Assessing beef carcass tissue weights using computed tomography spirals of primal cuts.

More than 800 beef primal cuts from 44 Aberdeen Angus and Limousin-cross steers carcasses were scanned using spiral computed tomography (CT) and dissected. Thresholds for the segmentation of fat, muscle and bone in the CT spirals were estimated with the objective of assessing the weight of these tissues in the primal cuts and in the entire carcasses. Thresholds were estimated using half of the dataset (DBE) and then validated in the other half (DBV). Automatic image analysis procedures were used to assess tissue weights. The R(2) of the regression between primal tissue weight by dissection and CT were high in both datasets for fat (DBE, 0.89; DBV, 0.92), muscle (DBE, 0.99; DBV, 0.99) and bone (DBE, 0.95; DBV, 0.97). The estimation of total carcass tissue weights were also very accurate for the three tissues (R(2) values of 0.95 to 0.96), indicating that CT scanning may deliver very accurate information on beef carcass composition faster and with lower cost than physical dissection and without damaging or depreciating the primal joints.

Efficiency of functional regression estimators for combining multiple laser scans of cDNA microarrays.

The first stage in the analysis of cDNA microarray data is estimation of the level of expression of each gene, from laser scans of hybridised microarrays. Typically, data are used from a single scan, although, if multiple scans are available, there is the opportunity to reduce sampling error by using all of them. Combining multiple laser scans can be formulated as multivariate functional regression through the origin. Maximum likelihood estimation fails, but many alternative estimators exist, one of which is to maximise the likelihood of a Gaussian structural regression model. We show by simulation that, surprisingly, this estimator is efficient for our problem, even though the distribution of gene expression values is far from Gaussian. Further, it performs well if errors have a heavier tailed distribution or the model includes intercept terms, but not necessarily in other regions of parameter space. Finally, we show that by combining multiple laser scans we increase the power to detect differential expression of genes.

Accuracy of in vivo muscularity indices measured by computed tomography and their association with carcass quality in lambs.

Assessments of muscle mass and skeletal dimensions by Computed Tomography (CT) enable the development of new muscularity indices for the hind leg (HL) and lumbar region (LR) in lambs. Compared to previous CT muscularity indices, the accuracy was much higher with the new index in the HL (correlations between CT and dissection indices of 0.89 vs 0.51). The accurate measurement of femur length by CT used in the calculation of the new HL index made an important contribution to the higher accuracy of the index. The improvement in accuracy was smaller for the LR (0.55 vs 0.44). The association of CT muscularity indices and carcass quality in Texel and Scottish Blackface lambs showed that improved muscularity is not phenotypically correlated with detrimental effects on carcass composition. CT muscularity indices provide an alternative method to improve carcass conformation and leanness, using measurements that at a constant weight are independent of fatness.

Combinatorial image analysis of DNA microarray features.

MOTIVATION: DNA and protein microarrays have become an established leading-edge technology for large-scale analysis of gene and protein content and activity. Contact-printed microarrays has emerged as a relatively simple and cost effective method of choice but its reliability is especially susceptible to quality of pixel information obtained from digital scans of spotted features in the microarray image. RESULTS: We address the statistical computation requirements for optimizing data acquisition and processing of digital scans. We consider the use of median filters to reduce noise levels in images and top-hat filters to correct for trends in background values. We also consider, as alternative estimators of spot intensity, discs of fixed radius, proportions of histograms and k-means clustering, either with or without a square-root intensity transformation and background subtraction. We identify, using combinatoric procedures, optimal filter and estimator parameters, in achieving consistency among the replicates of a gene on each microarray. Our results, using test data from microarrays of HCMV, indicate that a highly effective approach for improving reliability and quality of microarray data is to apply a 21 by 21 top-hat filter, then estimate spot intensity as the mean of the largest 20% of pixel values in the target region, after a square-root transformation, and corrected for background, by subtracting the mean of the smallest 70% of pixel values. AVAILABILITY: Fortran90 subroutines implementing these methods are available from the authors, or at http://www.bioss.ac.uk/~chris.

Estimators of tissue proportions from X-ray CT images.

Estimators are derived of tissue proportions from X-ray computed tomography (CT) images. These take into account that many pixels in such images are responses to mixtures of tissue types. The problem is motivated by an application involving estimation of sheep tissue weights. The standard estimator, a count of the number of pixels in a particular range of values, is compared with the maximum likelihood fit of a mixed-pixel distribution and a moment-based estimator. Both simulations and the application show the moment estimator to be best.

Towards automatic cell identification in DIC microscopy.

A general method is proposed for constructing templates of cells in differential interference contrast (DIC) microscopy. This takes account of the optics which generate DIC images, and is applicable to both transparent and semi-transparent cells of simple and complex shapes. Then, a template matching methodology is presented, which uses fast Fourier transforms to fit templates of a range of sizes and orientations to images. For illustration, this is used to automatically identify and measure individual Candida yeast cells in clusters.

