US5987094A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Computer-assisted method and apparatus for the detection of lung nodules
Assignee: University of South Florida
•
Filing: 1997-10-30
•
Granted: 1999-11-16
•
Expires: 2017-10-30
Abstract
Abstract A computer-assisted diagnostic (CAD) method and apparatus are described for the enhancement, detection, and classification of suspicious regions in digital x-ray images, with particular emphasis on lung nodule detection using chest x-ray images. An objective is to improve the sensitivity and specificity of detection of suspicious areas such as nodules, while maintaining a low false positive detection rate. A modular CAD technique has been developed to be potentially automatic and to be used as a second-opinion method for lung nodule detection. The method consists of using a plurality of CAD modules to preprocess and enhance image features, including image preprocessing, selective enhancement, segmentation, and feature extraction using a multiresolution/multiorientation wavelet transform and a computationally efficient filter, a 1.5 D circular pattern filter.
Inventors
- 👤 Laurence P. Clarke
- 👤 Wei Qian
- 👤 Fei Mao
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-07 22:11
CPC Classifications
G; G01; G01N; G01N23/00; G01N23/02; G01N23/04; G01N23/046; G06; G06T; G06T7/00; G06T7/0002; G06T7/0012; G01N2223/00; G01N2223/40; G01N2223/419; G01N2223/60; G01N2223/612
Claims Analysis
Total Claims: 37
Claim 1
Independent Claim
70 words
A computer-assisted method for detecting a lung nodule comprising the steps of: suppressing noise in a digital chest image; enhancing and segmenting anatomical structures in the noise-suppressed image; enhancing a potentially suspicious area in the noise-suppressed image; and further enhancing and segmenting the potentially suspicious area with the use of a filter having sensitivity to a shape feature of a lung nodule and having relative insensitivity to an overlapping object.
Claim 2
Dependent Claim
70 words
A computer-assisted method for detecting a lung nodule comprising the steps of: suppressing noise in a digital chest image; enhancing and segmenting anatomical structures in the noise-suppressed image; enhancing a potentially suspicious area in the noise-suppressed image; and further enhancing and segmenting the potentially suspicious area with the use of a filter having sensitivity to a shape feature of a lung nodule and having relative insensitivity to an overlapping object.
Claim 3
Dependent Claim
26 words
The method recited in claim 1, wherein the suspicious area further enhancing and segmenting step comprises applying a 1.5-dimensional filter having sensitivity to a circular pattern.
Claim 4
Dependent Claim
26 words
The method recited in claim 1, wherein the suspicious area segmenting step further comprises performing adaptive clustering to segment out a boundary of the suspicious area.
Claim 5
Dependent Claim
16 words
The lung-nodule detection method recited in claim 1, wherein the noise-suppressing step comprises adaptively filtering noise.
Claim 6
Dependent Claim
21 words
The lung-nodule detection method recited in claim 4, wherein the adaptive noise-filtering step comprises applying an adaptive nonlinear multistage nonlinear filter.
Claim 7
Dependent Claim
26 words
The lung-nodule detection method recited in claim 5, wherein the adaptive nonlinear multistage nonlinear filter is adapted to adaptively select a filter on a pixel-by-pixel basis.
Claim 8
Dependent Claim
28 words
The lung-nodule detection method recited in claim 6, wherein the filter-selecting step comprises selecting among a plurality of filters comprising a linear operator and an order statistic operator.
Claim 9
Dependent Claim
23 words
The lung-nodule detection method recited in claim 1, wherein the anatomical structure enhancing step comprises applying a wavelet transform to the noise-suppressed image.
Claim 10
Dependent Claim
29 words
The lung-nodule detection method recited in claim 8, wherein the wavelet transform applying step comprises applying one or more of a tree-structured wavelet transform and a directional wavelet transform.
Claim 11
Dependent Claim
23 words
The lung-nodule detection method recited in claim 1, wherein the anatomical structure enhancing step comprises applying a wavelet transform to the enhanced image.
Claim 12
Dependent Claim
30 words
The lung-nodule detection method recited in claim 10, wherein the wavelet transform applying step comprises applying a tree-structured wavelet transform and the anatomical structures segmenting step comprises applying adaptive clustering.
Claim 13
Dependent Claim
23 words
The lung-nodule detection method recited in claim 1, wherein the suspicious area enhancing step comprises subtracting enhanced anatomical structures from the noise-suppressed image.
Claim 14
Dependent Claim
24 words
The lung-nodule detection method recited in claim 12, wherein the suspicious area enhancing step further comprises applying a wavelet transform to the noise-suppressed image.
Claim 15
Dependent Claim
31 words
The lung-nodule detection method recited in claim 13, wherein the wavelet transform applying step comprises applying a wavelet-based multiresolution/multiorientation transform comprising a bank of wavelet filters implemented on two-stage adaptive combiners.
