Deep Learning and Entropy-Based Texture Features for Color Image Classification.

In the domain of computer vision, entropy-defined as a measure of irregularity-has been proposed as an effective method for analyzing the texture of images. Several studies have shown that, with specific parameter tuning, entropy-based approaches achieve high accuracy in terms of classification results for texture images, when associated with machine learning classifiers. However, few entropy measures have been extended to studying color images. Moreover, the literature is missing comparative analyses of entropy-based and modern deep learning-based classification methods for RGB color images. In order to address this matter, we first propose a new entropy-based measure for RGB images based on a multivariate approach. This multivariate approach is a bi-dimensional extension of the methods that have been successfully applied to multivariate signals (unidimensional data). Then, we compare the classification results of this new approach with those obtained from several deep learning methods. The entropy-based method for RGB image classification that we propose leads to promising results. In future studies, the measure could be extended to study other color spaces as well.

Parameter Analysis of Multiscale Two-Dimensional Fuzzy and Dispersion Entropy Measures Using Machine Learning Classification.

Two-dimensional fuzzy entropy, dispersion entropy, and their multiscale extensions (MFuzzyEn2D and MDispEn2D, respectively) have shown promising results for image classifications. However, these results rely on the selection of key parameters that may largely influence the entropy values obtained. Yet, the optimal choice for these parameters has not been studied thoroughly. We propose a study on the impact of these parameters in image classification. For this purpose, the entropy-based algorithms are applied to a variety of images from different datasets, each containing multiple image classes. Several parameter combinations are used to obtain the entropy values. These entropy values are then applied to a range of machine learning classifiers and the algorithm parameters are analyzed based on the classification results. By using specific parameters, we show that both MFuzzyEn2D and MDispEn2D approach state-of-the-art in terms of image classification for multiple image types. They lead to an average maximum accuracy of more than 95% for all the datasets tested. Moreover, MFuzzyEn2D results in a better classification performance than that extracted by MDispEn2D as a majority. Furthermore, the choice of classifier does not have a significant impact on the classification of the extracted features by both entropy algorithms. The results open new perspectives for these entropy-based measures in textural analysis.

Non-marking Collection of Amino Acids from Fingerprints Using Hydrogels.

The amino acid profile obtained from a fingerprint may provide valuable information on its donor. Unfortunately, the collection of chemicals from the fingerprint is often destructive to the fingerprint ridge detail. Herein we detail the use of cross-linkable solutions of dextran-methacrylate to form hydrogels capable of collecting amino acids from surfaces followed by extraction and quantification with UPLC-MS. This method allows for the amino acid profile analysis of fingerprints while allowing for their increased visualization at a later stage using the standard method of cyanoacrylate fuming followed by basic-yellow dyeing.

Collection of amino acids and DNA from fingerprints using hydrogels.

The amino acid profile obtained from a fingerprint may provide valuable information on its donor. For forensic scientists, recovering evidence relating to the amino acid profile of a suspect can potentially be valuable for identification and exclusion purposes. Herein we detail the use of cross-linkable solutions of dextran-methacrylate to form hydrogels capable of collecting amino acids from surfaces followed by extraction and quantification with UPLC-MS. This method allows for the amino acid profile analysis of fingerprints while allowing for their increased visualisation at a later stage using the standard method of cyanoacrylation. We will demonstrate this method to also be capable of collecting DNA from fingerprints with a 20-60% yield in comparison to using a conventional cotton swab.

Cell cycle perturbation and acquired 5-fluorouracil chemoresistance.

Acquired chemoresistance is one of the obstacles for success of 5-fluorouracil (5-FU)-based cancer chemotherapy. Some molecular mechanisms of acquired 5-FU resistance are still unknown. We have recently demonstrated down-regulation of a group of cell cycle related genes in acquired 5-FU resistant human cancer cell lines. In this study, the bivariate distribution of propidium iodide versus BrdU in acquired 5-FU resistant colon (H630R10) and breast (T47DFU2.5) cancer cell lines was compared with their parental cell lines using flow cytometric analysis. The resistant cell lines showed significantly lower labelling index (T47DFU2.5) and cell cycle delay in G1 and G1/S boundary and prolonged DNA synthesis time (H630R10). Both resistant cell lines demonstrated significantly prolonged potential doubling time (Tpot). The protein expression levels of some G1 and S phase transition-related genes were also analysed by Western blot. CDK2 protein and Thr-160 phosphorylated CDK2 were remarkably reduced in the resistant cell lines. Cyclin D3 and cyclin A were also decreased in the resistant cells. Total pRB expression was unaltered but hypophosphorylation of pRB (Ser780, Ser795 and Ser807/811) was detected in the resistant cancer cells. Our data suggest that there may be a slow down in cell cycle traverse preventing incorporation of 5-FU metabolites into DNA and also providing cancer cells with sufficient time to correct the mis-incorporated nucleotides. The cell cycle perturbation may be involved in acquired 5-FU resistance.

