Patients (2 subjects) with no clinical evidence for blisters and ICS titres < 1:40 at the end of the therapy were considered as PV in remission (recovered). Informed consent was obtained from all participants. Blood samples were taken by using a butterfly needle and collected into 10 ml blood serum separator tubes under routine conditions. Blood was separated into a lighter phase (serum) and a heavier phase after centrifugation at 1000 �� g for 5 min. Serum was transferred to fresh tubes and samples were then stored at -20��C until use. For Raman examination a drop of serum was placed on a microscope glass and covered by a 170 ��m thick cover glass, Samples were examined using a micro-Raman spectrometer equipped with a confocal microscope (Horiba-Jobin Yvon). A He-Ne laser operating at a light wavelength �� = 632.8 nm with a laser maximum delivery of about 3.5mW at sample level was used as exciting source. The spectrograph included monochromator, a CCD detector with a chip size of 1024��256 equipped with a Peltier cell and with a grating of 1800 grooves/mm. The laser light was focused on the sample surface by means of a 50X long working distance optical objective (Olympus MPLAN 50x/0.75) on an excitation spot size with a diameter of about 50 ��m. A notch filter (Kaiser Optic) was used in the collimated scattered beam to reduce the laser background. This device was directly controlled by the data acquisition software. A color video camera integrate within the microscope enabled the user to visualize the sample in reflection. For each patient two blood samples were prepared and several spectra were acquired for each samples. The ��-RS spectrum was generally evaluated in the wavenumber shift regions of 1000�C1800 cm-1 and 2700-3000 cm-1 where peaks assigned to vibration modes of amide I (near 1650 cm-1), CH2 and CH3 bending modes (near 1445 cm-1, 2850-2930 cm-1) and amide III (1240�C1260 cm-1) are placed [5, 8]. Accumulation times ranging in 60�C300 s were used for the ��-RS acquisitions. It should be noted that we have used visible low power laser, in contrast with other authors who prefer near infrared laser source to further minimize fluorescence effects, although the latter approach needs the use of more powerful lasers as reported in literature [5-7, 10].2.2. Wavelet deconvolutionLight dispersion effects due to the presence of liquid in the samples considered, affected the quality of the ��-RS spectra, which featured typically a large background signal and a relatively high level of noise. An automatic numerical data treatment based on wavelet algorithm was used in order to suppress the non-correlated signal, to subtract the background signal and to increase the quantitative readibility of the Raman signal. As reported by Camerlingo et al.