2.1.

Interstitial Laser Irradiation

An 805-nm diode laser (Delta-30, AngioDynamics, Queensbury, New York) was used to deliver near-infrared light for all tissues. Continuous laser power was delivered through a 7-m optical fiber with an active cylindrical tip (Pioneer Optics, Windsor Locks, Connecticut). Active tips of 0.5 and 1.0 cm were used. To verify the procedures, bovine liver was used for temperature measurements during laser irradiation with the 1.0-cm active tip, which allowed a more comprehensive temperature assessment (more image slices along the active tip) and an easier slice selection process. In the rat experiments the 0.5-cm active tip was used due to the physical constraint of the small tumor size. Figure 1a shows the configurations of the long and short active tips in tissue. Prior to all laser irradiation, actual power outputs for various power settings were determined using a spherical power meter (Ophir-Spiricon, Logan, Utah).

Fig. 1

(a) Schematics of the configuration for 1- and 0.5-cm active tips in tissue and the MRI slice selection. Ten 1-mm transverse slices were selected along the laser fiber (z-axis) for image acquisition. The 1 and 0.5-cm active tips are used for the ex vivo bovine liver and in vivo rat experiments, respectively. (b) Schematic of the active tip placed inside a rat tumor, protected by a sheath. (c) Irradiation of a rat tumor using an interstitial fiber.

2.2.

Tumor Preparation

Wistar Furth female rats were injected with 10⁵ DMBA-4 metastatic mammary tumor cells on their upper backs. Interstitial laser irradiation was applied when the tumors grew to a size of 1 to 1.5 cm in diameter. For the PRF experiment, the rat was allowed to rest in a prone position in an acrylic tube placed inside the bore of the magnetic resonance imager. In addition, gating sensors were attached to the rat's body. A dedicated computer in the control room was used to monitor the rat's respiration using a respiration pillow and SAM PC gating software (SAI Instruments, Manchester, England). During treatment, anesthesia (1.5% 2% Isofluorane, 800 to 1000 ml/min O₂) was administered to keep a stable respiration rate (∼35 breaths/min).

2.3.

Experimental Setup for PRF

For bovine liver, the active tip was inserted directly into a preselected location. For animal tumors, a small needle (20 gauge) was used to make the preliminary incision into the tumor, followed by a larger needle (14 gauge). The 0.5-cm cylindrical active tip was inserted into the center of the tumor guided by the preinserted 14 gauge needle. A transparent plastic sheath was used to protect the laser fiber. The position of the fiber and the rat experiment setup are shown in Figs. 1b, 1c. The laser power and irradiation duration were as follows: 1 W for 10 min, 1.25 W for 10 min, or 1.5 W for 10 min.

2.4.

Experimental Setup for Thermocouple Measurements

For the thermocouple measurement, each rat was placed in a prone position during laser irradiation. Experiments were conducted using the 0.5-cm active tip with a laser power of 1 or 1.5 W for 10 min. Before the experiment, the thermocouples were calibrated on InstruNET software (Omega Engineering, Stamford, Connecticut) using freezing and boiling baths at 0°C and 98°C, respectively. To acquire the temperature data, the thermocouple needles were first sterilized and then inserted into the target tissue at varying distances from the center of the cylindrical active tip. Two thermocouple needle probes were placed 2 and 5 mm away from the laser fiber.

2.5.

Calibrations and Verification

To calibrate the equipment and to verify the PRF method, ex vivo bovine liver tissue was cooled to 0°C overnight in an ice water bath and then brought back to 4°C before MRI imaging. Different points in the tissue were chosen for photothermal assessment with respect to the active tip. PRF was used to measure bovine liver tissue temperature profiles during a 10 min laser irradiation with 1.5 W using the 1-cm active tip. A similar experiment was performed on bovine liver tissue with temperature measurement using thermocouples with needle probes inserted 2, 3, and 5 mm from the active tip.

2.6.

