US5873824A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Apparatus and method for computerized analysis of interstitial infiltrates in chest images using artificial neural networks
Assignee: Arch Development Corp
•
Filing: 1996-11-29
•
Granted: 1999-02-23
•
Expires: 2016-11-29
Abstract
Abstract An automated computer-aided diagnosis (CAD) method and system using artificial neural networks (ANNs) for the quantitative analysis of image data. Three separate ANNs were applied for detection of interstitial disease on digitized two-dimensional chest images. The first ANN was trained with horizontal profiles in regions of interest (ROIs) selected from normal and abnormal chest radiographs. The second ANN was trained using vertical output patterns obtained from the 1st ANN for each ROI. The output value of the 2nd ANN was used to distinguish between normal and abnormal ROIS with interstitial infiltrates. If the ratio of the number of abnormal ROIs to the total number of all ROIs in a chest image was greater than a certain threshold level, the chest image was considered abnormal. In addition, the third ANN was applied to distinguish between normal and abnormal chest images where the chest image was not clearly normal or abnormal. The ANN trained with image data learns some statistical properties associated with interstitial infiltrates in chest radiographs. In addition, the same technique can be applied to higher-dimensional data (e.g., three-dimensional data and four-dimensional data including time-varying three-dimensional data).
Inventors
- 👤 Kunio Doi
- 👤 Takayuki Ishida
- 👤 Shigehiki Katsuragwa
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-09 22:54
CPC Classifications
G; G16; G16H; G16H50/00; G16H50/20; G06; G06T; G06T7/00; G06T7/0002; G06T7/0012; G06V; G06V2201/00; G06V2201/03; Y; Y10; Y10S; Y10S128/00; Y10S128/92; Y10S128/925
Claims Analysis
Total Claims: 30
Claim 1
Independent Claim
73 words
An apparatus for generating a partial diagnosis based on image data, the apparatus comprising: a plurality of one-dimensional neural networks connected in a tree of levels, the tree of levels comprising a) a plurality of first-level one-dimensional neural networks receiving strips of two-dimensional image data and outputting one-dimensional image data, and b) a second-level one-dimensional neural network receiving the one-dimensional data output from the plurality of first-level networks and outputting a partial diagnosis.
Claim 2
Independent Claim
73 words
An apparatus for generating a partial diagnosis based on image data, the apparatus comprising: a plurality of one-dimensional neural networks connected in a tree of levels, the tree of levels comprising a) a plurality of first-level one-dimensional neural networks receiving strips of two-dimensional image data and outputting one-dimensional image data, and b) a second-level one-dimensional neural network receiving the one-dimensional data output from the plurality of first-level networks and outputting a partial diagnosis.
Claim 3
Dependent Claim
56 words
The apparatus as claimed in claim 1, further comprising: an evaluator for pre-classifying a plurality of images into a corresponding complete diagnosis of one of normal, abnormal, and undecided; and a classification neural network for receiving statistics on the images classified as undecided and for determining a corresponding complete diagnosis of one of normal and abnormal.
Claim 4
Dependent Claim
70 words
The apparatus as claimed in claim 2, wherein the evaluator comprises: a normal evaluator for determining that the complete diagnosis is normal if a ratio is below a minimum value, the ratio obtained by dividing a number of images which are abnormal by a number of the plurality of images; and an abnormal evaluator for determining that the complete diagnosis is abnormal if the ratio is above a maximum value.
Claim 5
Dependent Claim
41 words
The apparatus as claimed in claim 1, wherein each neural network of the plurality of one-dimensional neural networks comprises: a first layer including 32 input units; a second layer including 16 hidden units; and a third layer including one output unit.
Claim 6
Dependent Claim
39 words
The apparatus as claimed in claim 1, further comprising: means for training the plurality of one-dimensional neural networks on a level-by-level basis using plural training iterations until an area under an ROC curve is greater than a minimum value.
Claim 7
Dependent Claim
21 words
The apparatus as claimed in claim 1, further comprising: a pre-filter for pre-filtering raw image data to create the image data.
Claim 8
Dependent Claim
36 words
The apparatus as claimed in claim 6, wherein the pre-filter comprises an optical density correction device for performing an optical density correction using a gradient curve of a medium from which the image data was generated.
