Chapter 10 per tissue type (see also Table 10.1). The sterile fibre probe was handled by the surgeon and gently brought into direct contact with one of the designated tissues (see Figure 10.1). If blood was visibly present on the tissue surface, it was dapped away using a sterile gauze. Between the measurements on different locations, the probe tip was swiped with a clean sterile gauze wetted with saline. Acquired data was labeled according to the tissue type description of the attending surgeon. To correct for dark current14, the spectrometer was calibrated prior to in vivo data acquisition. After the completion of in vivo spectroscopy, a reference spectrum was acquired, for calibration purposes, by direct contact measurement on a white reference phantom (Optical‐grade spectralon reference; Labsphere, Inc., North Sutton, New Hampshire USA). The integration times of the Silicon and InGaAs sensor were individually optimized during the Spectralon calibration. No correction for ambient light was performed. Data processing For inter‐patient comparability, all raw in vivo spectra were calibrated using the dark current and reference reflectance spectrum. This normalizes the radiance spectrum to yield the reflectance, which manages the problem of spectral non‐uniformity of the illumination device and influence of the dark current25. Since both reflectance intensity and spectral shape are related to the composition of the tissue, no further normalization steps were performed. To identify possible distinctive features for tissue‐specific 150 enhancement, 36 features (i.e. 18 gradients and 18 amplitude differences at predefined points in the tissue spectra) were extracted based on known wavelengths related to characteristic absorption features for blood, water and fat26‐28. These features are described in more detail in our previous publication23. Figure 10.2 illustrates the characteristic wavelengths and features in a mean spectrum for human adipose tissue (anatomical region: neck). All data processing was performed by in‐house developed software (using MATLAB environment Version 7.7.0, MathWorks Inc., Natick, Massachussetts, USA). To evaluate the results for application in imaging technology for identification of parathyroid within its natural surroundings, we performed three separate classification steps: i.e. Si‐sensor (≤1000 nm) features, InGaAs‐sensor (≥900 nm) features, and features covering the whole range (350 – 1830 nm). Considering the number of subjects, the number of extracted features (n=36) is too large to perform a statistically meaningful classification, as the extracted features could be redundant in the information they retain. Therefore, using combinations of all 36 features to build a classifier would result in a dimensionality problem and over‐fitting29. To omit this
proefschrift_Schols_SLV
To see the actual publication please follow the link above