Image segmentation method

• US 6,839,462 B1
• Filed: 12/22/1997
• Issued: 01/04/2005
• Est. Priority Date: 12/23/1996
• Status: Expired due to Fees

• Current Assignee: Norman Smith, Richard I. Kitney
• Original Assignee: Norman Smith, Richard I. Kitney
• Inventors: Smith, Norman, Kitney, Richard I.
• Primary Examiner(s): Werner, Brian
• Assistant Examiner(s): Edwards, Patrick
• Application Number:
  US09/331,667
• Time in Patent Office: 2,570 Days
• Field of Search: 382/128, 382/154, 382/170, 382/171, 382/173, 382/225, 382/299
• US Class Current: 382/173
• CPC Class Codes: G01B 11/2518 Projection by scanning of the object
• CPC Class Codes: G06F 18/23213 with fixed number of clusters, e.g. K-means clustering
• CPC Class Codes: G06T 2207/10088 Magnetic resonance imaging [MRI]
• CPC Class Codes: G06T 2207/30008 Bone
• CPC Class Codes: G06T 2207/30016 Brain
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/143 involving probabilistic approaches, e.g. Markov random field [MRF] modelling
• CPC Class Codes: G06V 10/267 by performing operations on regions, e.g. growing, shrinking or watersheds
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
• CPC Class Codes: G06V 2201/031 of internal organs

Abstract

A method of producing a simulated 3-dimensional image, the method comprising: assimilating 2-dimensional image data, e.g. from MR image slices; assigning a feature, e.g. gray scale value to each datum; reducing the resolution of each gray scale value of the data; generating a histogram of the reduced resolution gray scale data; performing a fast fuzzy c-means clustering on the histogram data. The resolution of each pixel object in the images may be reduced from 12 bit to 8 bit resolution, and the histogram may be generated from these. Subsequently a value may be assigned for each entry in the histogram, each entry value being equal to the number of objects of any given feature (e.g. gray scale) in the reduced resolution (8 bit) image. The image may be displayed using a novel color blending technique. A 3D image is produced quickly and without supervisory intervention and can be used in endoscopic surgery and in diagnostic methods as well as in understanding healthy anatomical features better.

Claims

1. A method of classifying grey scale image data according to different region types within the image, the method comprising:

assigning a single feature value to each datum;

reducing the resolution of each value of the data;

generating a histogram of the reduced resolution value data; and

,

performing a fast fuzzy c-means clustering algorithm on all of the histogram data, so as to minimize a fuzzy object function J_m(W, v), wherein J_m(W, v) is defined by the equation $J_m<br><br>\left(W,v\right)=\underset{g=G_{\min }}{\overset{G_{\max }}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>\underset{i=1}{\overset{c}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>f<br><br>\left(g\right)<br><br>{\left(w_{\mathrm{ig}}\right)}^{\mathrm{nl}}<br><br>{\left(d_{\mathrm{ig}}\right)}^2,$

 

using generated values f(g) of said reduced resolution histogram, and wherein;

c represents a number of clusters in the image;

G_min and G_max represent respectively a minimum and maximum grey scale value q in the image;

m is a fuzzy weighting exponent;

w is a fuzzy partition of the histogram data with entries w_ig which correspond to a membership value of each grey scale value g in a cluster i of the image; and

,

d_ig is a Euclidian distance between a grey scale value q and v₁.

2. A method according to claim 1 wherein the image data is assimilated from magnetic resonance imaging data.

3. A method according to claim 2 wherein the magnetic resonance data is segmented by generating a histogram of reduced resolution object data.

4. A method according to claim 1 wherein the feature assigned to each object datum is a grey scale value.

5. A method according to claim 1 wherein the resolution is reduce from 12 bits to 8 bits.

6. A method according to claim 1 wherein a value is assigned for each entry in the histogram, each entry value being equal to the number of objects of the given feature in the reduced resolution histogram.

7. A method of classifying grey scale image data according to claim 1, further comprising defining a color space of c-1 equally spaced dimensions, assigning the cluster corresponding to no signal to the dimensional origin, assigning a dimension in the color space to each other cluster giving a value in each dimension to each pixel, which value is proportional to the fuzzy membership of the pixel in the respective cluster to which the dimension is assigned, and displaying a visual image by re-generating the pixels according to their assigned values and dimensions.

8. A method according to claim 7 wherein the output of the fast fuzzy c-means (ffcm) clustering is translated into a grey scale LUT of RBG (Red/Blue/Green) values.

9. A method according to claim 8 further comprising the step of displaying the color blended fuzzy segmented data on a VDU.