Characterization and classification of tumor lesions using computerized fractal-based texture analysis and support vector machines in digital mammograms

Objective: This paper presents a detailed study of fractal-based methods for texture characterization of mammographic mass lesions and architectural distortion. The purpose of this study is to explore the use of fractal and lacunarity analysis for the characterization and classification of both tumor lesions and normal breast parenchyma in mammography. Materials and methods: We conducted comparative evaluations of five popular fractal dimension estimation methods for the characterization of the texture of mass lesions and architectural distortion. We applied the concept of lacunarity to the description of the spatial distribution of the pixel intensities in mammographic images. These methods were tested with a set of 57 breast masses and 60 normal breast parenchyma (dataset1), and with another set of 19 architectural distortions and 41 normal breast parenchyma (dataset2). Support vector machines (SVM) were used as a pattern classification method for tumor classification. Results: Experimental results showed that the fractal dimension of region of interest (ROIs) depicting mass lesions and architectural distortion was statistically significantly lower than that of normal breast parenchyma for all five methods. Receiver operating characteristic (ROC) analysis showed that fractional Brownian motion (FBM) method generated the highest area under ROC curve (A z = 0.839 for dataset1, 0.828 for dataset2, respectively) among five methods for both datasets. Lacunarity analysis showed that the ROIs depicting mass lesions and architectural distortion had higher lacunarities than those of ROIs depicting normal breast parenchyma. The combination of FBM fractal dimension and lacunarity yielded the highest A z value (0.903 and 0.875, respectively) than those based on single feature alone for both given datasets. The application of the SVM improved the performance of the fractal-based features in differentiating tumor lesions from normal breast parenchyma by generating higher A z value. Conclusion: FBM texture model is the most appropriate model for characterizing mammographic images due to self-affinity assumption of the method being a better approximation. Lacunarity is an effective counterpart measure of the fractal dimension in texture feature extraction in mammographic images. The classification results obtained in this work suggest that the SVM is an effective method with great potential for classification in mammographic image analysis.

Differences in glucose-dependent insulinotropic polypeptide hormone and hepatic lipase in subjects of southern and northern Europe: implications for postprandial lipaemia

The aim was to determine in 32 healthy young men from northern and southern Europe whether differences in the secretion of insulin and glucose-dependent insulinotropic polypeptide (GIP) might explain these findings through the actions of these hormones on lipoprotein lipase. In a randomized, single-blind, crossover study the effects of 2 test meals of identical macronutrient composition but different saturated fatty acid (SFA) and monounsaturated fatty acid (MUFA) contents were investigated on postprandial GIP, insulin, the ratio of incremental triacylglycerol to apolipoprotein B-48 (a marker of chylomicron size), and the activity of postheparin lipases. Fasting and postprandial GIP concentrations and postheparin hepatic lipase (HL) activities were higher in the southern Europeans (P<0.001 and P<0.02, respectively). Lipoprotein lipase activity after the SFA-rich meal was higher in the northern Europeans (P<0.01). HL activity 9 h after the SFA-rich meal and the area under the curve (AUC) for the postprandial insulin response correlated with the AUC for the postprandial GIP response (r=0.44 (P<0.04) and r=0.46 (P<0.05), respectively). There were no significant differences in chylomicron size between the 2 groups for either meal, but when the groups were combined there was a difference in chylomicron size between the SFA- and MUFA-rich meals (P<0.05), which could be due to the formation of larger chylomicrons after the MUFA-rich meal. The significantly higher GIP and insulin responses and HL activities in southern Europeans may provide an explanation for a previous report of attenuated postprandial triacylglycerol and apolipoprotein B-48 responses in them.

