Entropy Segmentation Research Articles

Abstract 4961: Longitudinal bioluminescence imaging of primary versus abdominal metastatic tumor growth in orthotopic pancreatic tumor models in NOD/SCIDγ(-/-) mice

Abstract Pancreatic ductal adenocarcinoma (PDAC) has a very poor prognosis and is currently the fourth leading cause of cancer death in the United States. The lethal nature of PDAC is strongly associated with metastases to distant organs. The purpose of the present study was to develop and validate noninvasive bioluminescence imaging methods for differentially monitoring the kinetics of primary and abdominal metastatic tumor growth in mouse orthotopic models of pancreatic cancer. A semiautomated maximum entropy segmentation method was implemented for the primary tumor region-of-interest (ROI), and a rule-based method for manually drawing an ROI for the abdominal metastatic region also was developed. The two ROI methods were first validated by having two observers independently construct ROIs for the tumors of animals implanted orthotopically with Panc-1 cells, and the results compared with the number of mesenteric lymph node metastatic nodules counted upon necropsy. The findings were extended to orthotopic tumors of the more metastatic MIA PaCa-2 and AsPC-1 cells where different groups of animals were implanted with different numbers of cells. When the data were expressed as the total photon flux (Ph/sec) in the ROIs for the primary tumor and metastases, the total flux within the metastasis ROI was larger in magnitude than the total flux from the primary tumor ROI, at times by as much as several orders of magnitude. However, when the data were expressed as the average flux density (Ph/sec*mm2) within the ROIs, the density of the flux within the smaller primary tumor ROI was larger in magnitude than the density of the flux from the larger metastasis ROI, by as much as several orders of magnitude. Interobserver assessments for total flux and flux density from ROIs for both the primary tumors and metastatic region were highly concordant, with correlation coefficients of r > 0.98, coefficients of variation of ≤ 0.02, and limits of agreement within <5%. Further, there were statistically significant differences in the growth kinetics of AsPC-1 and MIA PaCa-2 orthotopic tumors when mice were implanted with different numbers of cells for each cell line. The present results demonstrated that the segmentation methods were highly reliable, reproducible and robust, and allowed statistically significant discrimination in the growth kinetics of primary and abdominal metastatic tumors of different cell lines implanted with different cell numbers. Thus, primary tumors and abdominal metastatic foci in orthotopic pancreatic cancer models can be reliably monitored separately and noninvasively over time with bioluminescence imaging. The novel segmentation methods reported here will facilitate investigations of the biology of primary and metastatic tumor growth, as well as the effects of novel treatments, over time, in individual animals. Citation Format: Harlan E. Shannon, Melissa L. Fishel, Jingwu Xie, Dongsheng Gu, Brian P. McCarthy, Amanda A. Riley, Anthony L. Sinn, Jayne M. Silver, Mark R. Kelley, Helmut Hanenberg, Murray Korc, Karen E. Pollok, Paul R. Territo. Longitudinal bioluminescence imaging of primary versus abdominal metastatic tumor growth in orthotopic pancreatic tumor models in NOD/SCIDγ(-/-) mice. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 4961. doi:10.1158/1538-7445.AM2014-4961

Image Segmentations for Through-the-Wall Radar Target Detection

Detection of stationary targets using pixel-wise likelihood ratio test (LRT) detectors has been recently proposed for through-the-wall radar imaging (TWRI) applications. We employ image segmentation techniques, in lieu of LRT, for target detection in TWRI. More specifically the widely used between-class variance (BCV) thresholding, maximum entropy segmentation, and K-means clustering are considered to aid in removing the clutter, resulting in enhanced radar images with target regions only. For the case when multiple images of the same scene are available through diversity in polarization and/or vantage points around a building structure, we propose to use image fusion, following the image segmentation step, to generate an enhanced composite image. In particular, additive, multiplicative, and fuzzy logic fusion techniques are considered. The performance of the segmentation and fusion schemes is evaluated and compared with that of the assumed LRT detector using both electromagnetic (EM) modeling and real data collected in a laboratory environment. The results show that, although the principles of segmentation and detection are different, the segmentation techniques provide either comparable or improved performance over the LRT detector. Specifically, in the cases considered, the maximum entropy segmentation produces the best results for detection of targets inside building structures. For fusion of multiple segmented images of the same scene, the fuzzy logic fusion outperforms the other methods.