PRF Method

PRF temperature dependency was first investigated by Hindman.³³ However, the application of PRF for temperature mapping was first proposed by Ishihara ³⁷ The PRF temperature mapping theory describes how the relationship between the external magnetic field and the local magnetic field are linked by the temperature dependent shielding factor. As described by Hindman and Ishihara, the shielding factor of nuclei, which expresses the relationship between the external and local magnetic fields, is sensitive to temperature. Hence, in PRF experiments the temperature is measured via changes in signal phase which result from variations in the shielding constant, according to:

PRF was used to evaluate temperature profiles during laser irradiation. All experiments were performed using a 7 T Bruker 730 USR horizontal-bore MR imaging system equipped with a 15-cm diameter quadrature volume coil and controlled by Para Vision 5.0 Software (Bruker BioSpin MRI GmbH, Germany). A fast low angle shot (FLASH) gradient-echo pulse sequence was used for optimal temporal resolution and reliability.^{38, 39, 40} The echo time (TE) and repetition time were set at 4 and 200 ms, respectively. Ten 1-mm slices were partitioned along the length of the active cylindrical tip, as shown in Fig. 1a. For the best temporal resolution, the number of averages was set to 1. The flip angle was set at 30 deg based on the temporal resolution priority. To minimize motion artifacts in the rat experiments, the scans were gated to the respiration of the animal. Consequently, the temporal resolution can only be approximated by an average with uncertainty depending on the steady state of the rat's respiration. The data acquisitions were obtained during the longest steady state, which was the inhalation period. The approximate temporal resolution was 1 min/cycle. The actual average temporal resolution was determined by recording the final time elapsed and dividing it by the number of cycles. An abnormal breathing rate could prolong scans and affect the overall reliability of the temporal resolution measurement. The field of view (FOV) was set to 50×50 mm, with 128×128 pixels, providing a spatial resolution of 2.56 pixels/mm or 2 mm for every 5 pixels. The system was configured to acquire between 20 to 25 cycles with an estimated time of 20 to 25 min. MRI imaging was performed 5 min before and 5 min after laser irradiation. Prior to each data acquisition, high-resolution (256×256) magnitude scans were acquired to locate the laser fiber.

A similar MRI configuration was used for the bovine liver experiment. Ten 1-mm slices were partitioned along the length of the 1-cm active cylindrical tip, as shown in Fig. 1a. The FOV was set to 60×23 mm, providing a spatial resolution of 2.13 and 5.57 pixels/mm along the horizontal and vertical axes, respectively. The total acquisition cycle was 44, with a total time of 18 min. Without gating, the temporal resolution for the liver experiment was 25.7 s/cycle.

2.7.

Image Processing

Using the real (Re_n) and imaginary (Im_n) parts of the acquired imaging data, the phase difference (Δϕ) between the phase at frame number n (ϕ_n) and the initial phase (ϕ_o) was calculated using the following equation,

Subsequently, the temperature elevation was obtained from the phase-temperature relation given in Eq. 1. Temperature maps were calculated using MATLAB 2008 version 7.6 (The MathWorks Inc, Boston, Massachusetts). A 9-pixel region of interest (ROI) was sampled and averaged over time in close proximity to the laser fiber (1, 3, 5, or 10 mm). The laser fiber location was determined using the magnitude map. However, these distances are under the physical constraint of the target tissue.

3.1.

Bovine Liver Temperature Profiles

To obtain temperature maps in bovine liver, 10 transverse MR thermographic isoplots of tissue temperature were acquired along the 1-cm active tip. Two-dimensional (2D) temperature distributions in the three middle slices of the liver tissue are given in Fig. 2a. Taking advantage of the 2D temperature profile, temperature data points on the transverse cross-section of the fifth slice along the x and y directions were extracted and analyzed, as shown in Figs. 2b, 2c. Sharp temperature gradients were observed in close proximity to the active tip (10°C over just 1 mm of displacement). To test the reliability of the PRF measurement, thermocouples were used in a separate experiment. Choosing a specific transverse plane, temperature elevations at different locations measured by the two different methods were compared in Figs. 3a, 3b.

Fig. 2

Bovine liver 1.5 W PRF spatial assessment. (a) Two-dimensional temperature distributions in liver tissue immediately after interstitial laser irradiation (1.5 W for 10 min) at different locations: Slice 6 (left panel), slice 5 (middle panel), and slice 4 (right panel). See Fig. 1a for the selection of slices with the 1-cm active tip. Liver tissue temperature elevations along the x-axis (b) and y-axis (c) on the transverse plane of the fifth slice, immediately after the interstitial laser irradiation (1.5 W for 10 min). A large temperature gradient is observed in close vicinity to the active tip.

