Evaluations of Oxygenation and Vascular Perfusion in Pre-Clinical Models of Cancer and Wound Healing Using OS-DCE OAI

Summary

Here, we present a protocol for performing Oxygen Sensitive – Dynamic Contrast Enhanced Optoacoustic Imaging (OS-DCE OAI) for evaluating pre-clinical models of cancer and wound healing. OS OAI can image oxy‑ and deoxyhemoglobin, mapping tissue oxygenation. DCE-OAI can image the pharmacokinetics of an exogenous contrast agent, thereby mapping vascular perfusion.

Abstract

Oxygen-sensitive optoacoustic imaging (OS OAI) and dynamic contrast–enhanced optoacoustic imaging (DCE OAI) provide complementary, noninvasive readouts of tissue physiology. OS OAI leverages multispectral detection of endogenous oxyhemoglobin and deoxyhemoglobin to generate maps of blood oxygen saturation (%sO₂), while DCE OAI tracks the in vivo pharmacokinetics of a near-infrared absorber to quantify vascular perfusion and permeability. The goal of this protocol is to present a unified OS-DCE OAI workflow that enables simultaneous assessment of oxygenation and perfusion in two widely used pre-clinical settings: orthotopic breast cancer tumors and full-thickness cutaneous wounds.

In this protocol, we demonstrated OAI of mouse models of breast cancer or a laceration wound. For OS OAI, multi-slice multispectral scans are acquired at 700–875 nm to allow spectral unmixing and estimation of %sO₂. For DCE OAI, repeated acquisitions at ≤5-s temporal resolution are performed before and after intravenous bolus injection of indocyanine green (ICG), using either multispectral sampling at 700-875 nm or single-wavelength sampling at 800 nm to generate time versus OA signal amplitude curves. The development of animal models, preparation for OA imaging, OS OAI and DCE OAI acquisitions, and analyses to measure characteristics of oxygenation and vascular perfusion are detailed, along with key troubleshooting guidance for motion artifacts, superficial absorbers, and failed injections.

Together, this OS-DCE OAI protocol yields an integrated functional portrait of hypoxia and vascular transport that can detect physiological changes in tumors and wounds earlier than gross anatomical measures. OS-DCE OAI is readily adaptable to other disease models.

Introduction

Optoacoustic imaging (OAI), also referred to as photoacoustic imaging or multispectral optoacoustic tomography, is a relatively new modality that couples optical excitation with ultrasound (US) detection to enable deep tissue imaging^1,2. In OAI, brief nanosecond laser pulses excite endogenous chromophores such as oxy- and deoxyhemoglobin, producing localized thermoelastic expansion and wideband acoustic waves that are detected by US transducers. These signals enable OAI to yield images with optical contrast at ultrasound-like spatial resolution while achieving centimeter-scale depth of view in vivo. When data are acquired at multiple wavelengths and analyzed with spectral unmixing, Oxygen Sensitive (OS) OAI can estimate relative concentrations of oxy- and deoxyhemoglobin and thereby infer tissue oxygen saturation (%sO₂)². Blood oxygen saturation provides a functional evaluation of microvascular physiology that is difficult to access noninvasively with other methods.

Vascular perfusion is also useful for evaluations of microvascular physiology. Vascular perfusion represents a combination of vascular flow, permeability, and vessel surface area³. To quantify perfusion, OAI can be used with an exogenous contrast agent to perform Dynamic Contrast Enhanced OAI (DCE OAI). In this approach, temporal imaging before, during, and after contrast agent injection monitors the pharmacokinetics (PK) of the agent. These results can be modeled to measure the wash-in rate (NK^trans) from the blood plasma compartment to the tumor extracellular extravascular compartment, and washout rate (k_ep) from the tumor extracellular extravascular compartment to the plasma compartment. We have developed a PK modeling method that avoids the notorious problem of light fluence (i.e., scattering and absorbance in tissues that cause variable radiant energy through the tissue)⁴. Our method normalizes the change in dynamic OAI signal from 0% (pre-injection) to 100% (maximum enhancement) and then uses differential equations to estimate NK^trans wash-in rate and the k_ep. Other DCE OAI studies have used standard PK modeling methods that suffer from the major assumption that light fluence is uniform throughout the tissue^5,6.

DCE OAI results can also be used to assess wash-in rates by making empirical measurements of Maximum Signal Enhancement (MSE), Time-to-Peak (TTP) to reach MSE, slope (typically for injection to TTP), and Initial Area Under the Curve (IAUC; again typically for injection to TTP post-injection)⁷. It should be noted that if the TTP occurs later in the PK curve, the slope or IAUC can be taken during the linear portion of the curve or from the time of injection to 1 min post-injection. We have shown that empirical measurements can also be useful when evaluating treatments administered to tumor models or when evaluating wound healing^8,9. Other studies have also shown value in using empirical measurements to evaluate tumor vascular permeability and liver fibrosis^10,11. The shape of the PK curve can also be used to qualitatively categorize tumor vascular perfusion as high or low^12,13,14.

In addition, the OAI signal at a single post-injection time point can be used as an empirical evaluation of perfusion, known as non-dynamic Contrast Enhanced OAI (CE OAI)¹⁵. A CE OA image that shows a bright tumor or wound is simplistic and highly visual and yet is often only used for a qualitative assessment of high vs. low perfusion. In general, the dynamic change in OAI signals after TTP is not analyzed to empirically evaluate vascular perfusion in tumors. However, as shown below, the wash-in rates in wounds are extremely fast, so that NK^trans, TTP, slope, and IAUC are challenging to measure. Therefore, we have used the linear or exponential slope of the washout phase as an empirical measure of vascular perfusion in wounds⁹.