Signalling cell cycle arrest and cell death through the MMR System.

Loss of DNA mismatch repair (MMR) in mammalian cells, as well as having a causative role in cancer, has been linked to resistance to certain DNA damaging agents including clinically important cytotoxic chemotherapeutics. MMR-deficient cells exhibit defects in G2/M cell cycle arrest and cell killing when treated with these agents. MMR-dependent cell cycle arrest occurs, at least for low doses of alkylating agents, only after the second S-phase following DNA alkylation, suggesting that two rounds of DNA replication are required to generate a checkpoint signal. These results point to an indirect role for MMR proteins in damage signalling where aberrant processing of mismatches leads to the generation of DNA structures (single-strand gaps and/or double-strand breaks) that provoke checkpoint activation and cell killing. Significantly, recent studies have revealed that the role of MMR proteins in mismatch repair can be uncoupled from the MMR-dependent damage responses. Thus, there is a threshold of expression of MSH2 or MLH1 required for proper checkpoint and cell-death signalling, even though sub-threshold levels are sufficient for fully functional MMR repair activity. Segregation is also revealed through the identification of mutations in MLH1 or MSH2 that provide alleles functional in MMR but not in DNA damage responses and mutations in MSH6 that compromise MMR but not in apoptotic responses to DNA damaging agents. These studies suggest a direct role for MMR proteins in recognizing and signalling DNA damage responses that is independent of the MMR catalytic repair process. How MMR-dependent G2 arrest may link to cell death remains elusive and we speculate that it is perhaps the resolution of the MMR-dependent G2 cell cycle arrest following DNA damage that is important in terms of cell survival.

Public health challenges in Kyrgyzstan: developing a new curriculum.

INTRODUCTION: The public health challenges facing the Central Asian Republic of Kyrgyzstan are rooted in the social, economic and political conditions that emerged after the collapse of the Soviet Union in 1991. Geographically remote and with a substantial part of it's population living in mountainous rural villages, economic recovery and the maintenance of basic standards of public health is now a major problem for the Kyrgyz people. This project report sets out the case for restructuring public health education in Kyrgyzstan. It also explains how a new public health curriculum will equip Kyrgyz students with the knowledge and skills to work effectively with urban and rural communities in this geographically remote region of Central Asia. METHODS: With financial support from the European Union's Tempus program St Martin's College, the Kyrgyz State Medical Academy (KSMA) and Pirkanmaa Polytechnic (Finland) worked together to develop a new public health curriculum for Kyrgyzstan. Project activities took place in Kyrgyzstan and Europe throughout 2002-2004 and included English language training, fact finding visits to partner institutions and health services. Seminars and workshops were used to develop curriculum content and to support the design of new programs. A core curriculum team from KSMA, supported by European project staff, devised the new curriculum now on offer in Kyrgyzstan. RESULTS: The project achieved its three main goals: (1) development of a new public health curriculum; (2) establishment of an international forum for public health in Kyrgyzstan; and (3) dissemination of the project's findings via an international conference and the provision of web based support services. CONCLUSIONS: New courses in Preventive Medicine and Public Health Nursing at KSMA represent a significant cultural shift within public health in Central Asia, complementing the structural reforms of health care introduced in the 1990s. Emphasis on community engagement, health promotion and preventive action enhances knowledge, develops new skills and refocuses work practices for public health staff serving Kyrgyzstan's remote rural communities.

Transactivation of the cyclin A promoter by bovine papillomavirus type 4 E5 protein.

