Multispectral characterization of tissues encountered during laparoscopic colorectal surgery type per subject. Directly after completion of the measurements (average 10‐15 minutes) the specimen was sent, undamaged, to the pathology department. Data processing of spectral measurements Data were processed using MATLAB (Version 7.7.0, MathWorks Inc., Natick, Massachussetts, USA). All spectra were normalized at a wavelength of 810 nm, which was chosen since, at this isobestic wavelength, the absorption of oxygenized and de‐oxygenized blood are equal; water absorption is also low3. This choice of normalization wavelength minimized the influence of blood smears on the fiber probe top. The acquired spectra were divided by the baseline reference spectrum acquired on spectralon. This resulted in optical reflectance spectra which are largely independent of the optical characteristics of the applied illumination source. Determination of an appropriate classification method The classification methods applied for this study consisted of two parts – training and validation – like many classification methods generally do. During the training part a set of representative measurements and the corresponding outcome were used to learn how to classify the measurements under consideration. In this case the measurements were reflectance spectra (X) of different types of tissue; the classification was the tissue type (Y) itself. We applied a Total Principal Component Regression (TPCR), a method that is well described in literature9,10. Within the Supplementary section our application of TPCR is discussed in more detail. 95 Results A total of 929 spectral measurements were performed on fresh colonic specimens of 19 patients who underwent open or laparoscopic colorectal surgery. The following tissue types were distinguished: mesenteric fat (n=269), blood vessels (n=377), colonic tissue (n=213), ureter (n=10) and tumorous tissue in colon (n=60). See Table 7.1 for study subject characteristics and the number of collected spectra per tissue type. The spectral profile of each tissue type is plotted combined in Figure 7.3A and separately in Figure 7.3B – 7.3F.
To see the actual publication please follow the link above