1 Photoacoustic Spectroscopy David Birtill 1 , Anant Shah 1,2 , Michael Jaeger 1,2 , Andreas Gertsch 1 , Jeffrey Bamber 1 1 Joint Department of Physics, 2 CRUK-EPSRC Cancer Imaging Centre, Institute of Cancer Research and Royal Marsden NHS Foundation Trust, Downs Road, Sutton, Surrey, SM2 5PT Abstract—A photoacoustic (PA) spectroscopy system has been built to study the differences between the PA spectra of oxy- genated and deoxygenated blood, and various PA contrast agents, with a view to optimising the identification of these media in clinical PA images. A variable-wavelength laser delivers short (ns) pulses of light, via a fibre optic cable, into a sample held between 75 μm transparent membranes. The optical wavelength is con- trolled by a computer, which scans the wavelength range 400-700 nm using a different pulse for each wavelength. Each pulse causes the sample to momentarily expand and emit a pressure wave, the energy of which is measured by the computer using a digital oscilloscope that samples the signal from a strongly focused 7.5 MHz ultrasound transducer. The resulting optical spectra are corrected for some system variables, such as the wavelength- dependent laser energy. Further corrections are planned, so that the measurement is truly of optical absorption coefficient at each wavelength. Even without these additional corrections however, the measured PA spectrum of oxygenated blood strongly resembles the published optical absorption spectrum. These results suggest that, in addition to its intended use for determining the optimum wavelengths for clinical PA imaging of blood oxygenation level and contrast agent concentration, this system may have applications as a laboratory spectrophotometer. Unlike traditional transmission spectrophotometers, which mea- sure the extinction coefficient, the PA spectrometer will measure the absorption coefficient. This will make it suitable for use with (a) optically dark samples such as normal blood, which cannot be analysed in a standard spectrophotometer without dilution, and (b) turbid media, which normally require an optical scatter- correction to convert the extinction coefficient to an absorption coefficient. Index Terms—Photoacoustic spectroscopy, photoacoustic imag- ing, optoacoustic imaging, blood spectrum, optical absorption. I NTRODUCTION Photoacoustic Imaging (PAI) inspects the optical absorption of the tissue. Tissue is irradiated using short laser pulses and ultrasound waves are generated within the tissue upon optical absorption (Wang 2009, Lai and Young 1982, Sigrist and Kneubuhl 1972, Jaeger 2007). An image is formed of the optical contrast based on the arrival times of the acoustic waves. As blood has a high optical absorption over much of the optical spectrum, photoacoustics can be used to detect blood vessels and could be used to determine the cancer’s stage, as advanced tumours create their own blood vessel network to sustain the tumour. Photoacoustics can be taken further by changing the wavelength of the emitted laser pulse. An image acquired at different wavelengths enables spatially resolved spectrometry. Therefore it should be possible to separate signals from oxygenated blood and deoxygenated blood to determine hypoxia. Hypoxia is important in oncology as it has been shown to be associated with tumour aggressiveness, angiogenesis and local recurrence; it affects and is affected by radiation therapy and some chemotherapy agents (Tatum at al. 2006, Semenza et al. 2007, Vaupel and Mayer 2007). A photoacoustic (PA) spectroscopy system has been built to study the differences between the PA spectra of oxygenated and deoxygenated blood, and various PA contrast agents, with a view to optimising the identification of these media in clinical PA images. METHOD This system comprises of a multiwavelength OPO (Optical Parametric Oscillator) laser system (Continuum Powerlite plus Panther OPO) which emits a laser pulse at a given wavelength. This light is guided by a fibre optic cable into the water bath, directed at a thin box with two 75 μm optically transparent membranes 2mm apart (Opticell) filled with 10ml of human venous blood (figure 1a). Then the emitted photoacoustic wave is detected by a focused 7.5 MHz single element ultrasound transducer. The acoustic signal was recorded by an oscilloscope using the laser Q-switch trigger for starting the acquisition. For each illuminating wavelength 100 traces were averaged to reduce noise. Then a low pass filter was used to reduce high frequency noise (figure 1b). The peak to peak amplitude was recorded over the visible wavelength range of 420 nm to 700 nm with a step size of 10 nm, shown in figure 2b. Also the energy of the laser output at the fibre optic cable exit was measured 100 times for each illuminating wavelength to produce a plot of the mean energy spectrum with the standard error as shown in figure 2a. The relative absorption coefficient for venous blood, calculated by dividing the peak to peak amplitude by the illuminating energy for each wavelength, is shown in figure 2c. This relative absorption coefficient μ * a assumes the following relation: μ * a (λ)= V (λ) E(λ) = μ a (λ)ΓK where V is the signal amplitude (voltage), E is the en- ergy of the laser output both of which are wavelength (λ) dependent. Before the true optical absorption coefficient μ a can be calculated the other system dependent variable K, which includes the transducers response to the intensity of the acoustic wave, and the Gruneisen coefficient Γ, which is a combination of the thermal expansion coefficient, sound speed, and specific heat capacity of the sample medium, needs to be known and accounted for (Soroushian 2010). As the Gruneisen coefficient is not wavelength dependent and to first order does not vary greatly between water based samples, this small variation is neglected at this stage. The other system dependent