Methods for conducting dietary assessment in the United States date back to the early twentieth century. Methods of assessment encompassed dietary records, written and spoken dietary recalls, FFQ using pencil and paper and more recently computer and internet applications. Emerging innovations involve camera and mobile telephone technology to capture food and meal images. This paper describes six projects sponsored by the United States National Institutes of Health that use digital methods to improve food records and two mobile phone applications using crowdsourcing. The techniques under development show promise for improving accuracy of food records.

Quantification of immunostaining is a widely used technique in pathology. Nonetheless, techniques that rely on human vision are prone to inter- and intraobserver variability, and they are tedious and time consuming. Digital image analysis (DIA), now available in a variety of platforms, improves quantification performance: however, the stability of these different DIA systems is largely unknown. Here, we describe a method to measure the reproducibility of DIA systems. In addition, we describe a new image-processing strategy for quantitative evaluation of immunostained tissue sections using DAB/hematoxylin-stained slides. This approach is based on image subtraction, using a blue low pass filter in the optical train, followed by digital contrast and brightness enhancement. Results showed that our DIA system yields stable counts, and that this method can be used to evaluate the performance of DIA systems. The new image-processing approach creates an image that aids both human visual observation and DIA systems in assessing immunostained slides, delivers a quantitative performance similar to that of bright field imaging, gives thresholds with smaller ranges, and allows the segmentation of strongly immunostained areas, all resulting in a higher probability of representing specific staining. We believe that our approach offers important advantages to immunostaining quantification in pathology.

Atherosclerotic lesions are heterogeneous in terms of their cellular and lipid composition. While heterogeneity can be the result of stochastic noise, an alternate hypothesis is that the differences observed among individual lesions arise from deterministic chaos. Five New Zealand white rabbits were fed a diet containing 0.15% cholesterol for 6 months. Segments of the aorta were fixed in formalin, stained en bloc with Nile red (NR) and filipin (F), and en face fluorescence microscopy was used to map the distribution of lipids in fatty streaks (FS). The smallest lesions detected stained only with filipin. Larger lesions stained with both Nile red and filipin and two distinct regions of Nile red staining, NR-orange (rich in polar lipids) and NR-yellow (rich in neutral lipids) were observed. Digital overlays revealed a “nested” arrangement of F, NR-orange, and NR-yellow. The lesions also showed marked heterogeneity in their lipid composition. Thus, although initially similar, as FS increased in size, their composition became divergent, suggesting that the ultimate composition of a FS was highly sensitive to its initial composition. Sensitivity to initial condition is one of the hallmarks of deterministic systems. To determine if FS were self-similar, another hallmark of deterministic chaos, the borders of the different regions defined by NR and F staining were subjected to fractal analysis. For each lesion, the borders of the F, NR-orange, and NR-yellow regions were found to be fractal. Return maps were constructed for the differently stained regions. Analysis of the entire 104-lesion data set showed that although the data could be described by a four-parameter logistic model, the population was not chaotic. However, return maps drawn for the maxima of the NR-orange stained regions demonstrated chaos. Taken together, the data suggest that deterministic chaos plays a role in the evolution of atheromatous disease but, in common with most biologic systems, as the lesions progress, chaotic behavior is dampened.