Image analysis – Applications – Biomedical applications
Reexamination Certificate
2001-07-17
2003-10-28
Patel, Jayanti K. (Department: 2623)
Image analysis
Applications
Biomedical applications
C382S194000
Reexamination Certificate
active
06640001
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to the field of computer aided diagnosis of abnormal lesions in medical images. In particular, the invention relates to a fast algorithm for detecting spiculated or stellar lesions in a digital mammogram to assist in the detection of malignant breast cancer tumors at an early stage in their development.
BACKGROUND OF THE INVENTION
Breast cancer in women is a serious health problem, the American Cancer Society currently estimating 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 estimating 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 lesions.
The detection of spiculated, or stellar-shaped, lesions (“spiculations”) in mammograms is of particular importance because a spiculated breast tumor has a relatively high probability of being malignant. While it is important to detect the spiculated lesions as early as possible, i.e. when they are as small as possible, practical considerations can make this difficult. In particular, a typical mammogram may contain myriads of lines corresponding to fibrous breast tissue, and the trained, focused eye of a radiologist is needed to detect small spiculated lesions among these lines. Moreover, a typical radiologist may be required to examine hundreds of mammograms per day, leading to the possibility of a missed diagnosis due to human error.
Accordingly, the need has arisen for a computer-assisted diagnosis (CAD) system for assisting in the detection of abnormal lesions, including spiculations, in medical images. The desired CAD system digitizes x-ray mammograms to produce a digital mammogram, and performs numerical image processing algorithms on the digital mammogram. The output of the CAD system is a highlighted display which directs the attention of the radiologist to suspicious portions of the x-ray mammogram. The desired characteristics of a spiculation-detecting CAD system are high speed (requiring less processing time), high precision (the ability to detect subtle spiculations), and high accuracy (the ability to avoid false positives and missed spiculations). It may also be desired that the spiculation-detecting CAD system also be used as a mass-detecting and mass-classifying CAD system, and that the CAD system be capable of using spiculation information in conjunction with mass information for identifying suspicious masses in the digital mammogram and directing the attention of the radiologist to both the spiculations and the suspicious masses.
One method for detecting spiculations in digital mammograms, proposed by Kegelmeyer et al and referred to as the “Alignment of Local Oriented Edges” (ALOE) algorithm, is described in Kegelmeyer, “Computer-aided Mammographic Screening for Spiculated Lesions,” Radiology 191:331-337 (1994). The ALOE method first calculates local gradients in a digitized mammogram. For each “candidate point” in the image, a predetermined window around that point is selected, the window size being some fraction of the overall image size. An “ALOE signal” for each candidate point then is calculated based on information in the surrounding window, the ALOE signal being defined as the standard deviation of a histogram of the gradient directions of all pixels in the window. The next candidate point, offset from the previous candidate point by a distance corresponding to the desired resolution of the search, is then considered.
Keeping in mind that a spiculation is a roughly symmetric set of lines radiating from a central point or region, a histogram of gradient directions will tend to be a flat distribution from 0 to 360 degrees if a spiculated region is centered around the candidate point. Thus, because the ALOE signal is the standard deviation of the histogram, the ALOE signal will be lower for those candidate points which are at the centers of spiculations, and will be higher for those candidate points which are not at the centers of spiculations. After the ALOE signal is calculated for all candidate points in the image, local minima in a plot of the ALOE signals are used as a basis for identifying spiculations.
The ALOE algorithm has several disadvantages. The primary disadvantage is that, in addition to spiculations, many unwanted background objects can also produce a small ALOE signal. For example, a point that is surrounded by a circle, such as the border of a circumscribed mass, also produces gradients in all directions, and therefore will produce a local minimum ALOE signal. A false positive may result. Furthermore, a typical spiculation in an actual mammogram will not have lines radiating in every direction, but rather will have lines radiating in several discrete directions in rough symmetry about the center. Thus, because every direction may not be present in the histogram of gradient angles around the center of a spiculation, the standard deviation of the histogram may still be quite large, resulting in a larger ALOE signal. This spiculation may then be missed. Thus, the ALOE algorithm represents serious practical problems because it may yield false positives and may also miss certain spiculations.
The ALOE algorithm is representative of a class of “backward direction” spiculation detection algorithms. 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. Thus, for example, the ALOE algorithm computes the “ALOE signal” for each candidate point, and then moves on to the next candidate point.
As a general observation, “backward direction” algorithms are computationally intensive. This is because, for an image size of N×N, there will 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 where 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 get out of hand.
A second method for detecting spiculations in digital mammograms, proposed by Karssemeijer et al., is described 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 (Elsevier Science 1994), and “Detection of Stellate Distortions in Mammograms using Scale Space Operators,”
Information Processing in Medical Imaging
(Bizais et al., eds., Kluwer Academic Publishers 1995). Like the ALOE algorithm, the Karssemeijer approach is also a “backward direction” spiculation detection algorithm.
In the Karssemeijer algorithm, a “line image” and a “direction image” is first formed from the digital mammogram. As is known in the art, a line image contains line information for each pixel in the digital mammogram, while a direction image contains direction information for each pixel in the line image. The most basic form of line image, used in the Karssemeijer algorithm, contains line information which is a “1” if the pixel is located along a line and a “0” otherwise. The most basic form of direction image, also used in the Karssemeijer algorithm, contains direction information which, for those pixels having a “1” in the line image, equals the approximate angle of a tangent to the line passing through the pixel.
Consi
Roehrig Jimmy R.
Romsdahl Harlan M.
Zhang Wei
Patel Jayanti K.
R2 Technology, Inc.
LandOfFree
Method and apparatus for fast detection of lesions 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 apparatus for fast detection of lesions, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for fast detection of lesions will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3163597