Uses of digital image analysis in electrophoresis.

The information in electrophoretic gels can be extracted visually, or by using specialised computer hardware and software which treat the gels as digital images. We discuss issues which arise in applying digital image analysis to electrophoresis, and comment on the range of computer packages available. We also illustrate how image analysis can make gel interpretation easier and more reliable. Methods covered include gel registration and warping, mathematical morphology and deconvolution to sharpen images.

An algorithm for unwarping multitrack electrophoretic gels.

An algorithm is proposed for removing the effect of "frowns" and "smiles" (i.e. warping) from multitrack electrophoretic gels, so that the bands are horizontal and in alignment between tracks. It has been applied to a DNA sequencing autoradiograph, but could equally be applied to other types of gel separation where warping occurs. The algorithm uses a gradient filter to estimate band orientations. These are smoothed and interpolated by robust regression, and integrated to find the unwarping transform. No manual intervention is required.

Validation of the in vivo measurement of adipose tissue by magnetic resonance imaging of lean and obese pigs.

In vivo quantification of adipose tissue with magnetic resonance imaging (MRI) was validated with pigs. Thirteen transaxial MRI sections were collected, at intervals proportional to body length, from each pig, which was then killed, frozen, and sliced at the locations of the MRI sections. Adipose-tissue quantities were determined by dissecting each slice, and lipid contents of the dissected slices and of the tissue segments between slices were measured. Compared with dissection, MRI underestimated abdominal percent adipose tissue and overestimated cervical percent adipose tissue by less than 6%. When all 13 sections were used, MRI closely predicted percent lipid and dissected percent adipose tissue with small residual SDs (RSD = 1.9 and 2.1, respectively), which increased only slightly if two sections (4, upper thorax and 8, upper abdomen) were used (RSD = 2.3 and 2.6, respectively). In conclusion MRI accurately quantifies adipose tissue in vivo, matching values produced by dissection and chemical analysis.

Total and subcutaneous adipose tissue in women: the measurement of distribution and accurate prediction of quantity by using magnetic resonance imaging.

Total and subcutaneous adipose tissue in seven lean and seven obese women were quantified using magnetic resonance imaging (MRI). The distributions of adipose tissues along the body were closely correlated: subcutaneous with total, both within and between lean and obese groups. Lean women had proportionally less adipose tissue in the lower thorax and upper abdomen than did obese women. Reducing the number of MRI scans from 17 to 4 did not increase the residual SD of predicted body adipose tissue (2.9 percent) when body density was used as the reference measure. MRI gave an estimate of total-body adipose tissue significantly closer to the value for fat percent produced when the results from five other techniques (skinfold thickness, underwater weighing, 40K whole-body counting, isotopic water dilution, and tetrapolar bioelectrical impedance) were averaged than when any other technique was used alone. MRI-determined percent body adipose tissue in women is close to, and proportional to, estimates derived by underwater weighing.

The distribution of numbers of open channels in multi-channel patches.

Assumptions inherent in use of the binomial distribution are relaxed in order to identify alternative distributions with different variances, for fitting to the frequency distributions of numbers of open channels in multi-channel patches. The variance is less than that for the binomial distribution with the same expectation when either there is a negative interaction between channels or channels have different probabilities of being open. When the variance is greater, this may be because of positive interaction between channels or non-stationary behaviour. Use of the distributions and lack-of-fit statistics is illustrated.

Exploratory and confirmatory plots of single-channel records.

Graphical displays are used to study open- and closed-time durations of a single ion-channel. CUSUM plots, in conjunction with Kolmogorov-Smirnov type tests, provide checks for changes in channel behaviour. Distributional properties are examined by means of Q-Q and P-P plots. The methods are illustrated using GABA-activated channel records from Ascaris suum.

Heat balance of ox steers (Bos taurus) in steady-temperature environments.

Work was undertaken to investigate body heat storage in cattle subjected to steady temperature environments. Heat production (M) of and heat losses (H) from two steers were studied by simultaneous direct and indirect calorimetry over periods of up to 11 days. M and H followed similar trends except after feeding and watering. When the animals stood both M and H leveled approximately 20% higher than when they were lying down. Over 24-h periods total M and H agreed within the limits of experimental error. Heat storage of the calorimeter contents (MH) was calculated from the accumulated difference between M and H. Day-night cycles observed in MH were partly due to the heat of warming food and water and of cooling excreta (FF). The difference (TB = MH - FF) represented calorimetrically determined changes in heat storage in the animal's body. Day-night cycles in TB were small (less than 1 degree C range in equivalent change in mean body temperature) but agreed well with observed changes in body core (carotid artery) temperatures. It is concluded that day-night changes in heat storage in fixed-temperature environments are small but measurable by calorimetry and that the methods used are applicable for future studies with cyclic environmental temperatures.