Bovine papillomavirus type 4 (BPV-4) E5 (formerly E8) is a 42-residue hydrophobic, membrane-localised protein that can transform NIH-3T3 cells by a poorly defined mechanism. In E5-expressing cells, the observed up-regulation of cyclin A is underpinned by transactivation of the cyclin A promoter. Here we show that E5 transactivates the minimal cell cycle-regulated cyclin A promoter in cells both stably and acutely expressing the viral protein. There are no detectable differences between control and E5 cells in protein complexes binding the E2F-like cell cycle-dependent element (CDE)/cell cycle-regulated element (CCRE) of the cyclin A promoter and E5 does not transactivate E2F reporter plasmids in an E2F-dependent manner in vivo. CCAAT box integrity and functional NF-Y complexes are required for E5-mediated transactivation and a Mr approximately 110 K CCAAT-box binding factor (p110 CBF) associates with NF-YA only in E5 cells. This suggests that E5 sets the extent of cyclin A promoter activation by a mechanism similar to other, structurally unrelated, DNA tumour virus oncoproteins but distinct from the action of serum factors and so is inconsistent with E5 acting through constitutive activation of tyrosine kinase growth factor receptors.

Cyclin A expression and growth in suspension can be uncoupled from p27 deregulation and extracellular signal-regulated kinase activity in cells transformed by bovine papillomavirus type 4 E5.

As the biochemical detection of bovine papillomavirus type 4 E5 is problematic, a fusion form of E5 and the green fluorescent protein (GFP-E5) was constructed and its characteristics were examined. GFP-E5 was detected in cells by autofluorescence and immunoblotting. Like wild-type (wt) E5, GFP-E5 localized in the endomembranes and permitted anchorage-independent (AI) growth. However, unlike wt E5, cells expressing GFP-E5 became quiescent in low serum and failed to sustain expression of cyclins D1 and to inactivate retinoblastoma protein (pRb). The normal anchorage requirement for cyclin D1 and cyclin A expression was abolished in cells expressing wt E5 or GFP-E5, residual extracellular signal-regulated kinase (ERK 1/2) activity was not required to sustain cyclin D1 and cyclin A expression in suspension and deregulation of cyclin A-cyclin-dependent kinase (CDK) activity was sufficient to account for AI growth of cells expressing E5. Constitutive upregulation of the CDK inhibitor p27(KIP1), characteristic of cells expressing wt E5, was not observed in those expressing GFP-E5; therefore, p27(KIP1) deregulation is not required for E5-mediated AI growth.

Mechanistic and predictive profiling of 5-Fluorouracil resistance in human cancer cells.

Gene expression was analyzed in five pairs of 5-fluorouracil (5-FU) resistant and parental cancer cell lines on DNA microarrays. In unsupervised analysis, a prediction rule was built from the expression profiles of 29 genes, and 5-FU sensitivity class was predicted with 100% accuracy and high predictive strength. In supervised analysis of key 5-FU pathways, expression of 91 genes was associated with 5-FU sensitivity phenotype and segregated samples accordingly in hierarchical analysis. Key genes involved in 5-FU activation were significantly down-regulated (thymidine kinase, 2.9-fold; orotate phosphoribosyltransferase, 2.3-fold; uridine monophosphate kinase, 3.2-fold; pyrimidine nucleoside phosphorylase 3.6-fold) in resistant cells. Overexpression of thymidylate synthase and its adjacent gene, c-Yes, was detected in the resistant cell lines. The mRNA and protein overexpression of nuclear factor kappaB (NFkappaB) p65 and related antiapoptotic c-Flip gene was detected in resistant cells. The 5-FU-resistant cell lines also showed high NFkappaB DNA-binding activity. Cotransfection of NFkappaB p50 and p65 cDNA induced 5-FU resistance in MCF-7 cells. Both NFkappaB- and 5-FU-induced resistant cell lines manifested reduced expression of genes governing G(1)-S and S-phase transition. Expression of genes involved in DNA replication was also down-regulated in resistant cell lines. These findings were highly consistent with the slower growth rate, higher proportion of G(1), and lower proportion of S-phase cells in the resistant cell lines. This phenotype may protect resistant cells from cell death induced by incorporation of 5-FU into DNA chains, by allowing time to repair 5-FU-induced damage. Our findings may provide novel targets for tackling 5-FU resistance.