Fig. 3

Bovine liver 1.5 W thermocouple and PRF temporal assessments. Liver tissue temperature elevations at different locations during interstitial laser irradiation (1.5 W for 10 min), measured by thermocouple (a) and PRF (b). The transverse plane of these measurements was the fifth slice. See Fig. 1a for the selection of slices with the 1-cm active tip. The transverse distance from the active tip ranges from 2 to 5 mm for thermocouple and 1 to 10 mm for PRF.

Due to the difficulties in determining the exact location of the thermocouple in tissue, the two temperature profiles in Figs. 3a, 3b do not match exactly. However, the range and the trend of the temperature elevation indicate that PRF is a reliable method to allow a comprehensive assessment of the nuanced spatial temperature elevation throughout the target tissue.

3.2.

Temperature Measurement During Laser Treatment of Rat Tumors Using PRF

Rat tumors were treated with interstitial laser irradiation under a power of 1.25 W and a duration of 10 min. The active tip entry point can be clearly seen in the corresponding magnitude profile, as shown in Fig. 4a, left panel. The temperature increases within a specific transverse plane are also shown in Fig. 4a, right panel. The peak temperature elevation, 1 mm away from the active tip, was about 25°C.

Fig. 4

Rat 1.25 W PRF spatial assessment. (a) Two-dimensional map of temperature elevation in a rat immediately after interstitial laser treatment (1.25 W for 10 min). A magnitude image (left panel) was used to help locate the active tip's entry point for alignment purposes. See Fig. 1a for the selection of slices with the 0.5-cm active tip. The corresponding temperature increases on the fifth slice (right panel), indicated that the temperature elevation peaked at 25°C at a location 1 mm away from the active tip. (b) Two-dimensional temperature distribution of a rat tumor using five different transverse slices, immediately after interstitial laser irradiation (1.25 W and 10 min). The end of the active tip can be seen in the fifth slice. Higher temperature elevations are correlated with increasing slice number, which is consistent with our liver results. Each temperature distribution map (left column) is shown with its corresponding magnitude image (right column).

A set of images, representing various cross-sections along the active tip and providing a 3D view of the temperature distribution inside the tumor, is shown in Fig. 4b. The temperature reached its peak near the center of the active tip (slice 7). A closer look at points 1 and 3 mm from the active tip of the fifth slice revealed that the temperature elevation fluctuates over time, as shown in Fig. 5. The rate of temperature elevation reaching a steady state was lower than that of the bovine liver.

Fig. 5

Rat 1.25 W PRF temporal assessment. Temperature elevation in a rat tumor during interstitial laser treatment measured at positions 1 and 3 mm from the active tip using the fifth slice. See Fig. 1a for the selection of slices with the 0.5-cm active tip. The rate of temperature elevation reaching a steady state is slower compared with that of the liver tissue, due to different tissue thermal insulation factors such as fat content and skin thickness. The data fluctuations are partly due to the distorted temporal resolution of the measurement caused by the slow respiration rate of the rat.

MR thermographic isoplots of tissue temperature were acquired for a rat tumor before and after irradiation with a laser power of 1.0 W and duration of 10 min, as shown in Fig. 6a. The temperature elevation reached 7°C and 8°C with a distance of 3 and 1 mm away from the active tip, respectively, as shown in Fig. 6b. The temperature elevation differed only by 2°C for a 2-mm displacement. This temperature gradient is smaller compared to that of higher laser powers.

Fig. 6

Rat 1.0 and 1.5 W PRF spatial and temporal assessments. Temperature profiles in a rat during interstitial laser treatment with a power of 1.0 W for 10 min. See Fig. 1a for the selection of slices with the 0.5-cm active tip. MR thermographic isoplots of tissue temperature were acquired from the fifth slice, before [(a), left panel] and immediately after [(a), right panel] irradiation. A closer examination (b) shows an increase of 7°C and 8°C at positions of 3 and 1 mm away from the active tip, respectively. These smaller temperature differences correspond to a more even distribution of temperature. Temperature profiles of a rat during interstitial laser treatment with a power of 1.5 W for 10 min. MR thermographic isoplots of tissue temperature were acquired from the fifth slice, before [(c), left panel] and immediately after [(c), right panel] irradiation. A closer examination (d) shows an increase of 8°C and 17°C at positions of 3 and 1 mm away from the active tip, respectively.

