1. Introduction

The chemical compositions of tissue,such as DNA/RNA, hemoglobin,melanin,water and lipid,bring important information regarding the anatomical,functional,metabolic,and molecular/ genetic processes in biological systems. Therefore,tissue chemical compositions can be employed as the reference standards for the diagnosis of many diseases ^{[1, 2]}. Characterizing chemical information in biological tissue has always been the subject of intense research in chemistry as well as medicine.

Due to the unique optical spectra of atoms and molecules, optical spectroscopic techniques have been used to analyze the chemical properties of materials in analytical chemistry ^[3], material science ^[4] and astronomy ^[5] for a long time. With the advantages in detecting chemical information,optical modalities, such as optical coherence tomography ^[6],two-photon microscopy ^[7] and confocal laser scanning microscopy ^[8],have also shown great success in biomedical imaging ^{[9, 10, 11]}. Most biological tissues are strongly scattering for electromagnetic waves in the spectral range from ultraviolet to near-infrared. Within this range,the optical diffusion limit,i.e. optical mean free path,in most biological tissues is on the order of 1 mm. When beyond this depth,multiple light scattering would quickly randomize the trajectories of incident photons and prevent effective optical focusing in tissue ^{[12, 13, 14, 15]}. Therefore,pure optical modalities either suffer from poor spatial resolution or have limited imaging depth ^[16].

In comparison to optical wave,acoustic wave is much less scattered in biological tissue. In soft biological tissues,the acoustic scattering coefficient is about three orders of magnitude less than the optical scattering coefficient ^[17]. Benefitted from the lower scattering,acoustic method can achieve much better spatial resolution in deep tissue,usually equivalent to 1/200 of the imaging depth ^[12]. Conventional acoustic imaging,such as Bmode ultrasonography,however,is purely relying on the detection of mechanical properties and structures of biological tissues without considering tissue chemical information (i.e. the contents and changes in chemical components). Therefore,B-mode ultrasonography is not sensitive to tissue physiology and function which are highly valuable for the diagnosis of many diseases.

PAI is a recently developed technique whereby non-ionizing, non-invasive laser pulses are used to generate ultrasonic signals from biological tissues followed by signal detection via ultrasonic transducers to form images. Benefited from the low scattering of ultrasonic signals,PAI is able to achieve high resolution in optically scattering biological tissues at unprecedented depths. Moreover, like other conventional optical imaging modalities,PAI can be used to differentiate tissues containing various chemical substances based on their different optical absorption spectra. The hybrid PAI technology,by combining the advantages of both light and sound, shows tremendous potential applications in medicine and biology, and has already been used in evaluating the anatomical,functional, metabolic,molecular and genetic information of biological samples ^{[12, 13, 14, 15, 16, 17, 18, 19, 20]}.

In this mini-review article,we will first introduce the basic principles and two major implementations of PAI,photoacoustic tomography (PAT) and photoacoustic microscopy (PAM). The applications of PAI in detecting the chemical compositions and evaluating the microstructures in biological tissues will also be introduced. 2. Principles and implementations

When biological tissue is illuminated by laser light,the electrons transit from the ground state to an excited state as a result of photon absorption. The excitation is defined as thermally or stress confined if the laser pulse duration is much shorter than the thermal relaxation. The rapid deposition of laser energy causes an immediate increase in pressure within the radiated region. The release of this pressure transforms thermoelastic expansion to ultrasonic wave,named photoacoustic (PA) wave. The PA pressure p(r,t) at position r and time t can be described as ^{[21, 22]}:

where c is the speed of sound,t(t) is the waveform of the electromagnetic illumination,and p₀(r) is the initial PA pressure excited by optical illumination. p₀(r) is directly proportional to the Gru¨ neisen parameter Г(r) and the optical absorption coefficient A(r),i.e. p₀(r) = Г(r)A(r). After PA signals are picked up by an ultrasound transducer(s),the information of the location and the density of the optical absorbers in tissue can be recovered by the following equation ^[23]:

where t = ct,Ω₀ is a solid angle of the whole surface S₀ with respect to the reconstruction point inside S₀,r₀ is the measurement position,and dΩ₀ is the solid angle for a detection element dS₀ with respect to a reconstruction point at r.

