Medial axis transform-based features and a neural network for human chromosome classification

Lerner, Boaz; Guterman, Hugo; Dinstein, I.; Romem, Yitzhak

doi:10.1016/0031-3203(95)00042-x

Cited by 49 publications

(34 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The first database was derived by Lerner et al (1995) from chromosome images, which were gathered at the Soroka Medical Center, Beer-Sheva, Israel. The chromosome patterns were represented by 64 density profile (d.p.)…”

Section: A the Methodologymentioning

confidence: 99%

“…The second database is also obtained from chromosome images but since it consists of geometrical features, patterns of the database are independent of patterns of the former database, which consists of intensity features. The chromosome patterns are represented by four geometrical features, i.e., the centromeric index (the ratio of the short arm length to the total length), length, perimeter and area of the chromosome (Lerner et al, 1995).…”

Section: A the Methodologymentioning

confidence: 99%

See 1 more Smart Citation

A comparative study of neural network based feature extraction paradigms

Lerner

Guterman

Aladjem

et al. 1999

Pattern Recognition Letters

View full text Add to dashboard Cite

Section: A the Methodologymentioning

confidence: 99%

Section: A the Methodologymentioning

confidence: 99%

A comparative study of neural network based feature extraction paradigms

Lerner

Guterman

Aladjem

et al. 1999

Pattern Recognition Letters

View full text Add to dashboard Cite

“…A method on the basis of approximation of class conditional densities by a mixture of parameterized densities of a special type is used for feature selection [27]. Medical axis transformation with artificial neural network (ANN [28]) is used for human chromosome classification [29]. Adaptive partitioning based discriminability measure was used as a criterion in a classifierindependent feature selection procedure [30].…”

Section: Literature Surveymentioning

confidence: 99%

A Survey on Feature Selection Algorithms

Kumar¹

2015

IJRITCC

View full text Add to dashboard Cite

Abstract-One major component of machine learning is feature analysis which comprises of mainly two processes: feature selection and fe ature extraction. Due to its applications in several areas including data mining, soft computing and big data analysis, feature selection has got a reasonable importance. This paper presents an introductory concept of feature selection with various inherent approaches. The paper surveys historic developments reported in feature selection with supervised and unsupervised methods. The recent developments with the state of the art in the on-going feature selection algorithms have also been summarized in the paper including their hybridizations.

show abstract

“…Traditionally a medial axis transform [26] is performed on the chromosomes to straighten it, and then a density profile is measured by integrating the intensities along sections perpendicular to the medial axis of the chromosomes. This could be challenging though, since many bent chromosomes are not trivial to straighten.…”

Section: Classificationmentioning

confidence: 99%

Maximum-likelihood techniques for joint segmentation-classification of multispectral chromosome images

Schwartzkopf

Bovik

Evans

2005

IEEE Trans. Med. Imaging

View full text Add to dashboard Cite

The purpose of the research described in this dissertation is to develop new methods for automatic chromosome identification. In particular, I (1) develop a maximum likelihood hypothesis test that uses this multi-spectral information, together with conventional criteria, to select the best segmentation possibility, (2) use this likelihood function to combine chromosome segmentation and classification into a robust chromosome identification system, and (3) show that the proposed likelihood function can also be used as a reliable indicator of errors in segmentation, errors in classification, and the chromosomes anomalies that can be diagnosed with M-FISH imaging. I show that the proposed multispectral joint segmentation-classification method outperforms past grayscale segmentation methods in decomposing touching chromosomes. Furthermore, I show that it outperforms past M-FISH classification techniques that do not use segmentation information. OBJECTIVESThis work investigated what improvements in chromosome analysis could be obtained by using M-FISH multi-spectral images. One aim of this work was to develop algorithms that could take full advantage of the multi-spectral information and to quantify their improvement over grayscale chromosome image analysis methods. In order to achieve effective segmentation and classification, I studied how to combine segmentation and classification to make both more accurate. Furthermore, this work examined how to use this multi-spectral information to detect automatically abnormalities in the chromosomes that could not be detected without such chromosome labeling. CONTRIBUTIONSThe best approach for segmentation and classification of multi-spectral chromosome images is fundamentally different from the best approach for grayscale chromosome images. The additional information provided by multispectral chromosome images is useful for distinguishing touching and 2. This maximum likelihood test is used to propose a method that combines the task of locating and classifying chromosomes for improved performance in both tasks.3. The first two contributions in the dissertation are then used to achieve aberration scoring; that is, giving a score to each segment to indicate the likelihood of abnormalities in that image. 4 NOTATIONThis section gives the mathematical notation that is followed throughout the dissertation. I denote vectors with boldface and scalars with plain font. I use the function ( ) ⋅ p to denote a probability. Further, the functiondenotes the usual Gaussian probability density function. Thus the probability density function of a random vector x , with mean µ and covariance matrix Σ , is ( ) ORGANIZATIONThe structure of this dissertation is as follows. Chapter 2 introduces the features of chromosomes and chromosomes images and shows how these features CHROMOSOMESChromosomes are the body's information carriers. They are the structures that contain genes, which store in strings of DNA all of the data necessary for an organism's development and maintenance. Chromosome...

show abstract

Medial axis transform-based features and a neural network for human chromosome classification

Cited by 49 publications

References 9 publications

A comparative study of neural network based feature extraction paradigms

A comparative study of neural network based feature extraction paradigms

A Survey on Feature Selection Algorithms

Maximum-likelihood techniques for joint segmentation-classification of multispectral chromosome images

Contact Info

Product

Resources

About