Most OAI agents absorb in the near-infrared (NIR) wavelength range, which can penetrate deeper into tissues^16,17. However, water absorption at wavelengths greater than 900 nm limits the usable in vivo OAI wavelength range to 700–900 nm for most applications². Indocyanine green (ICG) is a prototypical DCE OAI agent because it offers strong NIR absorption at 780–800 nm, produces a moderately strong OAI signal, and has favorable biocompatibility with FDA approval for clinical use. These features enable intravascular contrast and interpretable PK curves.

We have developed a combined OS-DCE OAI protocol to evaluate oxygenation and vascular perfusion in pre-clinical models of solid tumors and wounds^8,9. The OS portion of this protocol can evaluate oxygenation in multiple imaging slices that can cover entire tumor and wound volumes in mouse models. The DCE portion of our protocol has an option to image a single slice of the tumor or wound using multiple laser wavelengths (improving the accuracy of contrast agent detection), or to image as many as 5 slices with a single laser wavelength (improving spatial coverage). We have used OS-DCE OAI to monitor the change in oxygenation and vascular perfusion after treating a tumor model with radiotherapy or with combretastatin A4 phosphate, a vascular disrupting agent^4,8. We have also used OS-DCE OAI to monitor wound healing⁹. DCE OAI gives OAI the ability to monitor progression via vascular perfusion that can be applied to a variety of diseases.

Protocol

Conduct all procedures under the approval of the Institutional Animal Care and Use Committee (IACUC) of your institution. All procedures in the protocol below were approved by the IACUC of our institution.

1. Anesthesia (for subsequent procedures)

1. Induce anesthesia. Initially induce and maintain anesthesia with 5% isoflurane in 100% oxygen carrier gas until the mouse is no longer awake.
2. Maintain anesthesia. Continue to anesthetize the mouse with 1.5–2% isoflurane in 100% oxygen carrier gas. Verify the absence of pedal reflex.
3. Maintain normothermia. Make sure to maintain animals on a warm surface and monitor temperature throughout the anesthetized period.
   NOTE: (Pause point) Proceed to tumor implantation or wound formation once an animal is fully anesthetized.

2. Preparing the model 

NOTE: All cell work should be performed in a biosafety cabinet to avoid exposure and potential contamination.

1. Establish a 4T1 orthotopic tumor model (Figure 1).
   1. Prepare cells on ice: Resuspend 4T1 cells in ice-cold PBS at a concentration of 2 × 10⁷ cells/mL.
   2. Mix cells with matrix: Combine the cell suspension 1:1 (v/v) with ice-cold biological hydrogel that has a rich extracellular matrix. The combination should yield a solution of cells at 1 × 10⁷ cells/mL.
   3. Load the syringe: Load the cell mixture into a sterile laboratory syringe with a volume no larger than 1.0 mL. Limit the volume to 100 µL per injection (1 × 10⁶ cells). As an option, gently warm the syringe to 30 °C, which makes the biological hydrogel slightly viscous and more focally retained at the injection site.
      CAUTION: Be mindful that syringes are sharp and avoid recapping.
   4. Anesthetize the mouse as described in section 1.
   5. Remove fur if present: Depilate the breast pad area. Omit this step if using athymic nude mice.
   6. Position the needle and inject: Sterilize the 4th inguinal mammary fat pad. Insert the needle bevel-up and inject 100 µL slowly into the fat pad to avoid reflux.
   7. Recover the animal from anesthesia: Return the mouse to a warmed cage and monitor the mouse until ambulatory.
   8. Allow the tumor to grow: Allow 8-10 days for 4T1 tumors to reach a minimum diameter of 0.3 cm before imaging.
      NOTE: Tumors with a larger diameter can also be imaged. The diameter of the tumor can be measured with a calibrated vernier caliper. Tumors that exceed the maximum diameter approved by the IACUC (or an equivalent institutional committee) should be euthanized. Follow orthotopic injection and adapt cell number/volume to the desired engraftment protocol.
2. Generation of an excisional wound in a mouse (Figure 2)
   1. Prepare the surgical field: Sterilize the work area and instruments.
   2. Anesthetize the mouse as described in section 1.
   3. Remove fur if present: Depilate the dorsal region. Omit this step if using athymic nude mice.
   4. Administer analgesia: Secure the mouse on a sterile surgical surface. Inject buprenorphine 0.05–0.1 mg/kg subcutaneously (SC) in the dorsal region.
   5. Disinfect the wound area: Disinfect the dorsal skin with povidone-iodine followed by 70% ethanol.
   6. Create an excisional wound: Use a 6-mm biopsy punch to outline the site. Excise the skin and panniculus carnosus to form a full-thickness defect.
   7. Seat the silicone splint:
      1. Center a donut-shaped silicone splint around the wound. Ensure that the inner diameter of the split is 2–4 mm larger than the diameter of the wound, and the width of the splint is ~5 mm.
      2. Cut a donut-shaped silicone splint from a sheet of silicone material. Use a clear suture material to sew the splint in place. If needed, use a thin adhesive film to seat the ring.
         NOTE: Pigmented suture materials may absorb NIR light and, therefore, should not be used.
   8. Suture the splint: Place eight evenly spaced interrupted sutures to secure the splint to the surrounding skin without strangulation.
   9. Apply an occlusive dressing: Cover the wound and splint with an occlusive dressing to protect the bed and standardize the wound environment.
   10. Recover the animal from anesthesia: Return the mouse to a warmed cage and monitor the mouse until ambulatory.
   11. Allow the animal to heal: 6 hours following initial wound formation, administer 1 dose of buprenorphine. The following day, administer 2 more doses of buprenorphine 6 hours apart. Continue daily care and assessments per study design or until the wound is completely healed.
   12. Replace the dressing each day: Replace occlusive dressing daily after imaging or photography to limit contamination and maintain sterility.
       NOTE: (Critical step) Splint the wound to suppress contraction and promote re-epithelialization, improving translational relevance to human healing. (Pause point) After applying the dressing, pause until the next scheduled assessment or imaging session.

