US5933519A — Cytological slide scoring apparatus

Claim 1

Independent Claim 174 words

A classifier development method comprising the steps of: a) developing a preliminary classifier for use in an automated, computer controlled, classification system using a set of collected objects; b) executing the preliminary classifier and any additional classifiers in the automated, computer controlled, classification system; c) evaluating the preliminary classifier and any additional classifiers using interactive data display and visualization methods to develop statistical quantities to characterize data distribution and decision boundaries, wherein said characterization includes the step of determining an error rate of the preliminary classifier and any additional classifiers; d) examining features and images of collected objects that the preliminary classifier and any additional classifiers failed to correctly classify so as to determine failure causes; e) comparing performance of the preliminary classifier and any additional classifiers with predetermined goals for feature selection to provide evaluation results; and f) building an additional classifier if the evaluation results show the current performance level falls short of the desirable goal and repeating steps b)-e) using only objects not previously classified correctly to build the additional classifier.

Claim 2

Dependent Claim 174 words

A classifier development method comprising the steps of: a) developing a preliminary classifier for use in an automated, computer controlled, classification system using a set of collected objects; b) executing the preliminary classifier and any additional classifiers in the automated, computer controlled, classification system; c) evaluating the preliminary classifier and any additional classifiers using interactive data display and visualization methods to develop statistical quantities to characterize data distribution and decision boundaries, wherein said characterization includes the step of determining an error rate of the preliminary classifier and any additional classifiers; d) examining features and images of collected objects that the preliminary classifier and any additional classifiers failed to correctly classify so as to determine failure causes; e) comparing performance of the preliminary classifier and any additional classifiers with predetermined goals for feature selection to provide evaluation results; and f) building an additional classifier if the evaluation results show the current performance level falls short of the desirable goal and repeating steps b)-e) using only objects not previously classified correctly to build the additional classifier.

Claim 3

Dependent Claim 45 words

The classifier development method of claim 1 further comprising the steps of further refining the preliminary classifier and any additional classifiers by acquiring additional data and refined features, and then rebuilding the preliminary classifier and any additional classifiers using the additional data and refined features.

Claim 4

Dependent Claim 22 words

The method of claim 1 wherein the step of developing a preliminary classifier includes the step of developing a preliminary cytological classifier.

Claim 5

Dependent Claim 23 words

The method of claim 1 wherein the step of building an additional classifier includes the step of developing a probable abnormal cell classifier.

Claim 6

Dependent Claim 22 words

The method of claim 1 wherein the step of building an additional classifier includes the step of developing an endocervical cell classifier.

Claim 7

Dependent Claim 21 words

The method of claim 1 wherein the step of building an additional classifier includes the step of developing an artifact classifier.

Claim 8

Dependent Claim 24 words

The method of claim 1 wherein the step of building an additional classifier includes the step of developing a probable endocervical cell group classifier.

Claim 9

Dependent Claim 23 words

The method of claim 1 wherein the step of building an additional classifier includes the step of building a probable metaplastic/parabasal group classifier.

Claim 10

Dependent Claim 23 words

The method of claim 1 wherein the step of building an additional classifier includes the step of building an uncertain endocervical/metaplastic/parabasal group classifier.

Claim 11

Dependent Claim 24 words

The method of claim 1 wherein the step of building an additional classifier includes the step of building a potential abnormal cell group classifier.

Claim 12

Dependent Claim 135 words

An object classifier training process for an automated classification system, the object classifier training process comprising the steps of: a) semi-automatically acquiring images of objects of interest; b) refining the acquired images by segmentation processing to produce segmented objects; c) labeling the segmented objects with their true classes to produce labeled objects; d) using the labeled objects for object classifier training for generating an object classifier; e) testing the object classifier and, if the performance of the classifier is unsatisfactory, then objects that the object classifier failed to classify correctly are acquired in an automated object acquisition mode; and f) building a subsequent classifier using only the objects that the current object classifier failed to classify correctly and repeating steps a)-f) until the performance of the initial object classifier and subsequent object classifiers are satisfactory.