Likewise, MR thermographic isoplots of tissue temperature were acquired for a rat tumor before and after irradiation with a laser power of 1.5 W and a duration of 10 min, as shown in Figs. 6c, 6d. The temperature elevation reached 8°C and 17°C at 3 mm and 1 mm away from the active tip, respectively, as shown in Fig. 6d. With the power increased, the thermal damage due to temperature elevation also increased. These relatively larger temperature differences may allow a wider range of immunological responses.

3.3.

Temperature Measurement During Laser Treatment of Rat Tumors Using Thermocouples

Temperature increases in a rat tumor were also measured at distances of 2 and 5 mm from the active tip using thermocouples to validate the results of PRF. For a tissue irradiation of 10 min, the elevation of a rat tumor temperature ranged from 8°C to 11°C for 1 W and 8°C to 15°C for 1.5 W, as shown in Fig. 7.

Fig. 7

Rat 1.0 and 1.5 W thermocouple temporal assessments. Temperature increases in a rat tumor during interstitial laser irradiation of 1.0 W (a) and 1.5 W (b) for 10 min measured using thermocouples. The measurement distances are approximately 2 and 5 mm from the active tip. The maximum temperature elevations are 8°C and 11°C for 1.0 W (a) and 8°C and 15°C for 1.5 W (b). These measurements are close to the results obtained using PRF, as shown in Figs. 6b and 6d.

4.1.

Calibrations and Observations of Phase Drift

One of the advantages of PRF is its near independence from tissue composition. It has been reported by Bruno that the thermal coefficient of swine liver, kidney, and muscle is approximately 0.01 ppm/°C.³² This is an assumption we used for both the bovine liver and the rat tumor in the PRF measurement of tissue temperatures.

It has been documented that chemical drift observed in an MR system is due to instability of the magnetic field,^{41, 42} and such an artifact is amplified over time.⁴³ Nevertheless, accurate and reliable temperature profiles can be extracted in the 20 to 30 min range depending on factors such as TE and tissue susceptibility.³⁶ As reported by Olsrud, during the first 30 min of their experiments using a 1.5 T MRI scanner, the corresponding temperature due to phase drift measured was approximately 1°C. Following this argument, for a stronger magnetic field of 7 T, the temperature corresponding to the phase drift should be even less significant (<1°C). Efforts to cope with these drifts are still being investigated. Instances of such endeavors are preacquisition K-space correction and post-acquisition linear regression.⁴⁴ These methods suppress the misalignment of the echo in the K-space, which according to Grimault ⁴³ contributes significantly to time-dependent background phase variations.

4.2.

Motion Artifacts

An unstable respiration rate and the treatment location (near the heart or stomach) are the main contributing factors to motion artifacts in PRF measurement. Even though the acquisition is gated, motion artifacts are sometimes unavoidable, as seen in Fig. 5. Tumors near the stomach and heart are also susceptible to motion artifacts. In addition, gating interference due to the rat's erratic respiration can distort the temporal resolution. To enhance image clarity, a masking magnitude filter was implemented in our experiment.

4.3.

Phase Wrapping in PRF Method

Phase wrapping was due to phase overlaps. This issue was resolved with a simple MATLAB code using a threshold setting, as shown in the Appendix. The code took the original 3D temperature data set, and then applied the unwrapping threshold to each individual pixel to calculate the unwrapped temperature. This unwrapping method is based on the simple assumption that the temperature elevation cannot be less than a certain negative value, since we are applying heat to the tissue. Thus, whenever the condition is met, the data will be offset by adding twice the maximum temperature increase. The result from this method is then processed as a normal data set.

4.4.

Temperature Measurement Using Bovine Liver Tissue

High spatial and temporal resolutions are attained with FLASH image sequencing for ex vivo experiments using a bovine liver tissue. The temperature is uniformly distributed in an elliptical fashion, as shown in Fig. 2a, which is consistent with the results reported by Olsrud ³⁶ The results from PRF and thermocouple measurements for the liver tissue were within a reasonable range for comparison. For example, at positions of 2, 3, and 5 mm, changes of 43°C, 33°C, 20°C were observed, respectively, using thermocouples, as shown in Fig. 3a. At positions of 1, 3, and 5 mm away from the active tip, temperature changes of 50°C, 40°C, and 35°C were observed, respectively, using PRF, as shown in Fig. 3b. The differences between the results are partly attributed to the thermocouple positioning error.

