부산대학교 도서관

Minimally invasive probe and optical biopsy system based on optical spectra recording and analysis seem to be a promising tool for early diagnostics of breast cancer. Light scattering and absorption spectra are generated continuously as far as the needle-like probe with one emitting and several collecting optical fibers penetrates through the tissues toward to the suspicious area. That allows analyzing not only the state of local site, but also the structure of tissues along the needle trace. The suggested method has the advantages of automated on-line diagnosing and minimal tissue destruction and in parallel with the conventional diagnostic procedures provides the ground for decision-making. 165 medical trials were completed in Nizhny Novgorod Regional Oncology Centre, Russia. Independent diagnoses were the results of fine biopsy and histology. Application of wavelet expansion and clasterization techniques for spectra analysis revealed several main spectral types for malignant and benign tumors. Automatic classification algorithm demonstrated specificity ∼90% and sensitivity ∼91%. Large amount of information, fuzziness in criteria and data noisiness make neural networks to be an attractive analytic tool. The model based on three-layer perceptron was tested over the sample of 29 'cancer' and 29 'non-cancer' cases and demonstrated total separation. [ABSTRACT FROM AUTHOR]