Claim 13

Dependent Claim 135 words

An object classifier training process for an automated classification system, the object classifier training process comprising the steps of: a) semi-automatically acquiring images of objects of interest; b) refining the acquired images by segmentation processing to produce segmented objects; c) labeling the segmented objects with their true classes to produce labeled objects; d) using the labeled objects for object classifier training for generating an object classifier; e) testing the object classifier and, if the performance of the classifier is unsatisfactory, then objects that the object classifier failed to classify correctly are acquired in an automated object acquisition mode; and f) building a subsequent classifier using only the objects that the current object classifier failed to classify correctly and repeating steps a)-f) until the performance of the initial object classifier and subsequent object classifiers are satisfactory.

Claim 14

Dependent Claim 25 words

The object classifier training process of claim 11 wherein the step of generating an object classifier comprises the step of generating an endocervical cell classifier.

Claim 15

Dependent Claim 24 words

The object classifier training process of claim 11 wherein the step of generating an object classifier comprises the step of generating an artifact classifier.

Claim 16

Dependent Claim 27 words

The object classifier training process of claim 11 wherein the step of generating an object classifier comprises the step of generating a probable endocervical cell group classifier.

Claim 17

Dependent Claim 26 words

The object classifier training process of claim 11 wherein the step of generating an object classifier comprises the step of generating a probable metaplastic/parabasal group classifier.

Claim 18

Dependent Claim 26 words

The object classifier training process of claim 11 wherein the step of generating an object classifier comprises the step of generating an uncertain endocervical/metaplastic/parabasal group classifier.

Claim 19

Dependent Claim 26 words

The object classifier training process of claim 11 wherein the step of generating an object classifier comprises the step of generating a potential abnormal cell group.

Claim 20

Dependent Claim 64 words

The object classifier training process of claim 11 wherein the step of semi-automatically acquiring images comprises the steps of: selecting a slide from a training slide set; qualifying training objects from the slide; establishing a slide level diagnosis; acquiring training object coordinates; acquiring training object images located at the training object coordinates; segmenting the acquired images; and extracting features corresponding to the segmented objects.

Claim 21

Independent Claim 94 words

A method for development of multi-stage object classifiers for use in an automated classification system, the method comprising the steps of: using acquired images from an object library to generate an initial classifier; operating the initial classifier in the automated classification system to identify objects; conducting an automated data collection process to acquire additional training samples; and using the additional training samples and the acquired images to train a plurality of subsequent classifiers by using newly acquired data wherein the newly acquired data are the samples that the current classifiers failed to classify correctly.

Claim 22

Independent Claim 94 words

A method for development of multi-stage object classifiers for use in an automated classification system, the method comprising the steps of: using acquired images from an object library to generate an initial classifier; operating the initial classifier in the automated classification system to identify objects; conducting an automated data collection process to acquire additional training samples; and using the additional training samples and the acquired images to train a plurality of subsequent classifiers by using newly acquired data wherein the newly acquired data are the samples that the current classifiers failed to classify correctly.

Claim 23

Dependent Claim 12 words

The method of claim 19 wherein the acquired images comprise cytological images.

Claim 24

Dependent Claim 23 words

The method of claim 1 wherein the step of building an additional classifier includes the step of developing an endometrial cell group classifier.

Claim 25

Dependent Claim 22 words

The method of claim 1 wherein the step of building an additional classifier includes the step of developing an intermediate cell classifier.

Claim 26

Dependent Claim 26 words

The object classifier training process of claim 11 wherein the step of generating an object classifier comprises the step of generating an endometrial cell group classifier.

Claim 27

Dependent Claim 25 words

The object classifier training process of claim 11 wherein the step of generating an object classifier comprises the step of generating an intermediate cell classifier.

Annuity Stage	Window Opens	Payment Deadline	Surcharge Period Until	Estimated Fee	Status
3.5 Years (Year 4)	2002-08-02	2003-02-01	2003-08-03	$2000.00	Window Closed / Lapsed
7.5 Years (Year 8)	2006-08-02	2007-02-01	2007-08-03	$3760.00	Window Closed / Lapsed
11.5 Years (Year 12)	2010-08-02	2011-02-01	2011-08-03	$7700.00	Window Closed / Lapsed

Cytological slide scoring apparatus

Abstract

Inventors

Bibliographic Details

Claims Analysis

References Cited (Backward Citations)

USPTO Annuity & Maintenance Windows

Assigned Categories & Tags

Assign Category Tag