Claim 16
Dependent Claim
58 words
The lung-nodule detection method recited in claim 1, further comprising the steps, following the enhancing and segmenting steps, of: performing feature extraction on the segmented anatomical structures and potentially suspicious areas; differentiating a suspicious area from the potentially suspicious areas based upon the feature-extraction step; and classifying the suspicious area as a true positive or a false positive.
Claim 17
Dependent Claim
42 words
The lung-nodule detection method recited in claim 15, wherein the feature extraction step comprises extracting a region containing a potentially suspicious area and performing a feature calculation in a grey-level domain, a direction texture domain, a morphological domain, and a connectivity domain.
Claim 18
Dependent Claim
20 words
The lung-nodule detection method recited in claim 15, wherein the feature extraction step further comprises calculating positional and connectivity measurements.
Claim 19
Dependent Claim
25 words
The lung-nodule detection method recited in claim 17, wherein the suspicious area selecting and classifying steps comprise applying a neural network to the extracted region.
Claim 20
Dependent Claim
25 words
The lung-nodule detection method recited in claim 17, wherein the neural network applying step comprises applying a mixed-feature neural network using backpropagation with Kalman filtering.
Claim 21
Independent Claim
74 words
A system for detecting a lung nodule contained in a digital chest image comprising: means for suppressing noise in a digital chest image; means for enhancing and segmenting anatomical structures in the noise-suppressed image; means for enhancing potentially suspicious areas in the noise-suppressed image; and filter means for segmenting the potentially suspicious area, the filter means having sensitivity to a shape feature of a lung nodule and having relative insensitivity to an overlapping object.
Claim 22
Independent Claim
74 words
A system for detecting a lung nodule contained in a digital chest image comprising: means for suppressing noise in a digital chest image; means for enhancing and segmenting anatomical structures in the noise-suppressed image; means for enhancing potentially suspicious areas in the noise-suppressed image; and filter means for segmenting the potentially suspicious area, the filter means having sensitivity to a shape feature of a lung nodule and having relative insensitivity to an overlapping object.
Claim 23
Dependent Claim
53 words
The system recited in claim 20, further comprising: means for performing feature extraction on the raw and segmented anatomical structures and potentially suspicious areas; means for classifying a suspicious area as a true positive or a false positive from the potentially suspicious areas based upon input of computed features from the feature-extraction step.
Claim 24
Independent Claim
80 words
A system for detecting a lung nodule contained in a digital chest image comprising: a filter for suppressing noise in a digital chest image; a wavelet transform adapted to enhance anatomical structures in the noise-suppressed image; a subtraction algorithm for removing the enhanced anatomical structures from the noise-suppressed image; a wavelet transform adapted to enhance a potentially suspicious areas in the noise-suppressed image; and a 1.5-dimensional circular pattern filter for additionally enhancing a suspicious area from the enhanced, anatomic-structure-subtracted image.
Claim 25
Independent Claim
80 words
A system for detecting a lung nodule contained in a digital chest image comprising: a filter for suppressing noise in a digital chest image; a wavelet transform adapted to enhance anatomical structures in the noise-suppressed image; a subtraction algorithm for removing the enhanced anatomical structures from the noise-suppressed image; a wavelet transform adapted to enhance a potentially suspicious areas in the noise-suppressed image; and a 1.5-dimensional circular pattern filter for additionally enhancing a suspicious area from the enhanced, anatomic-structure-subtracted image.
Claim 26
Dependent Claim
48 words
The system recited in claim 22, further comprising: means for performing a feature extraction in a grey-level domain, a direction texture domain, and a morphological domain, for performing feature extraction on the segmented anatomical structures and potentially suspicious areas; and a neural network for classifying the suspicious area.
Claim 27
Independent Claim
98 words
A system for detecting a lung nodule contained in a digital chest image comprising: means for accessing digital chest image data; a processor in electronic communication with the data accessing means, the processor having resident therein a plurality of computer-assisted diagnostic modules for processing the digital chest image data comprising: a filter for suppressing noise in a digital chest image; a wavelet transform adapted to enhance anatomical structures in the noise-suppressed image; a wavelet transform adapted to enhance a potentially suspicious area in the noise-suppressed image; and a circular pattern filter for selectively further enhancing the suspicious area.
Claim 28
Independent Claim
98 words
A system for detecting a lung nodule contained in a digital chest image comprising: means for accessing digital chest image data; a processor in electronic communication with the data accessing means, the processor having resident therein a plurality of computer-assisted diagnostic modules for processing the digital chest image data comprising: a filter for suppressing noise in a digital chest image; a wavelet transform adapted to enhance anatomical structures in the noise-suppressed image; a wavelet transform adapted to enhance a potentially suspicious area in the noise-suppressed image; and a circular pattern filter for selectively further enhancing the suspicious area.
Claim 29
Dependent Claim
76 words
The system recited in claim 24, wherein the modules further comprise software means for performing a feature calculation in a grey-level domain, a direction texture domain, a morphological domain, and a location/connectivity domain, for performing feature extraction on the enhanced and segmented anatomical structures and potentially suspicious area; and a neural network for classifying the suspicious area as a potential lung nodule or a false positive based upon input of computed features from the feature-extraction module.