3. OS-DCE OAI acquisition protocols

1. Prepare a mouse and the contrast agent for the OAI scan.
   1. Prepare the contrast agent.
      1. Prepare a solution of 1.62 mM of the ICG contrast agent in sterile water. Filter the solution with a 0.3 µm syringe filter. Fill a sterile laboratory syringe with 100 μμL of this solution.
         NOTE: This dose is 162 nmol or 125.5 μμg based on a 775 g/mol molecular weight of ICG. The dose is 6.28 mg/kg of mouse body weight, considering that a mouse weighs approximately 20 g. An animal equivalent dose is 12.3 times greater than a human equivalent dose¹⁸. Therefore, the human equivalent dose is 0.51 mg/kg of human body weight. The maximum tolerated dose for humans is 2 mg/kg¹⁹. Therefore, the dose used in this pre-clinical study is approximately ¼ of the maximum tolerated dose in humans.
   2. Prepare the OAI scanner.
      1. Fill the imaging tank with deionized, distilled water. Set the temperature of the tank to 36 °C for a mouse with <25 g body weight.
         NOTE: This tank temperature will maintain core body temperature near 37 °C because the mouse is wrapped in a plastic bag in the cradle. The temperature of the tank should be set to 35 °C for a larger mouse with >25 g body weight to avoid overheating the mouse, because a larger mouse has more subcutaneous adipose tissue that acts as an insulating layer.
   3. Anesthetize the mouse as described in section 1. For the remainder of the acquisition protocol, ensure the mouse remains anesthetized with 1.5–2% isoflurane in 100% oxygen carrier gas.
   4. If performing DCE OAI, place venous access.
      1. Flush a 27 G tail-vein catheter with heparinized saline. Insert the catheter into a lateral tail vein and secure the catheter to prevent it from uninserting from the tail vein (Figure 3A). Ensure there are no air bubbles present.
         NOTE: (Critical step) Use heparinized saline to maintain catheter patency.
      2. In addition, prepare a sterile laboratory syringe with a volume no greater than 500 µL that can connect to the catheter. Ensure the syringe contains 100 µL of 2.14 mM ICG (~322 nmol; ~16 µmol/kg for a 20 g mouse).
   5. Cover the mouse and cradle with ultrasound gel (Figure 3B).
      1. Apply a thin layer of clear ultrasound gel to the tumor or wound area (Figure 3C). Also, apply ultrasound gel to the cradle to make a seal between the cradle and the tumor or wound area.
         NOTE: Eliminate air bubbles. As an option, the ultrasound gel may be centrifuged to remove air prior to using the gel. Also, use a minimum amount of ultrasound gel to avoid bubbles that can form in thick gel layers.
   6. Place the mouse in the cradle (Figure 3D): Refer to the manufacturer instructions for the instrument. Place the mouse in a holder with the nose of the mouse firmly inside the nosecone. Use cuffs to hold the hands and feet of the mouse. Use the tooth loop to stabilize the head.
      1. (OPTIONAL) Use the cradle with a rat nose cone when imaging a mouse. The rat nose cone lifts the mouse head from the bottom of the cradle, which can avoid accidental drowning if a small amount of water leaks into the cradle.
   7. Place the mouse and cradle in the OAI instrument. Refer to the manufacturer's instructions for the instrument.
      1. Observe for any leaks in the mouse cradle. If leaks are present, remove the mouse and cradle from the instrument and re-seal the mouse in the cradle to avoid accidental drownings.
      2. If DCE OAI is desired, connect the catheter to the syringe with the contrast agent.
   8. Equilibrate the mouse prior to imaging: Allow 10 min for stabilization of the mouse physiology in the warm water tank (35-36 °C, depending on animal size, see section 3.1.2) prior to image acquisition (Figure 3E).
   9. Position the mouse relative to the OAI transducer:
      1. Move the mouse or the OAI transducer according to the manufacturer's instructions so that the tumor or wound is located in the multi-slice image set.
      2. Optimize the speed of sound for the image set. Complete this step while the mouse is being equilibrated prior to imaging (step 3.1.7).
      3. (Critical step) Maintain consistent animal orientation/slice position for all mice and at all time points to ensure rigorous results
2. Set the OS OAI parameters.
   1. Select the image slice parameters so that multiple slices are imaged to cover the entire tumor or wound area with 1 mm slice thickness.
   2. Select the absorption wavelengths. For OS OAI, select 700, 730, 760, 800, 850, and 875 nm absorption wavelengths.
   3. Employ signal averaging: Select parameters to acquire 10 averages per wavelength before acquiring the next wavelength, and before moving to the next imaging slice. The 10 averages per wavelength increase the signal-to-noise of each image slice.
   4. Set the number of repetitions:
      1. Set the number of repetitions as shown in Table 1 to maintain a total acquisition time of approximately 2 min. For users of an instrument that has a 10 Hz repetition rate, the total acquisition time is 6.0 s per image slice.
      2. Moving the mouse to the next imaging slice and then returning to the first slice position requires an average of 0.2 s, so that each slice requires 6.2 s.
      3. Perform a minimum of 3 repetitions to ensure measurement stability. A total acquisition time longer than 2 min is acceptable if more repetitions are desired.
3. Acquire the OS OA images.
4. Set the DCE OAI parameters.
   1. Set DCE OAI acquisition parameters for a single-slice imaging protocol with multiple wavelengths.
      1. Position the imaging slices. Select ≤5 imaging slices that are centered on the tumor or wound area with 1 mm slice thickness.
      2. Select the optimal absorption wavelength for the contrast agent. For indocyanine green within an in vivo environment, set the wavelength to 800 nm.
      3. Set the number of averages and repetitions as shown in Table 2. These recommendations will maintain a temporal resolution ≤ 5 s and a total acquisition time of 11.0–11.2 min for users of the OAI instrument, which has a 10 Hz imaging repetition rate.
         NOTE: Users of other instruments with a different imaging repetition rate should adjust the number of averages to maintain a temporal resolution of ≤ 5 seconds for each image set.
   2. As an alternative to step 3.4.1, set the DCE OAI acquisition parameters for a multi-slice imaging protocol with a single wavelength.
      1. Position the imaging slices. Select ≤ 5 imaging slices that are centered on the tumor or wound area with 1 mm slice thickness.
      2. Select the optimal absorption wavelength for the contrast agent. For indocyanine green within an in vivo environment, set the wavelength to 800 nm.
      3. Set the number of averages and repetitions as shown in Table 2. These recommendations will maintain a temporal resolution ≤ 5 s and a total acquisition time of 11.0–11.2 min for users of the OAI instrument, which has a 10 Hz imaging repetition rate.
         NOTE: Users of other instruments with a different imaging repetition rate should adjust the number of averages to maintain a temporal resolution of ≤ 5 s for each image set.
5. Acquire the DCE OA images.
   1. Record baseline images: Start the acquisition and collect 12 image sets (~1 min).
   2. Inject the contrast agent: Administer 100 µL via the catheter while continuing acquisition.
      NOTE: The bolus injection should be consistent for approximately 10–15 s to avoid a buildup of pressure in the catheter in the fragile tail vein. This consistency can be controlled with consistent pressure on the syringe plunger during manual injection, or by using an automated injector. Importantly, injection rates of 10–15 s smooth the initial delivery of agent to the tumor, rather than creating first-pass kinetics spikes in the PK curve that are caused by faster injection timings, which improves the PK modeling and analyses of empirical measurements.
      1. (Critical step) Avoid leakage of contrast agents from the blood vessel to surrounding tissues at the site of injection to preserve accurate bolus timing.
   3. Continue imaging: Acquire the remaining 128 image sets for 10.2 min post-injection to capture wash-in/washout kinetics.
   4. Monitor the imaging process to ensure that the images are acquired: Monitor the mouse breathing rate to ensure that the mouse physiology is stable during the scans.
   5. Remove the mouse and cradle from the instrument: Remove the mouse from the cradle. Allow the mouse to recover from anesthesia. Return the mouse to a warmed cage and monitor the mouse until ambulatory.
      NOTE: Use appropriate software and refer to the manufacturer's instructions for the OAI instrument. The steps below describe reconstruction and analysis steps for an OAI instrument that uses viewMSOT v4.0 software, and Matlab R2019b or a newer version of Matlab software.

