PubMed Health⌕ Search

Biomedical subjects

I Dinstein

Publications and source records attributed to I Dinstein.

2 recordsLinked to original sources
PubMed2000

RHIZOSCAN: A semiautomatic image processing system for characterization of the morphology and secondary metabolite concentration in hairy root cultures.

We describe the application of the newly developed RHIZOSCAN software for accurate morphological analysis and determination of overall and local secondary metabolite concentrations in two clones of Beta vulgaris throughout the growth period. Local secondary metabolite concentrations may be determined at any point in the root, and pigment gradients in each lateral root can be followed during culture and saved to the computer. Throughout the entire analysis process, an image appears in a graphical result window on the screen, which enables visual evaluation of the numerical output at each stage of the analysis. Biosynthetic data on concentrations obtained by image analysis were validated by spectrophotometric analysis. The importance of determining appropriate scanning and analysis conditions (scanning resolution, background color, threshold value, segmentation plane in the hue-saturation-intensity color system and pruning length) for obtaining accurate morphological measurements is examined and the means of fixing these parameters is described. Our results show that, using RHIZOSCAN, detailed and accurate information on root architecture and secondary metabolite concentrations can be obtained in a short time.

Automation↗
PubMed1995

Human chromosome classification using multilayer perceptron neural network.

A multilayer perceptron (MLP) neural network (NN) has been studied for human chromosome classification. Only 10-20 examples were required for the MLP NN to reach its ultimate performance classifying chromosomes of 5 types. The empirical dependence of the entropic error on the number of examples was found to be highly comparable to the 1/t function. The principal component analysis (PCA) was used, both for network initialization and for feature reduction purposes. The PCA demonstrated the importance of retaining most of the image information whenever small training sets are used. The MLP NN classifier outperformed the Bayes piecewise classifier for all the cases tested. The MLP classifier was found to be almost unsusceptible to the ratio of the number of training vectors to the number of features, whereas the piecewise classifier was highly dependent on this ratio.

Algorithms↗