Image analysis – Applications – Biomedical applications
Reexamination Certificate
1999-04-01
2001-11-20
Johns, Andrew W. (Department: 2621)
Image analysis
Applications
Biomedical applications
Reexamination Certificate
active
06320976
ABSTRACT:
BACKGROUND
1. Technical Field
The present invention relates generally to computer-assisted diagnosis (CAD) and, in particular, to a CAD method and system for automatically determining diagnostic saliency of digital images.
2. Background Description
Computer-assisted diagnosis is an important technology in many different clinical applications. However, one of the more prevalent clinical applications for computer-assisted diagnosis is in the detection of breast cancer in women. According to the American Cancer Society, breast cancer is the most common cancer among women, other than skin cancer. It is the leading cause of death among women aged 40 to 55. There are approximately 179,000 new cases of breast cancer in the United States each year and about 43,500 deaths from the disease.
While there are presently no means for preventing breast cancer, early detection of the disease prolongs life expectancy and decreases the likelihood of the need for a total mastectomy. Accordingly, the American Cancer Society recommends that all women aged 40 and older should have a mammogram every year.
Diagnostic images such as mammograms typically contain large, diagnostically unimportant regions. These regions may belong to the background or to body parts uninteresting for the purposes of the present study. A human diagnostician is able to quickly identify and focus only on diagnostically relevant patches in the image. Knowledge of relative diagnostic saliency of image regions can increase the effectiveness and efficiency of computer aided diagnosis (CAD) and other digital image processing.
When humans look at an image, certain locations in the image typically visually “stand out” from the rest. In medical images, however, diagnostically salient regions (i.e., image regions the content of which is likely to influence the outcome of diagnosis) can have visually insignificant appearances. Human diagnosticians learn by experience to recognize salient regions in diagnostic images. A medical image such as, for example, a mammogram, may contain background structures corresponding to healthy breast tissue. Accordingly, a trained, focused eye of a radiologist is needed to detect small lesions among these structures. However, a typical radiologist may be required to examine up to hundreds of mammograms on a daily basis, leading to the possibility of a missed diagnosis due to human error. Thus, it would be desirable and highly advantageous to have a CAD method and system for automatically determining diagnostic saliency of digital images.
A knowledge of the diagnostic saliency of regions in a digital image, in addition to guiding a human reader to the interesting portions of the image, is also useful for increasing the efficiency and effectiveness of many digital image processing methods. For example, image display can be improved by enhancing diagnostically salient regions, optionally using lesion-specific enhancement operators. Moreover, image matching (e.g., bilateral, temporal or inter-view change detection for mammograms), can be accomplished more robustly by de-emphasizing diagnostically unimportant regions. Also, higher compression ratios may be achieved for image storage or transmission, without the loss of diagnostic quality, by allotting more bits or storage for diagnostically salient regions than for non-salient regions. Additionally, computer-aided lesion detection methods will benefit from a knowledge of which portions of an image are more important. Further, a knowledge of the diagnostic saliency of image regions can help reduce false positive findings of automatic lesion detection methods. The preceding are but some of the many applications to which knowledge about diagnostic saliency of a digital image may be applied.
SUMMARY OF THE INVENTION
The present invention is directed to a computer-assisted diagnosis method and system for automatically determining diagnostic saliency of digital images.
In one aspect of the present invention, there is provided a computer assisted diagnosis system for automatically determining diagnostic saliency of regions in a digital image comprised of a plurality of pixels. The system includes: a memory unit, a plurality of filters stored in the memory unit, each of the plurality of filters designed to identify a specific type of diagnostic finding, and associated with a virtual window for defining regions in the image at which the filter is applied, a set of training image patches corresponding to typical appearances of the specific type of diagnostic finding, a distance measure between the training image patches and the regions in the image defined by the virtual window, and a feature set corresponding to the distance measure; and a processor for applying each of the plurality of filters to the image to compute distances between the regions in the image defined by the virtual window and the training image patches based on the distance measures and the feature sets, and ranking regions in the image corresponding to a particular type of diagnostic finding based on the computed distances.
In another aspect of the present invention, there is provided a computer assisted diagnosis method for automatically determining diagnostic saliency of locations in a digital image comprised of a plurality of pixels. The method includes the steps of: providing a plurality of filters for evaluating the image, wherein each of the plurality of filters is designed to identify a specific type of diagnostic finding, and is associated with a virtual window for defining regions in the image at which the filter is applied, a set of training image patches corresponding to typical appearances of the specific type of diagnostic finding, a distance measure between the training image patches and the regions in the image defined by the virtual window, and a feature set corresponding to the distance measure; applying the plurality of filters to the image to compute distances between the regions in the image defined by the virtual window and the training image patches based on the distance measure and the feature set, for each of the plurality of filters; and ranking regions in the image as corresponding to a particular type of diagnostic finding based the computed distances.
These and other aspects, features and advantages of the present invention will become apparent from the following detailed description of preferred embodiments, which is to be read in connection with the accompanying drawings.
REFERENCES:
patent: 5287275 (1994-02-01), Kimura
patent: 5680485 (1997-10-01), Loce et al.
patent: 5710833 (1998-01-01), Moghaddam et al.
patent: 5822742 (1998-10-01), Alkon et al.
patent: 5828769 (1998-10-01), Burns
patent: 5857030 (1999-01-01), Gaborski et al.
patent: 5870493 (1999-02-01), Vogl et al.
patent: 5923779 (1999-07-01), Ohmi et al.
patent: 5983218 (1999-11-01), Syeda-Mahmood
patent: 6058206 (2000-05-01), Kortge
patent: 6163629 (2000-12-01), Cheung et al.
Murthy Sreerama K. V.
Novak Carol L.
Qian Jianzhong
Johns Andrew W.
Nakhjavan Shervin
Paschburg Donald B.
Siemens Corporate Research Inc.
LandOfFree
Computer-assisted diagnosis method and system for... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Computer-assisted diagnosis method and system for..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Computer-assisted diagnosis method and system for... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2585949