US 7,321,796 B2 | ||
Method and system for training a visual prosthesis | ||
Wolfgang Fink, Montrose, Calif. (US); and Mark Humayun, Glendale, Calif. (US) | ||
Assigned to California Institute of Technology, Pasadena, Calif. (US); and Doheny Eye Institute, Los Angeles, Calif. (US) | ||
Filed on Apr. 30, 2004, as Appl. No. 10/837,163. | ||
Claims priority of provisional application 60/467037, filed on May 01, 2003. | ||
Prior Publication US 2004/0236389 A1, Nov. 25, 2004 | ||
Int. Cl. A61N 1/36 (2006.01) |
U.S. Cl. 607—54 | 33 Claims |
1. A method for training a visual prosthesis comprising:
presenting a non-visual reference stimulus corresponding to a reference image to a patient having a visual prosthesis, the
visual prosthesis including a plurality of electrodes;
generating a training data set by presenting a series of stimulation patterns to the patient through the visual prosthesis
and receiving a corresponding series of subjective patient selections of a preferred stimulation pattern among stimulation
patterns in the series of stimulation patterns previously presented to the patient until the patient identifies a successful
stimulation pattern, and storing the presented stimulation patterns and the received selections of the patient;
wherein at least one stimulation pattern in the series of stimulation patterns is determined at least in part based on a fitness
function optimization algorithm and at least one previously received subjective patient selection in the series of subjective
patient selections; and
after identifying the successful stimulation pattern, presenting off-line to a neural network at least the reference image
and the successful stimulation pattern from the training data set to determine a vision solution.
|