Claim 30
Dependent Claim
71 words
A modular software system for processing digital chest image data to detect a lung nodule, the system comprising: a filter module for suppressing noise in digital chest image data; a wavelet transform module adapted to enhance anatomical structures in the noise-suppressed image; a wavelet transform module adapted to enhance a potentially suspicious areas in the noise-suppressed image; and a 1.5-dimensional circular pattern filter module for further enhancing the enhanced suspicious area.
Claim 31
Dependent Claim
71 words
A modular software system for processing digital chest image data to detect a lung nodule, the system comprising: a filter module for suppressing noise in digital chest image data; a wavelet transform module adapted to enhance anatomical structures in the noise-suppressed image; a wavelet transform module adapted to enhance a potentially suspicious areas in the noise-suppressed image; and a 1.5-dimensional circular pattern filter module for further enhancing the enhanced suspicious area.
Claim 32
Dependent Claim
70 words
The system recited in claim 26, further comprising: a feature calculation module for performing feature extraction on the raw image and the segmented anatomical structures and potentially suspicious areas in a grey-level domain, a direction texture domain, a morphological domain, and a connectivity domain; and a neural network module for selecting a suspicious area from the potentially suspicious areas based upon the feature-extraction step and for classifying the suspicious area.
Claim 33
Dependent Claim
30 words
A data file containing a selected and classified suspicious area in a chest image, the data file having been created with the use of the software system of claim 26.
Claim 34
Independent Claim
138 words
A method for performing image processing to detect a lung nodule in a digital chest image at a first site, the image having been collected at a second site remote from the first site: electronically transmitting the image from the second site to the first site; preprocessing the image to segment out a potentially suspicious area by: suppressing noise in the image; enhancing and segmenting anatomical structures in the noise-suppressed image; enhancing a potentially suspicious area in the noise-suppressed image; and further enhancing the potentially suspicious area with the use of a filter having sensitivity to a shape feature of a lung nodule and having relative insensitivity to an overlapping object; segmenting the further enhanced potentially suspicious area with the use of adaptive clustering; and electronically transmitting the processed image from the first site to the second site.
Claim 35
Independent Claim
138 words
A method for performing image processing to detect a lung nodule in a digital chest image at a first site, the image having been collected at a second site remote from the first site: electronically transmitting the image from the second site to the first site; preprocessing the image to segment out a potentially suspicious area by: suppressing noise in the image; enhancing and segmenting anatomical structures in the noise-suppressed image; enhancing a potentially suspicious area in the noise-suppressed image; and further enhancing the potentially suspicious area with the use of a filter having sensitivity to a shape feature of a lung nodule and having relative insensitivity to an overlapping object; segmenting the further enhanced potentially suspicious area with the use of adaptive clustering; and electronically transmitting the processed image from the first site to the second site.
Claim 36
Independent Claim
205 words
A method for performing image processing to determine for a presence of a lung nodule in a digital chest image at a first site, the image having been collected at a second site remote from the first site: electronically transmitting the image from the second site to the first site; processing the image to determine whether a potentially suspicious area is present by segmenting out a potentially suspicious area by: suppressing noise in the image; enhancing and segmenting anatomical structures in the noise-suppressed image; determining whether a potentially suspicious area is present; if a potentially suspicious area is present, performing an image processing method on the area comprising the steps of: enhancing the area in the noise-suppressed image; and further enhancing the potentially suspicious area with the use of a filter having sensitivity to a shape feature of a lung nodule and having relative insensitivity to an overlapping object; and segmenting the further enhanced potentially suspicious area with the use of adaptive clustering; and if no potentially suspicious area is present, determining that no lung nodules are present; and electronically transmitting the processed image and a determination as to the presence or absence of a lung nodule from the first site to the second site.
Claim 37
Independent Claim
205 words
A method for performing image processing to determine for a presence of a lung nodule in a digital chest image at a first site, the image having been collected at a second site remote from the first site: electronically transmitting the image from the second site to the first site; processing the image to determine whether a potentially suspicious area is present by segmenting out a potentially suspicious area by: suppressing noise in the image; enhancing and segmenting anatomical structures in the noise-suppressed image; determining whether a potentially suspicious area is present; if a potentially suspicious area is present, performing an image processing method on the area comprising the steps of: enhancing the area in the noise-suppressed image; and further enhancing the potentially suspicious area with the use of a filter having sensitivity to a shape feature of a lung nodule and having relative insensitivity to an overlapping object; and segmenting the further enhanced potentially suspicious area with the use of adaptive clustering; and if no potentially suspicious area is present, determining that no lung nodules are present; and electronically transmitting the processed image and a determination as to the presence or absence of a lung nodule from the first site to the second site.
References Cited (Backward Citations)
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 2002-11-15 | 2003-05-17 | 2003-11-16 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2006-11-15 | 2007-05-17 | 2007-11-16 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2010-11-15 | 2011-05-17 | 2011-11-16 | $7700.00 | Window Closed / Lapsed |