Granted Patent B2

US 7,542,950 · App. 11/471,965 · Granted Jun 2, 2009

Abstract

Claims (38)

1. A method for automatically producing data representative of a modified head shape from data representative of a deformed head for utilization in a system for producing cranial remodeling devices, comprising:

providing a neural network trained with a first plurality of first sets of data, each said first set comprising captured data for a corresponding one first head shape, and with a second plurality of second sets of captured data, each said second set comprising captured data for a modified head shape for a corresponding one of said first head shapes;

utilizing said neural network to operate on data representative of a deformed head to produce data representative of a modified head shape;

providing a computer readable memory; and

storing said data representative of a modified head shape in said computer readable memory for use in producing a cranial remodeling device.

2. A method for automatically producing data representative of a modified head shape from data representative of a deformed head, comprising:

providing one or more computer readable memories comprising a first database and a second database;

providing a neural network;

providing to said neural network a first plurality of first sets of captured data from said first database, each said first set comprising captured data for a corresponding one first head shape cast;

providing to said neural network a second plurality of second sets of captured data from said second database, each said second set comprising captured data for a modified head shape cast for a corresponding one of said first head shapes;

training said neural network with said first plurality of first sets of captured data and said second plurality of second sets of captured data such that said neural network will operate on a set of captured data for a first head shape to produce a new data set defining a unique modified head shape corresponding to said first head shape; and

storing said new data set as a third database in one of said one or more computer readable memories.

3. A method in accordance with claim 2 , comprising:

providing a support vector machine;

utilizing said support vector machine to train said neural network.

4. A method in accordance with claim 3 , wherein:

said support vector machine comprises a least squares support vector machine.

5. A system for automatically producing data representative of a modified head shape from data representative of a deformed head, comprising:

a first database comprising a first plurality of first sets of data, each said first set comprising data for a corresponding one first head shape;

a second database comprising a second plurality of second sets of data, each said second set comprising captured data for a modified head shape for a corresponding one of said first head shapes;

a neural network program; and

a processor operable with said neural network program to train said neural network program with said first plurality of first sets of data and said second plurality of second sets of data such that said neural network operates on a set of data for a first deformed head shape to produce a data set defining a unique corresponding modified head shape.

6. A system in accordance with claim 5 , comprising:

a support vector machine;

said a support vector machine program operable to train said neural network program.

7. A system in accordance with claim 6 , wherein:

said support vector machine comprises a least squares support vector machine.

8. A method for automatically producing data representative of a modified three dimensional shape from data representative of a first three dimensional shape, comprising:

providing a neural network;

utilizing said neural network to operate on a first set of data representative of a first shape to produce a second set of data corresponding to a modified three dimensional head shape; and

storing said second set of data in a computer readable memory for utilization in forming a cranial orthosis.

9. A method in accordance with claim 8 , comprising:

training said neural network with first plurality of first sets of data and said second plurality of second sets of data such that said neural network operates on a set of captured data for a three dimensional shape to produce a corresponding modified three dimensional shape.

10. A method in accordance with claim 9 , comprising:

providing a support vector machine; and

utilizing said support vector machine to train said neural network.

11. A method in accordance with claim 10 , wherein:

said support vector machine comprises a least squares support vector machine.

Assignments (9)