Assuming Г(r) is homogenous,the spatially distributed optical absorption coefficient A(r) in a 3-dimensional space or a 2- dimensional cross-section can be obtained. Because the optical absorption A(r) of a biological tissue is closely related to its chemical composition,the spectroscopic form of PAI,i.e. imaging the target tissue with multiple optical wavelengths,allows quantitative measurements of the chemical and functional information in the tissue.

Based on the differentmethods for image acquisition aswell as the differences in targeted spatial resolution and imaging depth, PAI can be divided into two large categories which are PAT and PAM. Fig. 1(a) ^[24] shows a typical PAT system where an unfocused laser beam covering a large area on the sample generates PA signals and an unfocused ultrasound transducer detects the PA signals. The tomographic images are generated from PA signals by using a reconstruction algorithm based on the inversely solving the photoacoustic wave equation. Several reconstruction algorithms,such as back projection ^[23],Radon transform ^[25],and Fourier-domain algorithms ^[26],have been developed for PAT. The spatial resolution and the penetration depth of a PAT system are scalable with the ultrasonic frequency. The lateral resolution of PAT can be estimated as R_L = 0.71c/ (NA_c′f₀),where c is the speed of PA signal,f₀ is the center ultrasonic frequency,and NA_c is the numerical aperture of the transducer. The axial resolution of PAT can be estimated by R_A = 0.88c/△f, where △f is the bandwidth for photoacoustic signal detection and is proportional to f₀ ^[12]. The imaging depth is limited by not only the frequency-dependent acoustic attenuation but also the optical attenuation in the biological sample.

In contrast to PAT,Fig. 1(b) ^[27] shows a typical PAM system where a focused laser beam is used for signal generation and a spherically focused ultrasound transducer is used for signal detection. To generate an image,PAM scans along the sample surface point by point without involving tomographic image reconstruction. PAM is further divided into two categories including optical resolution PAM (OR-PAM),where the lateral resolution is determined by optical focus,and acoustic resolution PAM (AR-PAM),where the lateral resolution is determined by ultrasonic focus ^[12]. OR-PAM utilizing a strongly focused light beam for signal excitation results in a diffraction-limited optical resolution in the lateral direction which can be estimated by 0.51λ_opt/NA,where NA is the numerical aperture of the optical lens, and λ_opt is the wavelength of the laser. PA signals could be picked up by an unfocused ultrasonic transducer,or a focused ultrasonic transducer ^[12] aligned confocally with the optical lens to maximize the detection sensitivity. The axial resolution of ORPAM is still acoustically determined by the bandwidth of the ultrasonic transducer,△f,following the equation of 0.88c/△f ^[17]. The imaging depth of OR-PAM is determined by the optical mean free path which is usually within 1 mm for most biological tissues. Similar to OR-PAM or PAT,the axial resolution of AR-PAM is determined by the bandwidth of the ultrasonic transducer. Its lateral resolution,however,is determined by the acoustic focal diameter,following the equation of 0.71cl/(f₀·_c/2),where NA_c is the aperture of an ultrasound transducer and l is the focal length ^[17]. AR-PAM can provide spatial resolution on the order of micrometer and imaging depth of up to several millimeters ^[17]. Fig. 1(c) ^[28] shows a miniaturized PAM system for endoscopic imaging of internal organs. Compared to routinely used clinical ultrasound endoscopy,photoacoustic endoscopy (PAE) offers similar spatial resolution and imaging depth while providing additional functional information ^[29]. In recent years,PAE has been developed for potential applications in esophagus and colon ^{[28, 29]}. 3. Chemical information in photoacoustic images

In biological tissues,each chemical composition has its specific optical absorption spectrum from ultraviolet (UV) to nearinfrared (NIR). Electromagnetic radiation is more likely to be absorbed at wavelengths which match the energy difference between two quantum-mechanical states of the molecules. The optical absorption properties over a range of electromagnetic wavelengths,namely absorption spectroscopy,are primarily determined by the electronic and molecular structures of the sample. As a result,the optical absorption properties of amaterial are useful in chemical analysis because of its high specificity and quantitative nature. On the one hand,the excellent specificity associated with optical absorption properties allows different chemical compounds in biological tissue to be distinguished accurately. On the other hand,with the knowledge of the optical absorption coefficients of all the chemical components,the measurement of absorption spectroscopy of a sample can be used to quantitatively determine the absolute concentrations of chemical compounds. PAI is essentially probing the spectroscopic optical absorption properties of biological tissue by means of acoustic detection. Therefore,it naturally inherits the merits of optical spectroscopy in chemical analysis.

