural network. In this way, the post-operative IOL position was evaluated in comparison with the pre- and post-operative visual axis, in order to assess whether its centration is the desired one.

40 eyes were recruited in this study. All collected data was transferred to the neural network in order to be trained to evaluate the IOL centration. The assessment of the deep neural network showed if the coordinates of the postoperative visual axis of patients was placed within limits of the IOL’s central ring.

The accuracy of IOL centration is really important for the successful surgical correction of presbyopia. The developed neural network is a clinical assessment tool of this factor, that will improve the postoperative evaluation of the centration in order to evolve the presbyopia correction surgical technique.