Hybrid neural network and multiple fiber probe for in-depth 3-D mapping

• US 5,660,181 A
• Filed: 12/12/1994
• Issued: 08/26/1997
• Est. Priority Date: 12/12/1994
• Status: Expired due to Term

• Current Assignee: Physical Optics Corporation
• Original Assignee: Physical Optics Corporation
• Inventors: Lu, Taiwei, Ho, Zonh-Zen
• Primary Examiner(s): Kamm, William E.
• Assistant Examiner(s): EVANISKO, GEORGE ROBERT
• Application Number:
  US08/354,317
• Time in Patent Office: 988 Days
• Field of Search: 128/664, 128/665, 356/326, 356/328
• US Class Current: 600/408
• CPC Class Codes: A61B 1/00193 adapted for stereoscopic vision
• CPC Class Codes: A61B 1/00194 adapted for three-dimensional imaging
• CPC Class Codes: A61B 1/042 characterised by a proximal camera, e.g. a CCD camera
• CPC Class Codes: A61B 1/043 for fluorescence imaging
• CPC Class Codes: A61B 1/07 using light-conductive means, e.g. optical fibres
• CPC Class Codes: A61B 5/0059 using light, e.g. diagnosis by transillumination, diascopy, fluorescence A61B5/0093, A61B5/1455, A61B5/02416 and A61B5/0261 take precedence
• CPC Class Codes: A61B 5/7264 Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems neural networks per se G06N3/00; expert systems per se G06N5/00
• CPC Class Codes: A61B 5/7267 involving training the classification device
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: Y10S 128/925 Neural network

Abstract

An apparatus for in-depth three dimensional tumor mapping including (A) a light source; (B) a multi-fiber bundle including at least one illumination fiber and at least two receiving fibers, the at least one illumination fiber being connected to the light source; (C) a spectrometer connected to the at least two receiving fibers; and (D) a hybrid neural network connected to the spectrometer, said hybrid neural network including a principle component analysis processor and a neural network classifier.

Claims

1. An apparatus for providing data for in-depth three dimensional mapping of an object, said apparatus comprising:

(A) a light source;

(B) at least one illumination fiber, said at least one illumination fiber being connected to said light source so as to illuminate said object;

(C) at least two receiving fibers, said at least two receiving fibers being angularly displaced with respect to each other so as to receive light reflected from said object at different angles;

(D) a spectrometer connected to said at least two receiving fibers; and

(E) a hybrid neural network connected to said spectrometer, said hybrid neural network including a principle component analysis processor and a neural network classifier, and said hybrid neural network providing data for a three dimensional map of said object.

2. The apparatus of claim 1 further comprising a monitor for displaying said data as a three-dimensional map, said monitor being connected to said hybrid neural network.

3. The apparatus of claim 1 further comprising a robotic and which is connected to and which positions said at least one illumination fiber and said at least two receiving fibers inside a patient'"'"'s body.

4. The apparatus of claim 1, further comprising a charge coupled device, said charge coupled device being connected to said spectrometer, and wherein said spectrometer further includes

a holographic filter connected to said at least two receiving fibers, and

a wavelength division multiplexer connected to said holographic filter.

5. The apparatus of claim 4 wherein said charge coupled device includes a two-dimensional photoarray detector and said wavelength division multiplexer includes a Littrow grating and a lens geometrically arranged for paraxial transmission.

6. The apparatus of claim 1 wherein said light source includes a selfoc lens connected to said at least one illuminating fiber, an SMA connector attached to said selfoc lens and a monochromatic light source attached to said SMA connector.

7. The apparatus of claim 1 wherein said at least one illumination fiber and said at least two receiving fibers are part of a single multi-fiber bundle.

8. In an endoscope, the improvement comprising the apparatus of claim 1.

9. A method of in-depth three dimensional tumor mapping comprising:

(A) producing an initial representation;

(B) providing

i) a light source,

ii) a multi-fiber bundle including at least one illuminating fiber and at least two receiving fibers, said at least one illuminating fiber being connected to said light source,

iii) a spectrometer connected to said at least two receiving fibers,

iv) a hybrid neural network connected to said spectrometer and

v) a monitor displaying said initial representation, said monitor being connected to said hybrid neural network;

(C) positioning said multi-fiber bundle proximal an object to be analyzed, said object being fluorescent;

(D) illuminating said object with photons from said light source to obtain fluorescence;

(E) transmitting fluorescence from said object to said spectrometer through said at least two receiving fibers;

(F) transforming said fluorescence into spectra with said spectrometer;

(G) transmitting spectra from said spectrometer to said hybrid neural network;

(H) processing said spectra with said hybrid neural network to obtain a principal component analysis of said spectra by extracting a set of orthogonal feature vectors to represent said spectra;

(I) classifying said set of orthogonal feature vectors with said hybrid neural network to obtain a set of results; and

(J) transforming said initial representation to display said set of results with said monitor.

10. The method claim 9 wherein extracting said set of orthogonal feature vectors includes generating said set of orthogonal feature vectors with a Gram-Schmidt orthogonalization algorithm.