Technique for finding the histogram region of interest for improved tone scale reproduction of digital radiographic images

• US 5,426,684 A
• Filed: 11/15/1993
• Issued: 06/20/1995
• Est. Priority Date: 11/15/1993
• Status: Expired due to Term

• Current Assignee: CareStream Health Incorporated
• Original Assignee: Eastman Kodak Company
• Inventors: Barski, Lori L., Gaborski, Roger S.
• Primary Examiner(s): Dzierzynski, Paul M.
• Assistant Examiner(s): Bruce, David Vernon
• Application Number:
  US08/152,772
• Time in Patent Office: 582 Days
• Field of Search: 364/413.13, 364/413.23, 395/130, 395/924, 378/62, 378/87, 378/98, 378/901
• US Class Current: 378/62
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 2207/30004 Biomedical image processing
• CPC Class Codes: G06T 5/40 using histogram techniques
• CPC Class Codes: G06T 5/92 based on global image properties
• CPC Class Codes: G06T 7/41 based on statistical description of texture
• CPC Class Codes: G16H 30/40 for processing medical images, e.g. editing
• 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
• CPC Class Codes: Y10S 378/901 Computer tomography program or processor
• CPC Class Codes: Y10S 706/924 Medical

Abstract

An image processing technique especially useful in processing digital radiographic images. A method for finding a histogram region of interest for improved tone scale reproduction of digital radiographic images includes the following steps. A digital radiographic image is randomly sampled with a sample having an appropriate size to delineate an object of interest. Each sample is processed using texture analysis techniques to extract a plurality of texture features. Using the extracted texture features, each sample is classified with a previously trained neural network classifier to determine its class. Last, the pixel values belonging to the same class are accumulated to form separate histograms for each class. Each of the histograms are then used to optimize tone scale reproduction.

Claims

1. A method of finding a histogram region of interest for improved tone scale reproduction of digital radiographic images comprising the steps of:

pseudo-randomly sampling a digital radiographic image with a sample having an appropriate size to delineate a region of interest;

processing each sample using texture analysis techniques to extract a plurality of texture features;

using the extracted texture features, classifying said sample with a previously trained neural network classifier to determine its class; and

accumulating pixel values belonging to the same class to form separate histograms for each said class, wherein each said histogram can be used to optimize tone scale reproduction.

2. The method of claim 1 including the step of segmenting said digital radiographic image into a background region, a foreground region, and an object region of interest, and wherein said sampling step is performed on said object region of interest.

3. The method of claim 1 wherein said sampling step includes a linear pixel shuffling technique.

4. The method of claim 1 wherein, before said processing step, each said sample is preprocessed by reducing the digital value of each sample pixel, and performing an adaptive histogram equalization operation on each said sample, and wherein said processing step is performed on the central region of said preprocessed sample.

5. The method of claim 1 wherein said processing step is performed using a gray-level co-occurrence matrix texture analysis technique to extract a first set of texture features, and also using a neighboring gray-level dependence matrix texture analysis technique to extract a second set of texture features, and wherein said classifying step is performed using said first and said second sets of texture features.

6. The method of claim 5 wherein said first set of texture features include energy, contrast, correlation, local homogeneity, and entropy, and wherein said second set of texture features include small number emphasis, large number emphasis, number nonuniformity, second moment, and entropy.

7. The method of claim 6 wherein said first and second sets of features are determined at intersample distances of 1, 2, and 3, such that each set of features include at least fifteen features, and wherein said thirty features are used in said neural network classifier of said classifying step.

8. The method of claim 1 wherein the histograms of one or more classes can be selectively manipulated to change the tonescale reproduction of said one or more classes.

9. The method of claim 8 wherein said histograms are of one or more body regions of interest and the tonescales of said one or more body regions can be selectively changed to emphasize said on or more body regions of interest.