4. OS-DCE OAI result analysis

1. Reconstruct the OA images.
   1. Select image reconstruction parameters: Set up image reconstruction to use a filtered back-projection method with a 50 kHz–6.5 MHz band-pass filter and enable automatic pulse-to-pulse energy correction.
   2. Reconstruct all images: Obtain images at all laser wavelengths, all image slice positions, and all time points. This volume of data may require substantial time for image reconstruction.
2. Analyze OS OA images.
   1. Unmix chromophores: Perform spectral unmixing to generate parametric maps of oxyhemoglobin (HbO₂), deoxyhemoglobin (Hb), total hemoglobin (where HbT = HbO₂ + Hb), and %sO₂ (where %sO₂ = HbO₂/HbT).
   2. Define ROIs:
      1. Manually draw an ROI that represents the tumor or wound in each image slice. Draw a conservative ROI that errs on the side of excluding pixels of the tumor periphery or wound periphery rather than including pixels for normal tissue at the periphery.
      2. As an alternative, manually draw one ROI that represents the tumor rim and a second ROI that represents the tumor core. The tumor rim can have greater vascularization than the tumor core, providing for a potentially useful sub-region analysis.
         NOTE: Although additional ROIs of tumor or wound sub-regions may be drawn, or the pixelwise values of HbO₂, Hb, HbT, or %sO₂ may be evaluated (requiring no ROI), these smaller regions have lower signal-to-noise, which leads to imprecise measurements.
      3. Maintain the ROI positions consistent across all images at each wavelength and time point.
   3. Calculate average parameters for the entire tumor.
      1. Calculate the average value of HbO₂, Hb, HbT, and %sO₂ in the tumor ROI or wound ROI for each imaging slice and each time point.
      2. For multi-slice OS OAI, calculate the average value of HbO₂, Hb, HbT, and %sO₂ from all imaging slices to obtain a single average value for each parameter for the entire tumor or wound.
   4. Evaluate parameter stability. Plot the value of %sO₂ for each repetition (see step 3.2.4). Evaluate whether the %sO₂ value changed by more than 10% during the scans, which indicates a temporally unstable measurement.
3. Analyze DCE OA images.
   1. Unmix chromophores in multiwavelength DCE OA images: Perform spectral unmixing to generate parametric maps of HbO₂, Hb, and the contrast agent. This step is not needed for single-wavelength DCE OAI.
   2. Define ROIs.
      1. Manually draw an ROI that represents the tumor or wound in each image slice. Draw a conservative ROI that errs on the side of excluding pixels of the tumor periphery or wound periphery rather than including pixels for normal tissue at the periphery.
      2. As an alternative, manually draw one ROI that represents the tumor rim and a second ROI that represents the tumor core. The tumor rim can have greater vascularization than the tumor core, providing for a potentially useful sub-region analysis.
         NOTE: Although additional ROIs of tumor sub-regions may be drawn, or the pixelwise values of contrast agent may be evaluated (requiring no ROI), these smaller regions have lower signal-to-noise, which leads to imprecise measurements.
      3. Maintain the ROI positions consistent across all images at each wavelength and time point.
   3. Extract PK curves from DCE OA images.
      1. For the ROI in each image slice, export the time points versus the average ICG signal (from multiwavelength DCE OAI) or the OAI signal at the single selected wavelength (for single-wavelength DCE OAI).
      2. Calculate the average signal for all image slices. A plot of the average signal throughout the tumor or wound vs. time is known as the PK curve.
      3. Transfer the average signal for all image slices at each time point to a computer program that can evaluate the empirical parameters listed below. For example, Microsoft Excel and Graphpad Prism can perform simple analyses of graphs.
   4. Compute empirical parameters as described in steps 4.3.5–4.3.11.
   5. Normalize the PK curve: Determine the average value of the pre-injection time points, known as the baseline signal. Subtract the baseline signal value from all post-injection signal values.
   6. Determine the Maximum Signal Enhancement (MSE): Identify the highest post-injection signal value in the PK curve.
   7. Determine the Time-to-Peak (TTP): Identify the time point value of the MSE.
   8. Calculate the initial slope of the PK curve.
      1. Obtain this initial slope from the time of initial injection to TTP. Alternatively, calculate the slope at 1 min post-injection if the shape of the curve is too steep or if the TTP is too late.
      2. In this scenario, acquire the pre-injection images for 1 min, and calculate this slope from Minute 1.0 to Minute 2.0 on the PK curve.
   9. Calculate the Initial Area Under the Curve (IAUC): Calculate the average value of the signal values from the time of initial injection to TTP on the PK curve.
   10. Calculate an empirical rate of the washout phase.
       1. For the portion of the PK curve from 1 min after TTP to the last time point, fit this portion of the curve with a straight line or a monoexponentially decaying function with a constant offset term.
       2. Use the slope of the straight line or the exponential decay constant as an empirical measure of the PK rate of the washout phase.
   11. Calculate contrast-enhanced OAI.
       1. Select a time point longer than TTP. Determine the signal value at this time point as a measure of contrast enhancement.
          NOTE: The time point at 10 min post-injection is often selected, but other time points may be selected as needed.
       2. As an alternative, calculate the average signal from TTP to the selected time point, the average signal from TTP to the last time point, or another range of time points as needed.