Multi-threshold image segmentation using maximum fuzzy entropy based on a new 2D histogram

The traditional fuzzy entropy segmentation method is based on 1D histogram which cannot reflect spatial information and is affected by noise acutely. In this paper a new definition of 2D histogram is proposed. Different images are divided into grids of different densities, and then the gray value of each pixel in the image and the average gray value of the grid which the pixel is located are used as two components of the 2D histogram. To get thresholds value which can be used to segment the image into three levels, a new multi-threshold image segmentation method using maximum fuzzy entropy based on the proposed 2D histogram is presented. The parameters of the fuzzy entropy function can be determined by genetic algorithm with an appropriate coding method. The experiment results show that by using different grids of proper density to divide different images, the proposed method obtains better performance for target information extraction.

Detecting the borders between coding and non-coding DNA regions in prokaryotes based on recursive segmentation and nucleotide doublets statistics

BackgroundDetecting the borders between coding and non-coding regions is an essential step in the genome annotation. And information entropy measures are useful for describing the signals in genome sequence. However, the accuracies of previous methods of finding borders based on entropy segmentation method still need to be improved.MethodsIn this study, we first applied a new recursive entropic segmentation method on DNA sequences to get preliminary significant cuts. A 22-symbol alphabet is used to capture the differential composition of nucleotide doublets and stop codon patterns along three phases in both DNA strands. This process requires no prior training datasets.ResultsComparing with the previous segmentation methods, the experimental results on three bacteria genomes, Rickettsia prowazekii, Borrelia burgdorferi and E.coli, show that our approach improves the accuracy for finding the borders between coding and non-coding regions in DNA sequences.ConclusionsThis paper presents a new segmentation method in prokaryotes based on Jensen-Rényi divergence with a 22-symbol alphabet. For three bacteria genomes, comparing to A12_JR method, our method raised the accuracy of finding the borders between protein coding and non-coding regions in DNA sequences.

The Combination of Local Fuzzy Entropy and Gray Level Difference-Based Thresholding Segmentation

A novel transition region extraction and thresholding method based on both frequency and degree of gray level changes is proposed by analyzing properties of transition region. Frequent gray level based transition region extraction methods are greatly affected by noise. To eliminate the algorithm limitation, a modified descriptor taking both degree and frequency of gray level changes into account is developed. The proposed algorithm can accurately extract transition region of an image and get ideal segmentation result. The experimental results show its superiority and feasibility.

Automatic classification of radar targets with micro-motions using entropy segmentation and time-frequency features

Micro-Doppler (mD) signatures have great potential in the radar micro-dynamic target classification. An automatic classification method for radar targets with micro-motions is proposed based on the idea of entropy and feature extraction from the spectrogram. In this method, the measurement of entropy is firstly conducted over the time-frequency distribution associated with the minimum filtering operation, the threshold discrimination and the region focusing to obtain the region of interest corresponding to mD signatures in the original spectrogram. It helps acquire the valid region in the time-frequency domain and reduce the computational burden in the following processing. Next, invariant moments and geometric characteristics of time-frequency distribution of mD signatures are extracted from the segmented spectrogram to construct mD feature vectors. A support vector machine (SVM) with decision-tree architecture is then used for multiclass micro-dynamic target classification from radar echoes. Finally, some experimental tests with simulated mD data are carried out to confirm the effectiveness of the proposed method and evaluate the performance under different conditions of signal-to-noise ratio (SNR), training set and feature vector. In addition, issues related to the improvement of classification performance are also discussed.

Image segmentation of maximum inter-class cross entropy based on 2D histogram oblique segmentation

With the approach based on 2D histogram oblique segmentation, a fast image thresholding method based on maximum inter-class entropy was presented in this paper. Firstly the thresholding method was formulated, based on maximum inter-class entropy and 2D histogram oblique segmentation, and then its fast recurring algorithm was deduced, finally the defined array operations and the algorithm were combined to search the optimal threshold vector. The experimental results show that the proposed method can not only get better generalization, but also has higher efficiency than the current thrsholding method based on 2D oblique segmentation.