Method for the autonomous image segmentation of flow systems

Patent metadata

Field	Value
Publication number	US-9836849-B2
Application number	US-201615007510-A
Country	US
Kind code	B2
Filing date	Jan 27, 2016
Priority date	Jan 28, 2015
Publication date	Dec 5, 2017
Grant date	Dec 5, 2017

Abstract

Disclosed herein is a method that comprises obtaining an image of a network section through which flow occurs; where the flow is selected from a group consisting of fluid, electrons, protons, neutrons and holes; subjecting the image to a low pass filter to increase contrast in portions of the network sections; computing a local mean of visible light intensity at each pixel that is present in the image; calculating a visible light intensity difference between each pixel and the local mean of visible light intensity and producing a differentiated image using this calculation; creating a base image of the differentiated image; where the base image comprises a hand segmented gold standard dataset; removing objects below a minimum threshold size from the base image; and retaining the remaining objects if they approximate the line or spine.

First claim

What is claimed is:

1. A method comprising:
obtaining an image of a network section through which flow occurs; where the flow is selected from a group consisting of fluid, electrons, protons, neutrons and holes;
minimizing visible light intensity differences in the image;
subjecting the image to a low pass filter to reduce the images' high frequency components;
computing a local mean of visible light intensity at each pixel that is present in the image;
calculating a visible light intensity difference between each pixel and the local mean of visible light intensity and producing a differentiated image using this calculation;
performing a wavelet transformation on the differentiated image;
creating a base image of the differentiated image; where the base image comprises a hand segmented gold standard dataset;
removing objects below a minimum threshold size from the base image;
testing remaining objects in the base image for its ability to approximate a line; and
retaining the remaining objects if they approximate the line; or alternatively reassembling disconnected objects if they demonstrate a least change in direction to form a repaired image of the network section.

2. The method of claim 1 , where minimizing the visible light intensity differences in the image comprises subtracting a portion of light intensity from those portions of the image that have a greater light intensity than other portions.

3. The method of claim 1 , where the low pass filter is a Gaussian blur.

4. The method of claim 1 , where the low pass filter comprises convolving the image with a Gaussian function.

5. The method of claim 1 , where computing a local mean of visible light intensity at each pixel comprises taking a moving average in the shape of a circle of all pixels in a radius of 30 to 100 pixels.

6. The method of claim 1 , where performing a wavelet transformation comprises taking a weighted sum of each pixel, where the weight is decided by each pixel distance from a center pixel.

7. The method of claim 1 , where the wavelet transformation comprises a Mexican Hat wavelet.

8. The method of claim 1 , where the line is straight.

9. The method of claim 1 , where the line is curved.

10. A system for performing a constructal analysis, the system comprising a processor and a memory to perform a method comprising:
obtaining an image of a network section through which flow occurs; where the flow is selected from a group consisting of fluid, electrons, protons, neutrons and holes; where the network section comprises an apparent random pathway, pattern, or network;
minimizing visible light intensity differences in the image;
subjecting the image to a low pass filter to reduce the images' high frequency components;
computing a local mean of visible light intensity at each pixel that is present in the image;
calculating a visible light intensity difference between each pixel and the local mean of visible light intensity and producing a differentiated image using this calculation;
performing a wavelet transformation on the differentiated image;
creating a base image of the differentiated image; where the base image comprises a hand segmented gold standard dataset;
removing objects below a minimum threshold size from the base image;
testing remaining objects in the base image for its ability to approximate a line; and
retaining the remaining objects if they approximate the line; or alternatively reassembling disconnected objects if they demonstrate a least change in direction to form a repaired image of the network section.

11. The system of claim 10 , where the system is used to evaluate images for grading diabetic retinopathy, to evaluate therapeutic responses of anti-angiogenic drugs in choroidal neovascularization, to evaluate optic neuritis along with degeneration of the retinal nerve fiber layer that is highly associated with multiple sclerosis, to evaluate ocular migraines associated with systemic vascular disease and high blood pressure, to evaluate the condition of blood vessels and/or nerves when affected by hypertension, chronic kidney failure, atherosclerosis, pulmonary diseases such as emphysema, chronic bronchitis, asthma, chronic obstructive pulmonary disease, interstitial lung disease and pulmonary embolism, cardiovascular diseases, myocardial infarction, aneurysms, transient ischemic attack, brain diseases, concussions, Alzheimer's disease and/or strokes.

12. The system of claim 10 , where the apparent random pathway, pattern, or network is a vascular network of blood vessels in a living being.

13. The system of claim 12 , where the vascular network of blood vessels are present in a retina, a heart, a brain, breast, kidney, and/or a lung of a human being.

14. The system of claim 10 , where the image is obtained using magnetic resonance imaging, computed tomography, ultrasound, ultrasound thermography, opto-acoustics, infrared imaging, positron emission tomography, visible light photography and xray imaging.

15. The system of claim 14 , where the image is further subjected to at least one of filtering, thresholding, digitization, and image and/or feature recognition.

16. The system of claim 15 , further comprising deriving at least one quantitative measure from the smoothed network.

17. The system of claim 16 , where the at least one quantitative measure is an end to end distance of the apparent random pathway, pattern, or network; an end to end distance of a portion of the apparent random pathway, pattern, or network; a radius of gyration of at least one branch or a plurality of branches of the apparent random pathway, pattern, or network; a persistence length of a branch or a plurality of branches of the apparent random pathway, pattern, or network; an average length between branches of the apparent random pathway, pattern, or network; an average branch length of the apparent random pathway, pattern, or network; an average orientation of the apparent random pathway, pattern, or network with respect to another apparent random pathway, pattern, or network; or the tortuosity of a branch or a plurality of branches of the apparent random pathway, pattern, or network.

Assignees

• Univ Florida

Inventors

• Dickrell Iii Daniel John
• Panikulam Jacob J

Classifications

• G06T7/143

  involving probabilistic approaches, e.g. Markov random field [MRF] modelling · CPC title

• G06T2207/30041

  Eye; Retina; Ophthalmic · CPC title

• G06T2207/20104

  Interactive definition of region of interest [ROI] · CPC title

• G06T7/168 Primary

  involving transform domain methods · CPC title

• G06T2207/20064

  Wavelet transform [DWT] · CPC title