Kriščiukaitis, Algimantas

Įvairiais būdais registruotų stuburo ir jį supančių struktūrų vaizdų suliejimo metodikos ir informatyvių diagnostinių požymių sistemos kūrimas. Tyrimui naudoti DICOM standarto MRT vaizdų rinkiniai, užregistruoti pacientams, kenčiantiems nuo apatinės nugaros dalies skausmo. Vaizdai registruoti naudojant Siemens Magnetom Aera (MRT 1,5 T) įrangą. Vaizdų rinkiniai buvo registruoti T1 ir T2 režimais, o dalis vaizdų dar ir T2 TIRM bei T2-TSE-RST režimais. Atlikę pagrindinių komponenčių analizę specialia MatLabtm programa tarpusavyje sutapatintiems pagal anatomines struktūras vaizdams, gavome optimalius pažaidos zonos vaizdus pagrindinių komponenčių erdvėje.

Event-related potentials and particular case – visual evoked potentials are electrophysiological events reflecting sensory, cognitive or motor reaction. The observed electrical activity (brief deflections, and valleys) is concurrent with other background electrical activity registered from the brain. So, in reality we register the mixture of the signals and methods for separation of visual evoked potentials from background EEG should be elaborated. Quantitative evaluation of visual evoked potential parameters (latency and amplitude of the principal peaks and valleys) provides useful information for studying brain processes and is important for clinical diagnostics.

The aim of this work was to reveal most informative quantitative estimates of single sweep visually evoked potentials shape reflecting optical nerve function related pathologies.

Objectives: Theoretically analyze existing methods of visually evoked potentials shape analysis, in particular, methods for single sweep analysis; Analyze visually evoked potentials recordings using MATLAB computational environment based-software and select informative signal shape reflecting parameters; Reveal most informative signal shape parameters reflecting pathologies related with optic nerve malfunction (neuritis, SD and other). A review of publications about VEP genesis, methods of their registration, preprocessing and evaluation revealed need of further development of methods, based on single sweep analysis.

Methods: Set of clinical recordings of VEPs registered in “Akiu Klinika" of Lithuanian University of Health Sciences used for analysis supplemented by synthetic recordings generated to imitate selected pathologies caused changes in VEP shapes. Principal Component Analysis based method for extraction of VEPs from the background EEG was used for signal pre-processing. Wavelet transform based method was proposed to estimate time and frequency characteristics of pre-processed VEPs. Validation of the method was performed on simulated data.

Conclusions: Theoretical analysis and review of publications showed that VEPs and other event related potentials carry important diagnostic information, so development of advanced methods for their recording and analysis, especially single trial analysis, is very important for clinical use. Analyzing VEP for reconstruction, it is shown that every singlesweep with finite amount of sweeps, is possible by means of multivariate analysis methods (e.g. PCA, KLT). We proposed to use wavelet transform for evaluation of specific changes in VEPs related with optic nerve malfunction.