Fiber imaging apparatus, methods, and applications
A flexible, artifact-free, and lensless fiber-based imaging system for biological objects. This system combines image reconstruction by a trained deep neural network with low-loss image transmission through disordered glass-air Anderson localized optical fiber. High quality images of biological objects can be obtained using short (few centimeters) or long (more than one meter) segments of disordered fiber with and without fiber bending. The deep neural network can also be designed to perform image classification. The system provides the unique property that the training performed within a straight fiber setup can be utilized for high fidelity reconstruction/classification of images that are transported through either straight or bent fiber making retraining for different bending situations unnecessary. In addition, high quality image transport and reconstruction is demonstrated for cells that are several millimeters away from the fiber input facet eliminating the need for additional optical elements at the distal end of the fiber. This novel imaging system shows great potential for practical applications in endoscopy including studies on freely behaving subjects.
1. A deep convolutional neural network glass-air Anderson localizing optical fiber (DCNN-GALOF) optical imaging system, comprising:
an incoherent, continuous, broadband (non-monochromatic) object illumination source operationally coupled to an object illumination channel;
an optical measurement path and an optical reference path optically coupled to the object illumination channel,
a length of a disordered transverse Anderson localizing fiber (‘Anderson fiber’) having a lensless distal end and a proximal end disposed in the optical measurement path;
an object illumination scanner disposed in the object illumination channel; and
a deep convolutional neural network having an input operationally coupled to the proximal end of the Anderson fiber to receive an input image of an biological object, disposed in the object illumination channel, said image transmitted through the Anderson fiber and an output, wherein the neural network is configured to perform an image reconstruction process on the input biological image to generate an output image that is a high fidelity image of the biological object.
2. The optical imaging system of claim 1 , wherein the length of the disordered transverse Anderson localizing fiber is between 2 cm and 3 m.
3. The optical imaging system of claim 1 , wherein the neural network is configured to provide an object classification after performing the image reconstruction process.
4. The optical imaging system of claim 1 , wherein the incoherent, continuous, broadband (non-monochromatic) object illumination source is a light emitting diode that emits light with a wavelength between 350 nm and 2100 nm.
5. The optical imaging system of claim 1 , wherein the continuous, broadband (non-monochromatic) object illumination source is a white light source.
6. The optical imaging system of claim 1 , wherein the light emitting diode (LED) object illumination source has a center wavelength of 460 nm.
7. The optical imaging system of claim 1 , wherein the biological object is a cellular structure.
8. A deep convolutional neural network glass-air Anderson localizing optical fiber (DCNN-GALOF) method for imaging a biological object, comprising:
providing a biological sample object;
illuminating the biological sample object with continuous, incoherent light;
scanning the biological sample object with the continuous, incoherent illumination over a plurality of vertical and horizontal positions to obtain training, validation, and test image datasets;
obtaining an image dataset of the biological object at each position;
splitting each image dataset into at least two copies;
sending one image dataset copy into a reference optical path and recording said reference image with a reference recorder;
transmitting another image dataset copy through a length of GALOF disposed in a measurement optical path and recording the GALOF output image with a measurement recorder;
dividing a plurality of reference and measurement image datasets into the training, validation, and test image datasets;
using the training and validation image datasets to train a deep learning convolutional neural network disposed in the system;
record an image of a biological sample to be identified, transmit it through the GALOF, record it with the measurement recorder, and transmit it through the deep learning convolutional neural network; and
recover a high-fidelity image of the new biological sample in real time.
9. The method of claim 8 , comprising providing the continuous, incoherent light from a LED.
10. The method of claim 8 , comprising providing the continuous, incoherent light from a white light source.