       3. Note that the average signal is equivalent to the area under the PK curve for these two time points, which is technically known as a measure of signal intensity rather than a measure of signal amplitude.
   12. Perform PK modeling as described in steps 4.3.13–4.3.20.
       NOTE: The following steps analyze DCE OA images acquired with an inVision OAI instrument. Other OAI instruments may produce images in a different data format. The steps below can still be followed to analyze DCE OA images using other OAI instruments if the data format from the other instruments can be transferred into MatLab and the data format is adjusted to match the example data files provided with the software package.
   13. Download the software.
       1. For multiwavelength DCE OAI, download OE-DCE_MSOT_Matlab from the CAMEL-MartyPagel repository on Github, or https://github.com/CAMEL-MartyPagel/OE-DCE_MSOT_Matlab.
       2. For single-wavelength DCE OAI, download DCE-MSOT_SingleWavelength_Matlab from the CAMEL-MartyPagel repository on Github, or https://github.com/CAMEL-MartyPagel/DCE_MSOT_SingleWavelength_Matlab.
       3. Unzip the compressed file. Open the SOP file in the doc folder and follow the standard operating procedure for installing and using the software.
   14. Organize the software components into different folders. Store the following software files in a single folder:
       Main code: run_MSOT_nonneg.m
       Sub-codes: run_DCE_MSOT_MSP.m ; run_OE_MSOT_AK_nonneg.m
       Other codes: Negative_pixel_count.m ; Voxel_count.m ; ROI_modification.m
       Folder titled: com.itheramedical.msotlib_beta_rev723_20221128_M2022b
       Data exported from ViewMSOT after reconstruction and spectral unmixing.
   15. Launch the program in MatLab.
       1. Double-click on run_MSOT_nonneg.m to start the program in Matlab. Click the "Run" button at the top of the Matlab window. The code may report an error that run_OE_MSOT_AK_nonneg is not found in the current folder or on the MATLAB path.
       2. To fix this error, type addpath then put the file location in quotes (e.g. addpath 'C:\Users\yourname\Documents\Non negative analysis') before running the code.
   16. Select the type of analysis: Enter N for the OE analysis, and Y for the DCE analysis.
       NOTE: The program is capable of analyzing Oxygen Enhanced (OE) OA images, which consist of a combination of %sO₂ maps acquired with 21% O₂ breathing gas (medical grade air) or 100% O₂ breathing gas, which is not used for OS-DCE OAI studies described in this protocol.
   17. Select the data:
       1. Select the folder containing the DCE OAI files. Select the main scan folder and not the MSP subfolder.
       2. Then verify the name of the scan. Finally, select the folder to save the analyzed data.
   18. Enter 3 for the number of chromophores. The chromophores are Hb, HbO_2, and the contrast agent.
       NOTE: The program is designed to accommodate the DCE OAI analysis of multiple chromophores that are co-injected into the mouse model, if needed. However, the DCE OAI protocol in this protocol only uses one chromophore.
   19. Draw the ROIs as described in steps 4.3.19.1–4.3.19.4.
       1. Draw the ROI of the blood vessel that will serve as the Arterial Input Function for the analysis: A grayscale image of the scan will appear. Manually trace the ROI that represents the blood vessel.
          NOTE: Zoom in by clicking on the magnifying glass with the plus sign just above the upper right-hand corner of the image. Click on the magnifying glass again to go back to the mode to draw the ROI. The inferior vena cava near the spine can often be identified and used as the AIF.
       2. Draw the ROI of the tumor: A grayscale image of the scan will appear. Trace the ROI that represents the entire tumor, the tumor rim, or the tumor core. To analyze more than one ROI, repeat this entire PK modeling procedure (Step 4.3.12) for an additional ROI.
       3. Draw the ROI of a muscle region: Do not use this ROI in the DCE OAI calculation. Use the ROI of the muscle for DCE OAI analysis using the Linear Reference Region Model, which is not used in this report. This step must still be performed for the program to continue.
       4. Verify that the correct ROI is drawn. The modeling of the PK curve will commence.
   20. Visualize the analyzed DCE OAI results.
       1. To load analyzed DCE data, go to the browser bar and select DCE_MSOT_MSP.
       2. In the command window, enter clear all.
          NOTE: The current folder tab is now populated with analyzed DCE OAI data, which includes:
          Average_signal_Wavelength_ICG.mat: Double click on this to view the tabulated values of NK^trans and k_ep for the tumor ROI. The kinetic parameters are tabulated in 3 ways:
          NK^trans_avg: Average of NK^trans calculated for each pixel.
          NK^trans _avg_signal: Single NK^trans value calculated after averaging the signals for the entire tumor
          NK^trans _nz: NK^trans_avg calculated with only non-zero pixel values.
          Similarly, k_ep values can be displayed.
          NKtrans.fig: Displays the NK^trans _avg map superimposed on a grayscale image.
          kep.fig: Displays the k_ep_avg map superimposed on a grayscale image.
          Maps_Wavelength_ICG.mat: This needs to be loaded to visualize other NK^trans or k_ep maps. After loading this matrix, use the "overlap code" (included separately) to display the desired parameter.
          Muscle_Average_signal_Wavelength_ICG.mat and Muscle_Maps_Wavelength_ICG.mat : Display parameters for muscle ROIs.
          Normalized signal for ICG.fig: Denotes the average signal of ICG over time in the three ROIs, normalized to the maximum signal.
          Signal for ICG.fig: Denotes the average signal of ICG over time in the three ROIs.