Claim 9
Dependent Claim
62 words
The apparatus as claimed in claim 1, wherein the image data is initially three-dimensional image data, the apparatus further comprising: a plurality of third-level one-dimensional neural networks connected to the tree of levels, wherein the plurality of third-level one-dimensional neural networks receive strips of the three-dimensional image data and output the two-dimensional image data to the plurality of first-level one-dimensional neural networks.
Claim 10
Dependent Claim
68 words
The apparatus as claimed in claim 1, wherein the image data is initially at least three-dimensional image data, the apparatus further comprising: a plurality of higher-levels of one-dimensional neural networks connected to the tree of levels, wherein the plurality of higher-levels of one-dimensional neural networks receive strips of the at least three-dimensional image data and output the two-dimensional image data to the plurality of first-level one-dimensional neural networks.
Claim 11
Independent Claim
100 words
A method of generating a partial diagnosis based on image data, the method comprising the steps of: applying strips of two-dimensional image data to a plurality of first-level one-dimensional neural networks included in a plurality of one-dimensional neural networks connected in a tree of levels; outputting one-dimensional image data from the plurality of first-level one-dimensional neural networks; receiving the one-dimensional image data at a second-level one dimensional neural network of the plurality of one-dimensional neural networks connected in a tree of levels; and outputting a partial diagnosis from the second-level one-dimensional neural network of the plurality of one-dimensional neural networks.
Claim 12
Independent Claim
100 words
A method of generating a partial diagnosis based on image data, the method comprising the steps of: applying strips of two-dimensional image data to a plurality of first-level one-dimensional neural networks included in a plurality of one-dimensional neural networks connected in a tree of levels; outputting one-dimensional image data from the plurality of first-level one-dimensional neural networks; receiving the one-dimensional image data at a second-level one dimensional neural network of the plurality of one-dimensional neural networks connected in a tree of levels; and outputting a partial diagnosis from the second-level one-dimensional neural network of the plurality of one-dimensional neural networks.
Claim 13
Dependent Claim
64 words
The method as claimed in claim 10, further comprising the steps of: pre-classifying a plurality of images into a corresponding complete diagnosis of one of normal, abnormal, and undecided; applying to a classification network statistics of the images classified as undecided; and determining, using the classification neural network, a corresponding complete diagnosis of one of normal and abnormal for the images classified as undecided.
Claim 14
Dependent Claim
64 words
The method as claimed in claim 11, wherein the step of pre-classifying comprises: calculating a ratio by dividing a number of images which are abnormal by a number of the plurality of images; determining that the complete diagnosis is normal if the ratio is below a minimum value; and determining that the complete diagnosis is abnormal if the ratio is above a maximum value.
Claim 15
Dependent Claim
74 words
The method as claimed in claim 10, wherein the step of outputting one-dimensional image data comprises the steps of: applying a strip of two-dimensional image data to a first layer including 32 input units; applying outputs of the 32 input units to a second layer including 16 hidden units; and applying outputs of the 16 hidden units to a third layer including one output unit to produce one datum for the one-dimensional image data.
Claim 16
Dependent Claim
40 words
The method as claimed in claim 10, further comprising the step of: training the plurality of one-dimensional neural networks on a level-by-level basis using plural training iterations until an area under an ROC curve is greater than a minimum value.
Claim 17
Dependent Claim
21 words
The method as claimed in claim 10, further comprising the step of: pre-filtering raw image data to create the image data.
Claim 18
Dependent Claim
32 words
The method as claimed in claim 15, wherein the step of pre-filtering comprises performing an optical density correction using a gradient curve of a medium from which the image data was generated.
Claim 19
Dependent Claim
58 words
The method as claimed in claim 10, wherein the image data is initially three-dimensional image data, the method further comprising the steps of: receiving strips of the three-dimensional image data at a plurality of third-level one-dimensional neural networks connected to the tree of levels; and outputting the two-dimensional image data to the plurality of first-level one-dimensional neural networks.
Claim 20
Dependent Claim
72 words
The method as claimed in claim 10, wherein the image data is initially at least three-dimensional image data, the method further comprising the steps of: receiving strips of the at least three-dimensional image data at a plurality of higher-levels of one-dimensional neural networks connected to the tree of levels; and outputting, from the plurality of higher-levels of one-dimensional neural networks, the two-dimensional image data to the plurality of first-level one-dimensional neural networks.