By targeting the selective optical absorption fingerprint of molecules,PAI can acquire the information of a specific chemical composition in tissue by choosing appropriate optical wavelength. For example,based on the strong optical absorption of DNA and RNA in UV spectrum,PAM has successfully imaged cell nuclei without involving any contrast agent (i.e. label-free),as an example shown in Fig. 2(a) ^[30]. Since cancer cells lose their control of DNA replication and their nuclei are different from normal nuclei in morphology,successful imaging of cell nuclei may contribute to cancer diagnosis and prognosis ^[31].

Hemoglobin accounts for most of the optical absorption in mammalian tissue in the visible to near-infrared spectrum ^[12]. One advantage of PAI working in this spectrum is its highresolution and high-sensitivity in presenting the volumetric information of blood volume in biological tissues. By mapping hemoglobin distribution,PAI has demonstrated excellent sensitivity in imaging and quantifying hemodynamic parameters in brain. As an example shown in Fig. 2(b),imaging of the vasculature as well as a blood-rich lesion in a small-animal brain has been achieved by a PAI system working at 570-nm wavelength ^[19]. Since blood circulation is highly relevant to metabolism and functions of tissues,functional imaging of blood circulation without involving contrast agents could strongly impact the research and clinical management of a variety of diseases,for example,diabetes,coronary artery disease,hypertension,atherosclerosis, and peripheral vascular diseases ^[12].

Similar to optical spectroscopy broadly used in analytical chemistry,multi-wavelength spectroscopic PAI can quantify the absolute concentrations of chemical compounds,and reveal rich functional and physiological information in biological samples. Based on the spectroscopic optical absorption differences between oxygenated hemoglobin (HbO₂) and deoxygenated hemoglobin (HbR),two major forms of hemoglobin in blood, PAI can quantify the relative concentrations of HbO₂ and HbR within a single blood vessel,and subsequently extract the information of oxygen saturation (SO₂) and total relative concentration of hemoglobin (HbT). Using laser illumination with different opticalwavelengths,λ1 and λ2,PAI can achieve two optical absorption images,μa(λ1) and μa(λ2),of a sample. As a first-order approximation,the blood absorption μa(λi) can be related to the relative concentrations C_ox and C_de of HbO₂ and HbR by the following equations:

where eox and ede are the known molar extinction coefficients of HbO₂ and HbR. Then the relative concentrations C_ox and C_de of HbO₂ and HbR can be obtained by solving Eqs. (3) and (4). After quantifying the relative concentrations of HbO₂ and HbR,PAI can subsequently extract the information of oxygen saturation (SO₂) and total relative concentration of hemoglobin (C_Hbt) by SO₂ = C_ox/ (C_ox + C_de) and C_Hbt = C_ox + C_de. An example result of dualwavelength PAI of blood oxygen saturation in vivo is shown in Fig. 2(c) ^[32]. The functional hemodynamic information that can be quantitatively evaluated by PAI without involving optical contrast agent is of great interest to the studies of cancer ^[33],brain activities ^[34] and skin diseases ^[35].

When the optical wavelength is beyond 1000 nm,the optical absorption in biological tissues is determined mainly by molecular vibration. Multi-wavelength photoacoustic imaging within the overtone bands has demonstrated the capability in differentiating fat and collagen in biological tissues. As an example,Fig. 2(d) is a PA image at 1195 nm showing scattered lipid deposition in an arterial wall ^[36]. In another research using 1200 nm light,PAI has been employed to study the lipid content in fatty liver ^[37].