4.5.

Thermocouple Measurement

The rat thermocouple experiments yielded reasonable results, which helped guide our evaluation of the PRF results. The results are comparable to that of the PRF experiments, as shown in Figs. 6b, 6d, 7. The differences observed are largely attributed to tissue composition and thermocouple positioning.

Interestingly, one may observe the large difference in thermal response between the liver and rat tumor tissues. Indeed, a similar power of 1.5 W was applied to liver and rat tumor tissues [Figs. 3b, 6d]. However, as the data indicated, the liver tissue temperature increase was roughly three-fold higher in comparison with that of the rat tumor. This is not completely unexpected, however, partly for the following reasons. First, these are two different types of tissue. The liver tissue is relatively homogeneous and heavily pigmented compared to the rat tumor tissue. Therefore, we can reasonably expect different photothermal properties such as conduction and scattering. Second, these experiments were conducted under different natural conditions. The ex vivo liver allowed more control but lacked the blood-perfusion present in the in vivo rat tumor experiment. In the in vivo experiments, the tissue is constantly perfused by fresh blood, and heated blood is constantly evacuated out of the irradiated region by cardiac-induced flow. Thus temperature elevation in the rat tumor occurred at a lower rate than what is shown in Fig. 3 for liver tissue.

4.6.

Thermal Effects of Interstitial Laser Irradiation

Numerous PRF thermal profiles made it possible to assess the photothermal effect for various slices along the active tip. With this method, we were able to detect the uneven distribution of laser power along the active tip, as shown in Figs. 2b, 2c. Abnormal temperature distributions were also observed; they were mainly due to diversity in tissue composition including lipids, necrosis, and fluids. These factors may also affect the rate of the temperature reaching a steady state, which may explain the effect seen in Fig. 5.

Variations in tissue compositions not only affected the PRF results, but also the temperature conductance, as shown in Figs. 5, 6. The thermal resistance was relatively high when 1.5 W was used, as evidenced by a major drop in temperature elevation just 2 mm from the laser source. In these cases, a higher power or longer laser irradiation duration was needed to overcome these physical barriers and to increase the likelihood of temperature saturation. Nevertheless, moderate laser power and irradiation duration may be the best solution in terms of biological and thermal interactions.

Although we ascribed the major anomalies such as temperature fluctuations and distortions from the PRF results to tissue homogeneity and motion artifacts, it is worth noting other potential factors inherent in this modality. According to Maswadi, with the increase in laser power and temperature elevation, the signal-to-noise ratio (SNR) decreases as the phase change deviation increases.⁴⁵ In terms of temperature resolution, for a temperature increase of 20°C, the resolution in 2D MRT decreases to 1.4°C (calculation based on a 2 T MRT). In the treatment of cancer, temperature elevation above 20°C is not necessary and may be detrimental to a total immune response as discussed earlier. However, with the exception of the liver results and artifact from Fig. 5 at the 5–6 min marks, we can reasonable assume (based on a higher SNR of 7.1 T MRT) that the temperature excursion error contribution is much less than 1.4°C. In addition, as reported by Peters, another potential source of error is due to the physical presence of the heat source within the target. In their investigation, there was an orientation-dependent effect on the measured temperature that was attributed to the magnetic susceptibility of tissue along the 1-cm diffuser.^{46, 47} This is an issue that we will address in future studies.

Tumor size was an important factor affecting temperature increase. Due to a finite spatial resolution, the tumor size cannot be too small for reliable assessment.

Laser immunotherapy using a noninvasive application mode may not be optimal due to the intervening skin. It was also observed that the internal tissue composition had a major effect on photothermal activity. Care must be taken especially when administering treatment to large tumors. Increasing the power by 0.5 W will assure a major temperature increase within a 1-mm radius of the ROI, but not necessarily for a 3-mm radius of a ROI. It is worth noting that ILIT does not need to completely destroy all tumor cells in order to stimulate an effective immune response. Therefore, a balance of laser power and irradiation duration should be considered when determining treatment protocols. Furthermore, a nonuniform temperature distribution in the target tissue may lead to better immunological stimulation since tumor cells injured or disturbed at different levels may provide different immune responses. Control of residual tumors and metastasis after laser immunotherapy may rely on all the components of host immune system. Indeed, further studies are needed to correlate the immunological responses with tissue temperature.