Claim 21
Dependent Claim
150 words
A computer program product, comprising: a computer storage medium and a computer program code mechanism embedded in the computer storage medium for causing a computer to generate a partial diagnosis based on image data, the computer program code mechanism comprising: a first computer code device configured to implement a plurality of one-dimensional neural networks connected in a tree of levels, wherein the tree of levels includes a plurality of first-level one-dimensional neural networks and a second-level one-dimensional neural network; a second computer code device configured to apply strips of two-dimensional image data to the plurality of first-level one-dimensional neural networks of the plurality of one-dimensional neural networks; a third computer code device configured to output one-dimensional image data from the plurality of first-level one-dimensional neural networks to the second-level one-dimensional neural network; and a fourth computer code device configured to output a partial diagnosis based on the one-dimensional image data.
Claim 22
Dependent Claim
150 words
A computer program product, comprising: a computer storage medium and a computer program code mechanism embedded in the computer storage medium for causing a computer to generate a partial diagnosis based on image data, the computer program code mechanism comprising: a first computer code device configured to implement a plurality of one-dimensional neural networks connected in a tree of levels, wherein the tree of levels includes a plurality of first-level one-dimensional neural networks and a second-level one-dimensional neural network; a second computer code device configured to apply strips of two-dimensional image data to the plurality of first-level one-dimensional neural networks of the plurality of one-dimensional neural networks; a third computer code device configured to output one-dimensional image data from the plurality of first-level one-dimensional neural networks to the second-level one-dimensional neural network; and a fourth computer code device configured to output a partial diagnosis based on the one-dimensional image data.
Claim 23
Dependent Claim
98 words
The computer program product as claimed in claim 19, further comprising: a fifth computer code device configured to pre-classify a plurality of images into a corresponding complete diagnosis of one of normal, abnormal, and undecided; a sixth computer code device configured as a classification neural network; a seventh computer code device configured to apply to the classification network statistics of the images classified as undecided; and an eighth computer code device configured to classify an output of the classification neural network as a corresponding complete diagnosis of one of normal and abnormal for the images classified as undecided.
Claim 24
Dependent Claim
81 words
The computer program product as claimed in claim 20, wherein the fifth computer code device comprises: a ninth computer code device configured to determine that the complete diagnosis is normal if a ratio is below a minimum value, the ratio obtained by dividing a number of images which are abnormal by a number of the plurality of images; and a tenth computer code device configured to determine that the complete diagnosis is abnormal if the ratio is above a maximum value.
Claim 25
Dependent Claim
68 words
The computer program product as claimed in claim 19, wherein the first computer code device comprises: a fifth computer code device configured as a first layer including 32 input units; a sixth computer code device configured as a second layer including 16 hidden units; and a seventh computer code device configured as a third layer including one output unit to produce one datum for the one-dimensional image data.
Claim 26
Dependent Claim
46 words
The computer program product as claimed in claim 19, further comprising: a fifth computer code device configured to train the plurality of one-dimensional neural networks on a level-by-level basis using plural training iterations until an area under an ROC curve is greater than a minimum value.
Claim 27
Dependent Claim
27 words
The computer program product as claimed in claim 19, further comprising: a fifth computer code device configured to pre-filter raw image data to create the image data.
Claim 28
Dependent Claim
37 words
The computer program product as claimed in claim 24, wherein the fifth computer code device is configured to perform an optical density correction using a gradient curve of a medium from which the image data was generated.
Claim 29
Dependent Claim
73 words
The computer program product as claimed in claim 19, wherein the image data is initially three-dimensional image data, the computer program product further comprising: a fifth computer code device configured to receive strips of the three-dimensional image data at a plurality of third-level one-dimensional neural networks connected to the tree of levels; and a sixth computer code device configured to output the two-dimensional image data to the plurality of first-level one-dimensional neural networks.
Claim 30
Dependent Claim
86 words
The computer program product as claimed in claim 19, wherein the image data is initially at least three-dimensional image data, the computer program product further comprising: a fifth computer code device configured to receive strips of the at least three-dimensional image data at a plurality of higher-levels of one-dimensional neural networks connected to the tree of levels; and a sixth computer code device configured to output, from the plurality of higher-level one-dimensional neural networks, the two-dimensional image data to the plurality of first-level one-dimensional neural networks.
References Cited (Backward Citations)
US5622171A
Cited
US5491627A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 2002-02-22 | 2002-08-24 | 2003-02-23 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2006-02-22 | 2006-08-24 | 2007-02-23 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2010-02-22 | 2010-08-24 | 2011-02-23 | $7700.00 | Window Closed / Lapsed |