Besides imaging based on endogenous contrast,PAI can also employ exogenous contrast agent to enhance the optical contrast and achieve the imaging of ‘‘invisible’’ biomarkers. Exogenous contrast agents are artificial materials which act to enable or augment the visibility of the preferred objects throughout the electromagnetic spectrum. The applications of exogenous contrast agents give PAI the ability to measure chemical and biological information at the molecular level. A great variety of PA contrast agents have been developed in recent years,such as organic dyes ^[38],metallic nanoparticles ^{[39, 40]},carbon nanotubes ^[41],and oligomer ^[42]. These agents have been wildly used in bio-medical and bio-chemical research. Based on the passive targeting mechanism,agents can significantly increase the contrast of PAI in cancer detecting ^{[38, 42]} and other applications ^[43]. PAI based on Doppler effect can achieve accurate flow measurements when powered by nanoparticles ^[44]. Furthermore,PAI in combination with contrast agents has shown promise in guiding and monitoring the therapeutic procedures ^[39]. An example of PAI enhanced by oligomerizable agents is shown in Fig. 3 ^[42],where the contrast of PAI of mice tumors is increased significantly after the agent administration.

4. Histological information in photoacoustic images

Besides revealing the chemical compositions of tissue,PAI also facilitates the assessment of histological microstructures in tissue. A study based on Monte Carlo simulation suggests that the ultrasonic spectral properties of PA signals are related to the erythrocyte aggregation level ^[45]. The distinctive features of the PA spectrum can be used to rapidly quantify variations in the morphology of single red blood cell ^[46]. In another research,as the result shown in Fig. 4 ^[47],significant differences in PA spectral parameters between cancerous tissue and noncancerous background tissue have been observed,suggesting that PA spectral analysis may enable cancer detection and tissue characterization in vivo. In a more recent study,the feasibility of differentiating fatty and normal livers by PA spectral analysis has been demonstrated, as the result shown in Fig. 5. The changes of histological microstructures in the liver associated with hemoglobin and lipid were characterized respectively by analyzing the power spectra of the radio-frequency PA signals at 532 nm and 1200 nm wavelengths, respectively ^[37].

The study on well-controlled gel phantoms containing optically absorbing microspheres has facilitated the understanding of the relationship between the PA spectral parameters (intercept,midband fit and slope) and the physical properties (density and sizes) of the optical absorbers ^[48]. A more interesting finding,as the results shown in Fig. 6 ^[49],is that PA spectral analysis can evaluate sub-resolution microstructures and can efficiently assess detailed tissue featureswithout need for broad-bandwidth and high-frequency detection of signals. In another study,a method of PA spectral matching has been developed to quantitatively evaluate stochasticmicrostructure in turbid media ^[50]. The frequency domain PA technique offers fundamental advantages for addressing a number of practical problems faced by conventional PAI. First,being able to performat relatively low frequencies avoids the high cost and the limited imaging depth associated with high-frequency PAI systems. Second,PA spectral analysis separates the effects from system components and tissue properties on image features,and delivers system-independent quantitative results ^[50]. Third,averaged power spectra provide a method of addressing the stochastic nature of tissue microstructure,and lead to measurements that are quantitative and repeatable.

5. Summary

In this mini-review article,fundamental principles and two major implementations of PAI including PAT and PAM are introduced. Then the capability of multi-wavelength PAI in evaluating chemical information in tissues,and the feasibility of PA spectral analyses in evaluating histological microstructures in biological samples are demonstrated. Combining light and ultrasound,PAI inherits the high resolution of ultrasonography in deep tissue and the high sensitivity of optical imaging to functional and chemical information of biological tissues. By taking advantages from its unique multi-physics nature,PAI,unlike any other optical or ultrasonic technologies,enables comprehensive and objective evaluation of both chemical and physical features in the target tissue,and holds promise for achieving quantitative imaging. In the past decade,the very active researches on PAI have already demonstrated many potential applications in biology and medicine. As a non-ionizing and non-invasive biomedical technology, PAI is now ready for the translation from basic research to clinical settings.

Acknowledgments

This work was supported by the National Basic Research Program of China (No. 2012CB921504),the National Natural Science Foundation of China (Nos. 11422439,11274167, 11274171,and 11028408),and SRFDP (No. 20120091110001).