Comparison of CT and chemical-shift MRI for differentiating thymoma from non-thymomatous conditions in myasthenia gravis: value of qualitative and quantitative assessment.

AIM: To evaluate the usefulness of computed tomography (CT) and chemical-shift magnetic resonance imaging (MRI) in patients with myasthenia gravis (MG) for differentiating thymoma (THY) from thymic lymphoid hyperplasia (TLH) and normal thymus (NT), and to determine which technique is more accurate. MATERIALS AND METHODS: Eighty-three patients with generalised MG who underwent surgery were divided into the TLH/NT group (A; 65 patients) and THY group (B; 24 patients). Differences in qualitative characteristics and quantitative data (CT: radiodensity in Hounsfield units; MRI: signal intensity index [SII]) between groups were tested using Fisher's exact test and Student's t-test. Logistic regression models were estimated for both qualitative and quantitative analyses. At quantitative analysis, discrimination abilities were determined according to the area under the receiver operating characteristic (ROC) curve (AUROC) with computation of optimal cut-off points. The diagnostic accuracies of CT and MRI were compared using McNemar's test. RESULTS: At qualitative assessment, MRI had higher accuracy than CT (96.4%, 80/83 and 86.7%, 72/83, respectively). At quantitative analysis, both the radiodensity and SII were significantly different between groups (p<0.0001). For CT, at quantitative assessment, the AUROC of the radiodensity in discriminating between groups was 0.904 (optimal cut-off point, 20 HU) with an accuracy of 77.1% (64/83). For MRI, the AUROC of the SII was 0.989 (optimal cut-off point, 7.766%) with an accuracy of 96.4% (80/83), which was significantly higher than CT (p<0.0001). By using optimal cut-off points for cases with an erroneous diagnosis at qualitative assessment, accuracy improved both for CT (89.2%, 74/83) and MRI (97.6%, 81/83). CONCLUSION: Quantitative analysis is useful in evaluating patients with MG and improves the diagnostic accuracy of CT and MRI based on qualitative assessment. Chemical-shift MRI is more reliable than CT in differentiating THYs from non-thymomatous conditions.

Single-channel properties of cloned cGMP-activated channels from retinal rods.

Single-channel properties of a cloned channel activated by cyclic GMP have been analysed. The mRNA encoding for the channel was injected into oocytes of Xenopus laevis and the current flowing through a single ionic channel activated by cGMP was studied in excised patches under voltage-clamp conditions. The ionic channel activated by cGMP had a single-channel conductance of 32 +/- 2 pS at +120 mV and 25 +/- 4 pS at -120 mV, and its conductance was not significantly affected by increasing the cGMP concentration from 20 microM to 200 microM. The single-channel currents in the presence of NH+4, Na+, K+, Li+ and Rb+ in the medium bathing the cytoplasmic side of the membrane at +140 mV were 5.3, 4.7, 3.8, 1.3 and 0.8 pA, respectively. The single-channel current in the presence of Cs+ was less than 0.5 pA. Ca2+ and Mg2+ (both 0.5 mM) in the presence of 100 microM cGMP did not appreciably affect the channel activity at membrane potentials more negative than -80 mV, whereas at +100 mV they reduced the single-channel conductance by about threefold. The ionic selectivity and the blockage by divalent cations of the native channel found in amphibian rods and in the cloned channel from bovine rods are quite similar. However, the cloned channel has well-resolved openings, especially at positive membrane voltages, whereas the native channel is characterized by a continuous flickering between the open and closed state.

Jump-diffusion processes as models for neuronal activity.

Aiming at an improvement of the existing neuronal models, we consider a mixed process ensuing from the superposition of continuous diffusions and of Poisson time-distributed sequence of impulses and focus our attention on the moments of the firing time. In particular, we consider three different instances: the large jumps model in which each jump causes the neuron firing, the reset model characterized by jumps towards the resting potential and a more general model where constant amplitude excitatory and inhibitory jumps are superimposed on diffusion. By resorting to analytical arguments and to numerical computations, the main behavioral differences of the considered models are outlined.

Simulation methods in neuronal modelling.

The interspike distribution can be modelled as the first-passage-time distribution of suitable diffusion processes with biologically meaningful boundaries. Since various mathematical difficulties arise when one attempts to obtain closed form solutions for first-passage-time problems, one can resort to simulation methods in order to study the problem. In this paper we pinpoint possible overestimations connected with simulations of first-passage-times for diffusion processes and propose a suitable simulation technique to determine the moments and the distribution of the firing times. After checking the validity of the proposed method in some instances where numerical evaluations for such quantities are available, we apply the simulation algorithm to model the spiking activity by means of a particular diffusion process constrained by a suitable time varying threshold.