Effects of dietary fatty acid composition on basal and hormonestimulated hepatic lipogenesis and on circulating lipids in the rat

Thirty male rats were randomly assigned to one of three dietary groups in which the source of dietary fat was either a mixed oil, maize oil or fish oil. Effects of dietary fatty acid composition on in virro rates of [U-'4C]glucose incorporation into hepatic total lipids and into hepatic triacylglycerol were measured under basal, insulin (4 nM)-, gastric inhibitory polypeptide (GIP; 6 mi)- and insulin + GIP (4 nM + 6 n ~ ) - stimulated conditions. Effects of the three diets on postprandial plasma triacylglycerol, cholesterol, insulin and GIP concentrations were also measured. The fish-oil diet decreased rates of basal glucose incorporation into hepatic total lipids (P < 0.05) and hepatic triacylglycerol (P < 0.01) compared with the mixed-oil diet. The presence of insulin + GIP in the incubation medium stimulated glucose incorporation into hepatic total lipids in the maize-oil (P < 0.01) and fish-oil groups (P < OW), as well as into hepatic triacylglycerol in the maize-oil group (P < 0.005). In addition, the fish-oil diet decreased postprandial plasma triacylglycerol levels compared with both other dietary groups (P < 0-05 both cases), and the mixed-oil diet markedly increased postprandial plasma insulin levels compared with the other dietary groups (P c 0.001).

Assessing hepatic metabolic changes during progressive colonization of germ-free mouse by 1H NMR spectroscopy

It is well known that gut bacteria contribute significantly to the host homeostasis, providing a range of benefits such as immune protection and vitamin synthesis. They also supply the host with a considerable amount of nutrients, making this ecosystem an essential metabolic organ. In the context of increasing evidence of the link between the gut flora and the metabolic syndrome, understanding the metabolic interaction between the host and its gut microbiota is becoming an important challenge of modern biology.1-4 Colonization (also referred to as normalization process) designates the establishment of micro-organisms in a former germ-free animal. While it is a natural process occurring at birth, it is also used in adult germ-free animals to control the gut floral ecosystem and further determine its impact on the host metabolism. A common procedure to control the colonization process is to use the gavage method with a single or a mixture of micro-organisms. This method results in a very quick colonization and presents the disadvantage of being extremely stressful5. It is therefore useful to minimize the stress and to obtain a slower colonization process to observe gradually the impact of bacterial establishment on the host metabolism. In this manuscript, we describe a procedure to assess the modification of hepatic metabolism during a gradual colonization process using a non-destructive metabolic profiling technique. We propose to monitor gut microbial colonization by assessing the gut microbial metabolic activity reflected by the urinary excretion of microbial co-metabolites by 1H NMR-based metabolic profiling. This allows an appreciation of the stability of gut microbial activity beyond the stable establishment of the gut microbial ecosystem usually assessed by monitoring fecal bacteria by DGGE (denaturing gradient gel electrophoresis).6 The colonization takes place in a conventional open environment and is initiated by a dirty litter soiled by conventional animals, which will serve as controls. Rodents being coprophagous animals, this ensures a homogenous colonization as previously described.7 Hepatic metabolic profiling is measured directly from an intact liver biopsy using 1H High Resolution Magic Angle Spinning NMR spectroscopy. This semi-quantitative technique offers a quick way to assess, without damaging the cell structure, the major metabolites such as triglycerides, glucose and glycogen in order to further estimate the complex interaction between the colonization process and the hepatic metabolism7-10. This method can also be applied to any tissue biopsy11,12.

Identification of metabolites in human hepatic bile using 800 MHz 1H NMR spectroscopy, HPLC-NMR/MS and UPLC-MS

