METHODS AND APPARATUS TO INTEGRATE SYSTEMATIC DATA SCALING INTO GENETIC ALGORITHM-BASED FEATURE SUBSET SELECTION

• US 20100177943A1
• Filed: 08/02/2007
• Published: 07/15/2010
• Est. Priority Date: 08/11/2006
• Status: Active Grant

• Current Assignee: Koninklijke Philips N.V.
• Original Assignee: Koninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
• Inventors: Boroczky, Lilla, Zhao, Luyin, Lee, Kwok Pun
• Granted Patent: US 8,311,310 B2
• Time in Patent Office: Days
• US Class Current: 382/131
• CPC Class Codes: G06F 18/2111 by using evolutionary computational techniques, e.g. genetic algorithms
• CPC Class Codes: G06N 3/126 Evolutionary algorithms, e.g. genetic algorithms or genetic programming
• CPC Class Codes: G06V 10/771 Feature selection, e.g. selecting representative features from a multi-dimensional feature space
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

Methods and apparatus for training a system for developing a process of data mining, false positive reduction, computer-aided detection, computer-aided diagnosis and artificial intelligence are provided. A method includes choosing a training set from a set of training cases using systematic data scaling and creating a classifier based on the training set using a classification method. The classifier yields fewer false positives. The method is suitable for use with a variety of data mining techniques including support vector machines, neural networks and decision trees.

Claims

1. A method of improving classification accuracy and reducing false positives in data mining, computer aided-detection, computer-aided diagnosis and artificial intelligence, the method comprising:

choosing a training set from a set of training cases using systematic data scaling;

and,

creating a classifier based on the training set using a classification method, wherein the systematic data scaling method and the classification method produce the classifier thereby reducing false positives and improving classification accuracy.

2. The method according to claim 1 wherein the classifier is selected from the group consisting of:

support vector machines, neural networks, and decision trees.

3. The method according to claim 1, the method further comprising evaluating the classifier produced by the classification method based on the training set using a testing set.

4. The method according to claim 1 wherein choosing further comprises removing from the training set false nodules that form Tomek links with true nodules until a threshold is met.

5. The method according to claim 4 wherein the threshold is determined with respect to a downscaling factor, x, such that the number of false nodules remaining in the training set after systematic data scaling is no more than x times the number of true nodules in the training set.

6. The method according to claim 1 wherein the method further comprises validating the classifier with the set of training cases or a subset thereof.

7. A genetic algorithm that when executed implements the method of claim 1.

8. The genetic algorithm according to claim 7 wherein the genetic algorithm is the CHC algorithm.

9. A method of choosing features from a feature pool, the method comprising:

providing each of a first genetic algorithm and a second genetic algorithm according to claim 7, wherein the first genetic algorithm is used to determine the best size of the feature set; and

,

fixing the feature set size and using the second genetic algorithm to select the features.

10. The method according to claim 9 wherein in providing the first genetic algorithm, the method further comprises analyzing results using at least one of:

number of occurrences of chromosomes representing different feature subset sizes and number of average errors.

11. The method according to claim 10, wherein number of average errors is a number of misclassified lung nodules.

12. A computer-readable medium which when executed implements the method of claim 1.

13. An article of manufacture which is an imaging device or a false positive reduction device, wherein the device is a computer that is programmed to analyze image data by implementing the method of claim 1.

14. The article of manufacture according to claim 13, wherein the imaging device is selected from the group consisting of:

computed tomography (CT), computed axial tomography (CAT), multi-slice computed tomography (MSCT), body section roentgenography, ultrasound, magnetic resonance imaging (MRI), magnetic resonance tomography (MRT), nuclear magnetic resonance (NMR), X-ray, microscopy, fluoroscopy, tomography, and digital imaging.

15. The article of manufacture according to claim 13, wherein the article of manufacture is a lung nodule CAD system.