Measuring tongue volumes and visualizing the chewing and swallowing process using real-time TrueFISP imaging--initial clinical experience in healthy volunteers and patients with acromegaly.

This study assessed both two-dimensional (2D) TrueFISP imaging for quantifying tongue volume and real-time TrueFISP imaging for evaluating chewing and swallowing in healthy volunteers and patients with acromegaly. In 50 healthy volunteers, tongue volumes were measured using a 2D TrueFISP sequence. Chewing and swallowing were visualized using a real-time TrueFISP sequence. Ten patients with acromegaly were examined twice with the same magnetic resonance imaging protocol: once prior to therapy and a second time 6 months after therapy. Prior to therapy, healthy volunteers had an average tongue volume of 140 ml for men and 90 ml for women, and patients with acromegaly had an average tongue volume of 180 ml for men and 145 ml for women. However, 6 months after therapy the mean tongue volumes in patients with acromegaly had decreased to 154 ml in the men and to 125 ml in the women. The chewing and swallowing process was normal in all volunteers. Prior to therapy, just two patients showed a chewing and swallowing pathology, which disappeared after therapy. Patients with acromegaly had larger tongue volumes than healthy volunteers, and TrueFISP imaging proved feasible for visualizing chewing and swallowing in real time and is capable of detecting possible pathologies. Furthermore, TrueFISP imaging can be used to monitor therapeutic approaches in patients with acromegaly.

[The logistic regression model including interactions between the factor variables demonstrated for the detection of E. coli O157.H7 in artificially contaminated minced beef]. / Die logistische Regression mit Wechselwirkungen zwischen den Einflussfaktoren, angewandt am Beispiel des Nachweises von E. coli O157.H7 aus künstlich kontaminiertem Rinderhackfleisch.

Logistic regression is a powerful tool to analyse data sets with a dichotomous response variable. However, in most situations it is used as a model without interactions between the factor variables. This is done either by presumption or to avoid difficulties in the interpretation of the statistical results. In this article first the model of simple logistic regression without interactions is introduced followed by the expanded model with pairwise interactions between the factors. The application of both models is demonstrated at the present data set concerning the detection of E. coli O157.H7 in artificially contaminated minced beef. The influencing variables are the factors enrichment time, inoculation density, enrichment broth, subculturing medium, and state of samples (fresh vs. deep frozen). The statistical reanalysis displayed strongly differing results emphasizing the importance of interactions in logistic regression models. In particular, the odds ratio for E. coli detection dependant from the enrichment time (24 h vs. 6 h) (OR = 0.41) was strongly overestimated without simultaneous attention of the E. coli inoculation density (OR approximately equal to 0.2 to 0.02). In this context the possible interpretation of the interaction is discussed.