Results

OAI signal can generate maps of oxygen saturation in tumors and wounds
Following acquisition and reconstruction of multispectral OAI datasets, regions of interest (ROIs) can be defined within tumors or wound beds to quantify signal intensity at different wavelengths. Spectral unmixing algorithms are then applied to separate the contributions of oxyhemoglobin (HbO₂) and deoxyhemoglobin (Hb) from the composite optoacoustic signal. This processing enables the generation of quantitative maps of HbO₂, Hb, total hemoglobin (HbT), and blood oxygen saturation (%sO₂) (Figure 4).

The %sO₂ maps derived from OAI can be used to evaluate tissue oxygenation and hypoxia in both tumors and wounds. In tumors, regions of hypoxia are often observed in the interior of the tumor due to the abnormal and contorted blood vessels forming a poor vascular network, limiting oxygen delivery. The abnormal vessel architecture, combined with the high metabolic consumption of oxygen, results in a hypoxic region within the tumor. Hypoxic tumors are often malignant; therefore, OS OAI can provide valuable information on the classification of tumor malignancy. Immediately after creating the wound model, wound beds initially exhibit increased HbT and higher %sO₂ due to hyperemia and neovascularization during the healing process, which gradually decrease towards the levels of normal epithelial tissue as the wound heals. OS OAI can then be used to evaluate the ability of a wound to heal, potentially providing clinicians with insight into whether debridement of non-healing wound tissues is needed. Together, these OAI-derived maps provide a framework for longitudinal assessment of tumor physiology and wound healing dynamics.

DCE OAI can generate maps of vascular perfusion in tumors.
DCE OAI can provide maps of several empirical PK parameters, including the maximum signal enhancement (MSE), the time-to-peak (TTP) at MSE, the slope from injection time to the TTP, initial and total area under the curve (AUC), signal loss from MSE to the last timepoint, and contrast enhancement at 10 min (CE10) (Figure 5). These metrics can be interpreted to assess vascular perfusion and contrast retention within the tumor microenvironment. The total imaging duration and temporal sampling frequency can be adjusted according to experimental needs.

Alternatively, the OA image can be analyzed with PK modeling to estimate the kinetic rates of wash-in (NK^trans) and the washout (k_ep) (Figure 6). This modeling uses the dynamic signal of the arterial input function (AIF) relative to the tumor, providing a more rigorous analysis than empirical measurements. In particular, PK modeling is less susceptible to experimental conditions, such as the amount and rate of injection.

DCE OAI can generate maps of vascular perfusion in wounds.
The evaluation of vascular perfusion using DCE-OAI in wound models differs from that in tumor models. While tumor models exhibit moderately fast wash-in rates that can be estimated using PK models or empirically measured, wounds exhibit extremely fast wash-in rates that are too rapid for PK modeling or empirical measurements. Instead, empirical measurements of the washout rate from the wound bed can be evaluated from DCE OAI results. These evaluations often show a slower washout rate soon after the wound is created, indicating vascular damage that does not allow venous draining from the pooling of blood at the wound site (Figure 7). The washout rate increases during healing, indicating a successful revascularization process of the wound.

