Gian Carlo Filligoi

Automatic visual field : a software diagnostic procedure

A recently developed technique for the automatic acquisition of data on visual field losses was tested, with reliable neuro-ophthalmological parameters, on 300 subjects with and without eye disorders. Algorithms for acquisition and processing of the central visual field were implemented on an IBM-AT personal computer with standard peripherals. The contours of scotomata (visual field losses) were best estimated by probabilistic adaptive enhancement of data sampling in those areas with greater visual variability. Image processing tools were specially designed to extract information concerning the size, shape, position and number of scotomata in the visual field. The diagnosis of wider campimetric lesions required parameters such as symmetry or specularity coefficients, obtainable by analysing the visual field of both eyes. Data obtained were correlated to the pathology involved by univariate statistical tools.

Spatial and temporal distribution enhancement of movement-related brain macropotentials

Conventional brain maps suffer from severe limitations due to both the spatial blur of potential distributions and the dependence on electrical reference. The surface Laplacian (SL) has been used to deblur movement-related brain macropotentials (MRBM) since it acts as a high-pass spatial filter that reduces the head volume conductor effects. Moreover, the method usually employed to improve the signal-to-noise ratio (SNR) is the well-known synchronized average. However, this method is no longer valid when the object of the study is the sweep-by-sweep variability. In this case, the SNR of original and Laplacian-transformed single-sweep MRBM can be improved by autoregressive with exogenous input (ARX) filtering. In our study, isolated or combined ARX and SL are applied to enhance the spatial distributions of single-sweep MRBM associated with unilateral voluntary self-paced finger movements in humans. It shows that single-sweep brain mappings are more coherent to physiological findings when ARX is first used followed by SL.

Chromatic perception test: a computer-based approach

The analysis of luminance visual defects (i.e. obtained with white stimuli on a black background) in the central area of 200-300 around the fovea provide a valuable tool in neuro-ophthalmological diagnosis. In fact, distribution and depth of monochromatic visual defects reflect disease or damage to specific portion of the brain along the visual pathway. Concerning chromaticity tests, it is well known that they improve the diagnosis of many neurological diseases, such as multiple sclerosis, optic neuritis, diseases due to vitamin deficiency, etc. A new automatic technique for testing chromaticity within the Central Visual Field (CVF) by means of a PC has been developed. According to the ICO-1979 recommendation for perimeter standardization, the coloured light stimuli were isoilluminant with a 12 Lux white background. Stimuli differed from the background for hue (red, green, blue) and saturation (two levels each: one corresponding to 30% of the Maximal Saturation (MS) and the other to 10% of MS). The new test proved to be accurate and rapid. It also achieved high sensitivity, because it classified mild defects of color vision of which subjects were unaware. This pilot study on patients with multiple sclerosis proved that the test achieved significantly greater sensitivity than standard luminance visual field analysis.

Fast-Slow phase separation of Near InfraRed Spectroscopy to study Oxigenation v/s sEMG Changes

The possibility of studying non-invasively local muscle oxidative metabolism during exercise has been recently enhanced thanks to the use of Near-InfraRed Spectroscopy (NIRS). Moreover, the myoelectric manifestations of muscle fatigue occurring during sustained isometric contractions have been extensively studied with quantitative surface electromyography (sEMG) and are described by means of some sEMG global parameters extracted in time- and/or frequency-domain. With the aim of analyzing together both NIRS and sEMG data recorded on Biceps Brachii, an experimental protocol has been applied to seven subjects. During the experimental session, each subject had to perform two different kind of physical trials, a constant force and a cyclically varying isometric contraction. While examining the whole set of data with the aim of investigating the relationship between modifications of sEMG and underlying metabolic status, we faced up with the problem of objectively separating the two main phases of NIRS data. Results clearly indicated the presence of an initial fast phase of muscle desaturation followed by a slow phase, regardless of the kind of exercise. This behavior was paralleled by an analogous rate of change of sEMG parameters, thus suggesting a strong link between the two phenomena. As an objective criterion for separating the fast and the following slow phase an ad-hoc algorithm has been implemented. NIRS & sEMG data have been analysed by considering only those data pertaining rigorously to one of the two phases, whereas data collected during initial, transiction and final phases were discarded. In this paper we present some details on the algorithm for automatic separation of the two phases, with the most important results on statistical significance of the relationship between parameters extracted from sEMG in the fast phase v/s muscle oxigenation.

Differential diagnosis of central campimetric lesions by means of parametric and nonparametric discriminant analysis

Campimetry is a daily clinical examination used for diagnosis of neuro-ophthalmological disorders. An Automatic Campimeter based on a low-cost personal computer was designed and realized. Data from more than 1000 patients examined with our campimeter was analysed. Specific indexes of pathology were tried out. Multivariate Statistical tools were applied to asses the parameters skill in pathology recognition. Two different approachs were followed to this aim: Parametric and Nonparametric discriminant analysis.