In the present paper we discuss a new approach for the detection of microcalcification clusters, based on neural networks and developed as part of the MAGIC-5 project, an INFN-funded program which aims at the development and implementation of CAD algorithms in a GRID-based distributed environment. The proposed approach has as its roots the desire to maximize the rejection of background during the analytical pre-processing stage, in order to train and test the neural network with as clean as possible a sample and therefore maximize its performance. The algorithm is composed of three modules: the image pre-processing, the feature extraction component and the Backpropagation Neural Network module. The First module comprises the use of several algorithms: H-Dome Transformation, Masking, Binarisation of grayscale images, Connected Components Labeling; for the classification, initially 27 features are extracted from the output image, features that are statistically analyzed and reduced to 17, which are used as input to the Backpropagation Neural Network. The algorithm was trained (tested) on 139 (139) images respectively, containing 149 (152) true clusters and 146 (415) false
S C CHERAN, R CATALDO, P CERELLO, F DE CARLO, FAUCI F, G FORNI, et al. (2004). Detection and classification of microcalcifications clusters in digitized mammograms. In IEEE Nuclear Science Symposium Conference Record (pp.4136-4140).
Detection and classification of microcalcifications clusters in digitized mammograms
CATALDO, Renato;FAUCI, Francesco;RASO, Giuseppe;
2004-01-01
Abstract
In the present paper we discuss a new approach for the detection of microcalcification clusters, based on neural networks and developed as part of the MAGIC-5 project, an INFN-funded program which aims at the development and implementation of CAD algorithms in a GRID-based distributed environment. The proposed approach has as its roots the desire to maximize the rejection of background during the analytical pre-processing stage, in order to train and test the neural network with as clean as possible a sample and therefore maximize its performance. The algorithm is composed of three modules: the image pre-processing, the feature extraction component and the Backpropagation Neural Network module. The First module comprises the use of several algorithms: H-Dome Transformation, Masking, Binarisation of grayscale images, Connected Components Labeling; for the classification, initially 27 features are extracted from the output image, features that are statistically analyzed and reduced to 17, which are used as input to the Backpropagation Neural Network. The algorithm was trained (tested) on 139 (139) images respectively, containing 149 (152) true clusters and 146 (415) falseFile | Dimensione | Formato | |
---|---|---|---|
10.1109_NSSMIC.2004.1466803.pdf
Solo gestori archvio
Dimensione
2.73 MB
Formato
Adobe PDF
|
2.73 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.