Figure 1: Establishing the 4T1 orthotopic tumor model. (A) Skin is lifted with tweezers, and then (B) a needle is inserted in the 4th mammary fat pad. (C) Tumors grew for ~10 days to reach 0.5-0.7 cm in diameter, which is acceptable for imaging. Please click here to view a larger version of this figure.

Figure 2: An excisional wound model. (A) A photograph of an anesthetized mouse and (B) an image of the wound area show that the wound is held open with a silicone splint that is sutured with material that does not absorb NIR light. Please click here to view a larger version of this figure.

Figure 3: Placing the mouse in the OAI instrument. (A) A catheter is inserted into the tail vein to provide venous access during imaging. (B) The cradle is prepared with ultrasound gel. (C) The mouse was placed onto the cradle. The tumor was set down onto the gel. (D) Ultrasound gel was spread over the region of interest, and the tumor was identified. (E) The animal was placed in the instrument and allowed to acclimate to the water bath for 10 min. Please click here to view a larger version of this figure.

Figure 4: OS OAI of a tumor model and wound healing model. (A) The mouse was positioned with the tumor or (B) wound in the lower center of the imaging field of view, which aided in drawing a region of interest (ROI) around the tumor. Parametric maps of the ROI were generated by spectral unmixing of OS OA images for deoxyhemoglobin (Hb), oxyhemoglobin (HbO2), total hemoglobin (HbT), and oxygen saturation (%sO2). These parametric maps of the tumor ROI were overlaid on an OA image with absorbance at 800 nm (the left-most images in each row). Please click here to view a larger version of this figure.

Figure 5: DCE OAI of a tumor model. (A) The OA image at 800 nm absorption wavelength shows a post-injection increase in signal in the tumor region of interest relative to a pre-injection image. This signal enhancement is attributed to the ICG contrast agent. The color bar (B) a pharmacokinetics curve from DCE OAI is labeled with the empirical parameters Maximum Signal Enhancement (MSE, blue), Time-to-Peak (TTP, green), slope from injection time to TTP (dashed brown line), initial area under the curve (IAUC, orange area), total area under the curve (AUC, red striped area that includes the IAUC), signal loss (pink) and contrast enhancement at 10 min post-injection (CE10, purple). Please click here to view a larger version of this figure.

Figure 6: Pharmacokinetics modeling of DCE OAI. (A) The OA image at 1 minute post-injection shows contrast enhancement from ICG. (B) A Region of Interest (ROI) is manually drawn on the image for the Arterial Input Function (AIF, a vessel), tumor, and muscle. The muscle ROI is not used for the modeling in this protocol. (C) Normalized pharmacokinetics curves from DCE OAI for the AIF (red) and tumor (blue). (D) The NK^trans map and E) k_ep map of the tumor. Please click here to view a larger version of this figure.

Figure 7: DCE OAI of a wound model. (A) The OA image at 800 nm absorption wavelength shows a post-injection increase in signal in the wound region of interest relative to a pre-injection image. This signal enhancement is attributed to the ICG contrast agent. (B) A pharmacokinetics curve from DCE OAI is labeled with the empirical parameters Maximum Signal Enhancement (MSE, blue), total area under the curve (AUC, red striped area), signal loss (pink), contrast enhancement at 10 minutes post-injection (CE10, purple), and an exponential decay constant that approximates the washout rate (k, red). Please click here to view a larger version of this figure.

Figure 8: Potential pitfalls of DCE OAI of a tumor model. (A) A tumor without an epithelial wound can produce (B) an OA image at 800 nm absorbance wavelength with good quality. (C) A tumor with an epithelial wound (highlighted with an arrow) can produce (D) an OA image with poor quality. (E) A slow injection leads to a pharmacokinetics curve with high quality. (F) A slow, failed, or premature injection can lead to poor-quality pharmacokinetic curves. Please click here to view a larger version of this figure.

Number of slices	Recommended	Total
	Repetitions	Acquisition Time
1	20	120 s
2	10	124 s
3	6	117.6 s
4	5	128 s
5	4	124 s
≥6	3	≥111.6 s

Table 1: Recommended number of averages, repetitions, and acquisition times for OS OAI.

Number of slices	Recommended Averages	Temporal Resolution	Recommended	Total
			Repetitions	Acquisition Time
1	50	5.0 s	132	11.0 min
2	23	5.0 s	132	11.0 min
3	14	4.8 s	140	11.2 s
4	10	4.8 s	140	11.2 s
5	8	5.0 s	132	11.0 min
6	6	4.8 s	140	11.2 s

Table 2: Recommended number of averages, repetitions, and acquisition times for DCE OAI.

Discussion

DCE OAI PK curves may be affected by a variety of artifacts that reduce the reliability of PK analysis. One common pitfall is the presence of superficial wounds, scabs, or dried blood within the imaging field, which can introduce strong, non-specific absorption at the skin surface and obscure deeper signals (Figure 8). When drawing ROIs within the tumor, care should be taken to avoid including these superficial structures to ensure that the extracted curves reflect intratumoral contrast dynamics rather than surface artifacts.

