Detection of spicules in mammograms

Abstract

The objective of this paper is to propose a method to automatically detect spicule shadows in mammograms. The method is consisted of two steps, enhancement and feature selection. First, spicule shadows are enhanced by using a newly developed operation. An opening operation is applied to remove noises and a direction map is made for feature selection. Second, a concentration expression is given with gray levels and two features are selected for recognition of tumors with spicules. In the method, the direction of spicules is not only considered, but the density is also utilized for classification of tumors with spicules. The method was tested on 24 samples including seven tumors with spicules. The recognition rate for tumors with spicules was 100% without the false positives.

Application of an edge detection method to satellite images for distinguishing sea surface temperature fronts near the Japanese coast

“…Especially, we take notice of a method called gray scale partial reconstruction (PR) proposed by Jiang et al (1997Jiang et al ( , 1998. It is a method of ridge-line extraction using mathematical morphology to provide for image filtering and segmentation tasks.…”

“…Namely the method effectively enhances and detects ridge components (fronts here), i.e., ranges of pixels with higher values independently of the pixel value variation along the ridge line. PR is expressed as below (Jiang et al 1997):…”

<title>Recognition of lung nodules from x-ray CT images considering 3D structure of objects and uncertainty of recognition</title>

“…The probability is represented as follows:: p(o(w, a";)) = PB p(a, b, ra, rb) = PB • g(l;p4', o')g(ra; i4 o')g(rb; o-,'). (9) where 1, j4' and o' represent the length, the mean and the standard deviation, respectively. ra, rb, /t and a!…”

Eigendetection of masses considering false positive reduction and breast density information

“…On the other hand, unsupervised segmentation consists of partitioning the image into a set of regions which are distinct and uniform with respect to specific properties, such as gray level, texture or color. Supervised approaches to mass detection are usually based on a template matching scheme, 17,18 the extraction and classification of a set of features, 19,20 the detection of spicules radiating from a central mass, [21][22][23] the creation of a statistical model of the mass, 23,24 or more recently the use of neural networks. 25,26 On the other hand, unsupervised approaches are based on an initial rough segmentation to detect regions of interest ͑ROIs͒, 27,28 which are regions likely to contain a mass, and subsequently, another algorithm, typically snakes, is used to refine the initial boundary.…”

“…The most common feature is gray-level intensity, based on the fact that mass regions tend to have brighter and more homogeneous intensity compared to their surrounding tissue. 17,20,28,34 Gradient information is another commonly used feature, which is used for finding spicules, [21][22][23] or to refine the boundary of the mass, in which case, it is usually combined with graylevel information. [29][30][31] Although some authors have used texture information in the detection step, [35][36][37][38] this information is more commonly used for mass classification purposes ͑in false positive reduction algorithms or to determine if the mass is malignant or benign͒.…”