Font Size: a A A

Drug Testing Based On Terahertz Time-domain Spectroscopy

Posted on:2014-07-22Degree:MasterType:Thesis
Country:ChinaCandidate:S Q DuFull Text:PDF
GTID:2181330422486105Subject:Optical Engineering
Abstract/Summary:PDF Full Text Request
Terahertz radiation with a frequency range of0.1THz-10THz, sandwichedbetween microwave and infrared, belongs to far infrared region. It has high bandwidth,low-energy, non-destructive testing, high signal to noise ratio, sensitive to phaseinformation and other advantages, and using the new THz technique, we can get thecharacteristic of substance in THz band. The refractive index of the many drugs, suchas absorption coefficient abundant spectral and structural information, can be obtainedby terahertz time-domain spectroscopy based on the characteristics of the terahertzelectromagnetic wave. Compared with Fourier transform infrared spectroscopy,terahertz time-domain spectroscopy technique has higher signal-to-noise ratio and canbe used to do lossless non-ionizing and high sensitivity spectroscopic measurementswithout cryogenic bolometer instrument. The vibrational and rotational energy levelsof most macromolecules fall in terahertz band. Especially some biological andchemical macromolecules have their own physical groups and thus can be analyzed bythe structure of the material to identify the physical structure of its characteristicfrequency. This paper combines terahertz time-domain spectroscopy, efficacyrecognition technology and support vector machine technology, and applied them tothe study of medicine.In this paper, we used transmitted terahertz time-domain terahertz spectroscopysystem to measure four kinds of anti-diabetic pills easily, including gliquidone,glipizide, gliclazide, and glimepiride. Frequency domain spectrum was obtained byFourier transform, and got terahertz absorption spectrum the0.3-3.0THz range. Theexperimental results showed that an obvious resonance peak is found at1.37THz.Then, the common sulfonylurea group found by HIPHOP algorithm is calculated andthe result agrees well with the experimental value. Further, with the support vectormachine method, we used different kernel functions (linear kernel, polynomial kernelfunction, radial basis kernel function) to distinguish and classify terahertz absorptionspectrum of four kinds of anti-diabetic pills in the1.5-2.0THz range, the resultsshowed that four kinds of anti-diabetic pills could be distinguished well, and theaccuracy rate was100%.The subject work mainly has several aspects below: Firstly, the construction ofexperimental time-domain terahertz spectroscopy system, including the design,optimization and improvement of terahertz system, and participation in the preparation of software that experiment required. Secondly, preparation of experimental samples,including the selection of the material of sample box, selecting the high-densitypolyethylene as the material of the sample box, the THz transmittance of this materialwas more than65percent; the fabrication of the sample boxes, sample box wassandwich structure, the intermediate sample thickness is0.5mm. Thirdly, processingand analysis of experimental data, including the extraction of the spectrum signal andthe absorption coefficient. Finally, we distinguish the pharmacophore and identificationof anti-diabetic pills, including group identification, vibration mode simulation, featuredata extraction, classification model and on the identification of four western medicineclassification.
Keywords/Search Tags:Terahertz wave, Drug design, Support vector machine, Kernel function, Absorption spectrum
PDF Full Text Request
Related items