Vascular access and injection quality are also critical. Figure 8E shows representative PK curves that occur following failed or partial contrast agent injections. If the catheter does not maintain venous access, if the injection is extravasated, or if the bolus is not delivered in a relatively rapid, coherent manner, the expected sharp increase in signal may be absent or markedly blunted. Similarly, slow or inconsistent injections can lead to diminished early-phase slopes and broadened peaks, complicating the interpretation of wash-in kinetics. Establishing a stable signal baseline prior to injection is essential. If baseline imaging is not performed, then changes in signal may be difficult to interpret.

Additional sources of error include motion artifacts from respiration or abrupt animal movements under light anesthesia, and inconsistent ROI placement across time points²⁰. Motion artifacts can be mitigated by secure animal immobilization and increasing the number of average frames per repetition. If a failed injection or major artifact is suspected, the study should be repeated once the contrast agent has sufficiently cleared, typically 24–48 h later, while ensuring improved catheter placement and injection technique.

The clinical adoption of optoacoustic imaging is hampered by the stochastic scattering and absorbance of NIR light in tissues, leading to variable light fluence in tissues (where fluence is the light energy that reaches a specific voxel within the tissue). This problem limits OAI image acquisition to only ~4 cm in depth^21,22. While a ~4 cm depth of view is more than adequate for imaging mouse models, this depth of view is limiting for clinical OAI studies. Furthermore, the absorption of NIR light in the tissue yields wavelength-dependent fluence leading to spectral coloring, which can compromise spectral unmixing during the analysis²³. Algorithms can compensate for spectral coloring, although these algorithms require computation time and do not eliminate the source of spectral coloring²⁴. For this reason, we recommend relative measurements that use 2 or more OAI signals that are affected by spectral coloring in an equal or similar way, so that the relative comparison of signals cancels or reduces the effect of spectral coloring. The measurement of %sO₂ is a ratio of HbO₂ to HbT, and therefore, %sO₂ is less affected by spectral coloring. The measurement of PK rates, including NK^trans and k_ep from tumors and k washout rate from wounds, depends on the rate of change of signal rather than the absolute signal amplitude.

DCE OAI requires the use of strong NIR-absorbing contrast agents to generate OAI signals. Furthermore, a minimum concentration of approximately 1 µM of OAI contrast agent is needed to elicit a signal in a tissue voxel. ICG can meet these requirements^8,9,25. However, ICG is amphiphilic and can non-specifically bind to proteins in the plasma and interstitial spaces in tissues, slowing the wash-in and washout rates. Other agents that are hydrophilic may mitigate this problem^16,17. These other agents should also be small molecules that have good vascular permeability into tumor tissues and wound beds. This OS-DCE OAI protocol can be useful for evaluations of new OAI contrast agents.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
20 G Needle	Exel INT	26417	Hypodermic needle; 20G x 1", used for preparing heparin solutions
4T1 cells	ATCC	CRL-2539	Triple negative breast cancer cell line
Anti-Anti (100x)	Gibco	15240-062	Antibiotic cocktail
Biopsy punch	Integra	33-36	6 mm disposable biopsy punch
Cell Culture Grade Water	Corning	25-055-CV	500 mL; tested to USP sterile sater for inection specifications
Collagen Matrix Solution	Corning	354262	Collagen matrix for tumor fomation and injection
Ethanol	Deacon Laboratory	2701	General purpose antiseptic
Eye Lube	Optixcare	50-218-8442	20 g/.70 oz; for animal use only
Gauze	Fisherbrand	22-415-468	Cotton gauze pad; 3" x 3" (7.6 x 7.6 cm); 12-Ply
Gloves	VWR	76457-130	Nitrile examination gloves
Heparin	Meitheal	NDC71288-402-10	10,000 USP units per 10 mL, 1000 USP units per mL; 10 mL Multi-Dose Vial
Indocyanine green	TCI	I0535	Optoacoustic Contrast Agent
Isoflurane	Isospire	17033-091-25	A nonflammable, nonexplosive inhalation anesthetic for use in horses and dogs; 250 mL
Laboratory Film	Parafilm	PM-992	2" x 250'; 1" diameter core
Nair body bream Hair remover	Nair	LL4038	Depiliatory Cream; rich cocoa butter & vitamin E
Needle	Terumo	NN2732R	27G x 1 1/4" (6.40 mm x 32 mm); R.W. Reg. Bevel
Optoacoustic Imaging Machine	iTheraMedical	MSOT inVision 512	Optoacoustic scanner
PBS, 1x	Corning	21-040-CV	500 mL; Phosphate-buffered saline without calcium and magnesium
RPMI	Corning	10-040-CV	Cell culture media
Saline	Hospira	NDC0409-4888-02	10 mL single dose; 0.9% sodium chloride injection, USP
Stopwatch	iTheraMedical	4135333	Used to record minutes and seconds
Syringes	Cardinal Health	1188128012	1 mL Insulin Syringe; U-100 Insulin only; 27G x 1/2" (0.35 x 12.7 mm)
Transparent Wound Dressing	Tegaderm	1624W	transparent film dressing frame style; 2 3/8" x 2 3/4" (6 cm x 7 cm)
Tweezers	Termo Electron	7626	Used to pull skin back when injecting cells into mice
Ultrasound gel	EcoVue	283	32 g packet, prpylene glycol free, water soluble, paraben and dye free
Undyed Sutures	Ethicon	VR493	18" cutting; P-3 3/8 13 mm; undyed and braided, synthetic absorbable suture, non-USP
Warming pad	Repti Therm UTH	120VAC 60Hz 8W	Used to warm animals while under anesthetic;
Would Splints	Grace Bio-Labs	476684	Clear Silicone, 14 mm OD x 10 mm ID 0.5 thickness with suture sites
Wound clips	Reflex 7	203-1000	Reflex skin closure system, 7 mm, stainless stee