The first application of high field NMR spectroscopy (800 MHz for 1H observation) to human hepatic bile (as opposed to gall bladder bile) is reported. The bile sample used for detailed investigation was from a donor liver with mild fat infiltration, collected during organ retrieval prior to transplantation. In addition, to focus on the detection of bile acids in particular, a bile extract was analysed by 800 MHz 1H NMR spectroscopy, HPLC-NMR/MS and UPLC-MS. In the whole bile sample, 40 compounds have been assigned with the aid of two-dimensional 1H–1H TOCSY and 1H–13C HSQC spectra. These include phosphatidylcholine, 14 amino acids, 10 organic acids, 4 carbohydrates and polyols (glucose, glucuronate, glycerol and myo-inositol), choline, phosphocholine, betaine, trimethylamine-N-oxide and other small molecules. An initial NMR-based assessment of the concentration range of some key metabolites has been made. Some observed chemical shifts differ from expected database values, probably due to a difference in bulk diamagnetic susceptibility. The NMR spectra of the whole extract gave identification of the major bile acids (cholic, deoxycholic and chenodeoxycholic), but the glycine and taurine conjugates of a given bile acid could not be distinguished. However, this was achieved by HPLC-NMR/MS, which enabled the separation and identification of ten conjugated bile acids with relative abundances varying from approximately 0.1% (taurolithocholic acid) to 34.0% (glycocholic acid), of which, only the five most abundant acids could be detected by NMR, including the isomers glycodeoxycholic acid and glycochenodeoxycholic acid, which are difficult to distinguish by conventional LC-MS analysis. In a separate experiment, the use of UPLC-MS allowed the detection and identification of 13 bile acids. This work has shown the complementary potential of NMR spectroscopy, MS and hyphenated NMR/MS for elucidating the complex metabolic profile of human hepatic bile. This will be useful baseline information in ongoing studies of liver excretory function and organ transplantation.

Attaching and effacing lesions in the large intestine of an eight-month-old heifer associated with Escherichia coli O26 infection in a group of animals with dysentery

Escherichia coli O26:K60, with genetic attributes consistent with a potentially human enterohaemorrhagic E coli was isolated from the faeces of an eight-month-old heifer with dysentery. Attaching and effacing lesions were identified in the colon of a similarly affected heifer examined postmortem, and shown to be associated with E coli O26 by specific immunolabelling.

Attaching and effacing lesions caused by Escherichia coli O157:H7 in experimentally inoculated neonatal lambs

Four 6-day-old conventionally reared lambs were inoculated orally with a total of 10(9) cfu comprising equal numbers of four enterohaemorrhagic Escherichia coli (EHEC) O157:H7 strains. All animals remained clinically normal. Tissues were sampled under terminal anaesthesia at 12, 36, 60 and 84 h post inoculation (hpi). EHEC O157:H7 was cultured from most gastrointestinal tract sites. Small, sparse attaching and effacing (AE) lesions were found in the caecum at 12 and 36 hpi and in the terminal colon and rectum at 84 hpi. Organisms in the lesions were labelled specifically by an O157 antiserum. The results indicate that the well-characterised mechanisms for intimate attachment encoded by the locus for enterocyte effacement (LEE) of EHEC O157:H7 may contribute to the initial events. at least, of colonisation of sheep.

Production of attaching-effacing lesions in ligated large intestine loops of 6-month-old sheep by Escherichia coli O157:H7

Shiga-toxigenic Escherichia coli O157:H7 (STEC O157:H7) is associated with potentially fatal human disease, and a persistent reservoir of the organism is present in some farm animal species, especially cattle and sheep. The mechanisms of persistent colonisation of the ruminant intestine by STEC O157:H7 are poorly understood but may be associated with intimate adherence to eukaryotic cells. Intimate adherence, as evidenced by induction of attaching-effacing (AE) lesions by STEC O157, has been observed in 6-day-old conventional lambs after deliberate oral infection but not in older animals. Thus, the present study used a ligated intestinal loop technique to investigate whether STEC O157:H7 and other attaching-effacing E. coli may adhere intimately to the sheep large intestinal mucosa. To do this, four STEC O157:H7 strains, one STEC 026:K60:H11 and one Shiga toxin-negative E. coli O157:H7 strain, suspended in either phosphate-buffered saline or Dulbecco's modified Eagle's medium, were inoculated into ligated spiral colon loops of each of two lambs. The loops were removed 6 h after inoculation, fixed and examined by light and electron microscopy. AE lesions on the intestinal mucosa were produced by all the inoculated strains. However, the lesions were sparse and small, typically comprising bacterial cells intimately adhered to a single enterocyte, or a few adjacent enterocytes. There was little correlation between the extent of intimate adherence in this model and the bacterial cell density, pre-inoculation growth conditions of the bacteria or the strain tested.