Image analysis – Applications – Biomedical applications
Reexamination Certificate
1998-08-27
2001-03-06
Patel, Jayanti K. (Department: 2723)
Image analysis
Applications
Biomedical applications
C378S037000
Reexamination Certificate
active
06198838
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to the field of computer aided diagnosis of medical images. In particular, the invention relates to a method and apparatus for detecting suspicious lesions in digital mammograms using an algorithm that independently computes mass information and spiculation information for allowing faster and more reliable identification of suspicious lesions.
BACKGROUND OF THE INVENTION
Breast cancer in women is a serious health problem. The American Cancer Society currently estimates that over 180,000 U.S. women are diagnosed with breast cancer each year. Breast cancer is the second major cause of cancer death among women. The American Cancer Society also estimates that breast cancer causes the death of over 44,000 U.S. women each year. While, at present, there is 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. Mammography using x-rays is currently the most common method of detecting and analyzing breast tumors.
Detection of suspicious, i.e. possibly cancerous, areas in mammograms is an important first step in the early diagnosis and treatment of breast cancer. While it is important to detect suspicious lesions when they are in the early stages, practical considerations can make this difficult. One complicating factor is that a typical mammogram may contain myriads of lines corresponding to fibrous breast tissue. The trained, focused eye of a medical professional, such as a radiologist, is needed to detect suspicious features among these lines. A typical radiologist may be required to examine hundreds of mammograms per day, leading to the possibility of a missed diagnosis due to fatigue and human error.
Recently, medical professionals have begun to use Computer Aided Diagnostic (CAD) tools to assist them in detecting suspicious features. Experiments have shown that the performance of radiologists improve when they are assisted by detection software that marks suspicious areas. See Brake and Karssemeijer, “Detection of Stellate Breast Abnormalities,”
Digital Mammography
pp. 341-346 (Elsevier Science 1996), the contents of which are hereby incorporated by reference into the present disclosure.
FIG. 1
shows a continuum of lesions that may appear in mammograms, ranging from a pure mass or pure density lesion on the left to a spiculated lesion on the right. Other types of lesions include architectural distortions, which have radiating lines similar to spiculated lesions but are generally without a central mass, and radial scars, which appear as criss-crossed lines and also are generally without a central mass.
Sharply defined masses such as those shown at the left in
FIG. 1
are rarely associated with malignant tumors, while spiculated masses can be a strong indication of malignancy. Lesions having the characteristics of architectural distortions or radial scars may also be cancerous, depending on their size and shape.
Accordingly, there is value in locating and analyzing both the “mass” or “density” qualities and the “spiculatedness” qualities of shapes found in digital mammograms. CAD systems generally include “mass” (or density) focused algorithms and “spiculation” focused algorithms. Some algorithms attempt to use metrics of both “massness” and “spiculatedness” to identify suspicious portions of digital mammograms. Such algorithms are included among the following references: Yin and Giger et. al., “Computerized Detection of Masses in Digital Mammograms: Analysis of Bilateral Subtraction Images,”
Medical Physics
, Vol. 18, No. 5, pp. 955-963 (Sept./Oct. 1991); Sahiner et. al., “Classification of Masses on Mammograms Using a RubberBand Straightening Transform and Feature Analysis,”
Medical Imaging, SPIE Symposium on Medical Imaging
Paper No. 2710-06, at p. 204 (1996); and Huo and Giger et al., “Analysis of Spiculation in the Computerized Classification of Mammographic Masses,”
Medical Physics
, Vol. 22, No. 10, pp. 1569-1579 (Oct. 1995). The contents of the above references are hereby incorporated by reference into the present application.
Typical of the above algorithms, Huo and Giger take a serial and dependent approach by first identifying masses and subsequently identifying spiculatedness characteristics of those masses. Huo and Giger demonstrate how the edges of detected masses can be used to determine a measure of spiculation—i.e., spiculatedness. Huo and Giger, approach the problem serially by first detecting the mass signature, or density, in a mammogram, and then applying various filtering analyses to filter out benign masses and false detections, such as parenchymal structure. The Huo and Giger approach initially depends on density, which is a feature with low positive predictive value. It then attempts to improve upon itself by measuring features with higher positive predictive values, such as spiculation. The shortcomings of the Huo/Giger approach include the fact that the additional feature measurements typically depend on secondary algorithms that may be non-robust. These secondary algorithms may include algorithms for spiculation, region growing, or segmentation of the mass boundaries. Also, these algorithms may display poor sensitivity on architectural distortions and radial scars which have no central density. Finally, because the algorithm is inherently serial, wherein the spiculation information is computed after the mass information, the time for completion of the algorithm is the sum of the time for completion of the mass detection algorithm plus the spiculation detection algorithm, which can lead to disadvantageously slow results.
A direct “backward direction” algorithm for of spiculation detection is disclosed in Karssemeijer, “Recognition of Stellate Lesions in Digital Mammograms,”
Digital Mammography: Proceedings of the
2
nd International Workshop on Digital Mammography
, York, England, Jul. 10-12, 1994, pp. 211-219 (Elsevier Science 1994), the contents of which are hereby incorporated by reference into the present application. By “backward direction” it is meant that a “candidate point” is incrementally moved across the image by a distance corresponding to the desired resolution of the spiculation search. At each candidate point, a set of “window computations” for a window of pixels surrounding the candidate point is performed, and a metric corresponding to the presence and/or strength of a spiculation centered on the candidate point is computed.
“Backward direction” algorithms are computationally intensive. For an image size of N×N pixels, there generally need to be on the order of K(bN)
2
computations, where K is the number of window computations for each candidate point and b is the reciprocal of the number of image pixels between each candidate point. Because the number K is often proportional to the square or cube of the window size, the computational intensity of “backward direction” approaches can easily become unwieldy. Another example of a “backward direction” spiculation algorithm is described in Kegelmeyer, “Computer-aided Mammographic Screening for Spiculated Lesions,”
Radioloqy
, Vol. 191, pp. 331-337 (1994), the contents of which are hereby incorporated by reference into the present application. The computational complexity of “backward direction” spiculation algorithms may cause a CAD program to be too slow for practical use by medical professionals, such as radiologists. Additionally, a practical implementation of a CAD system using a backward direction algorithm for spiculation detection would lead to inevitable dependency between the mass and spiculation algorithms. This is because, due to its slowness, the spiculation algorithm could only be applied to a subset of interesting regions of the digital mammogram image, the interesting regions being pointed out by the presence of masses from the mass detection algorithm.
Accordingly, it would be desirable to provide a computer-assisted diagnosis (CAD) system for assisting in the detection of suspicious lesions in medical ima
Doi Takeshi
Karssemeijer Nico
Roehrig Jimmy R.
Te Brake Guido M.
Patel Jayanti K.
Pennie & Edmonds LLP
R2 Technology, Inc.
LandOfFree
Method and system for detection of suspicious lesions in... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and system for detection of suspicious lesions in..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for detection of suspicious lesions in... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2543747