INTERSTITIAL LASER THERAPY
UNDER MAGNETIC RESONANCE IMAGING
Diagnostic Radiology Department, Cancer Institute of Milan, Italy.

INTRODUCTION
Interstitial laser therapy (ILT) is a new technique for destroying small tumours in which implanted optical fibers deliver energy, usually from a DIODE laser emitting at 980 nm. There is considerable interest in this technique since the use of imaging modalities was suggested for monitoring in real-time the therapeutic effectiveness during laser irradiation. Since energy deposited by laser radiation is converted into heat following optical absorption, monitoring the temperature increase could be used as an indication of thermal damage.
It as been shown that magnetic resonance imaging (MRI) can be used to measure changes in tissue temperature (1,2). Nevertheless, several limitations prevent extensive use of laser for interstitial surgery since, at present, the available MRI techniques do not allow prediction of the actual lesion volumes. The main difficulty is to visualize in real-time, size and geometry of thermal lesion. Although MR ultra fast imaging techniques are presently being investigated as a means of temperature mapping (3,4), there is still a lack of relationship between lesion image and temperature increase and/or tissue damage.
Our purpose was to use a fast imaging technique, of the order of tens of seconds, that   i)  was rapid enough to see the thermal changes induced by the DIODE laser while they occured, and  ii) had enough sensitivity and spatial resolution to allow comparison between the area of laser heating and that of histologically proven tissue damage.

MATERIALS AND METHODS
The beam from a continuous wave DIODE laser emitting at 980 nm was used to irradiate a bovine liver specimen. The laser has a maximum power output of 25 W and allows simultaneous delivery of light energy into four optical fibers. In the following experiments, the beam was focused  into one plastic-clad silica optical fiber 5 m long with a core diameter 600 um. The distal end of the fiber was plane-cut, with 5 mm of cladding and jacket removed to avoid melting. The laser output power was measured using a power meter before the insertion and after the removal of the fiber. Delivered power was 2 W. The fiber was inserted into the specimen through a 18 G plastic cannula.
MRI was performed with an Helmholtz oval receiving coil, 1,5 T (64 MHz) superconducting magnet (Siemens, MAGNETOM C1, Italy). FLASH sequence (4) with a 256 x 256 matrix, flip angle of 80 degrees and an echo time (TE) of 14 ms, was used for monitoring the heating. The pixel size (i.e., spatial resolution) was about 1 x 1 mm, slice thickness 5 mm. A FLASH sequence with a repetition time (TR) of 200 ms was made before heating. During heating, FLASH sequences with TR of 50 ms were performed at intervals of about 15-20 s. The acquisition time was limited to about 12 s. The advantage of this sequence is the possibility to utilize short TR that permits to maximize T1 contrast. According to Cohen (4), the signal intensity S produced in FLASH imaging is:

where  K   depends on whole magnetization, proton density and the characteristics of the machine, and is the RF flip angle.
It is well known that evaluation of T1 is possible following the acquisition of at least two images with different TR at fixed TE. This procedure is valid only if K and T1 do not vary during acquisition time. Thus, during laser irradiation the temperature dependence of T1 does not allow the application of the standard procedure for its calculation. This difficulty can be overcome by relating T1 at temperature K2 (T1(K2)) with the signal intensity and relaxation time T1 (K1) that is evaluated by standard procedure, for instance at room temperature K1. Considering an image acquired at temperature K1 with signal intensity S(K1) and an image acquired at temperature K2 with signal intensity S(K2), T1(K2) can be related to the signal intesities and to T1(K1) by the following expression:

Obviously, the two sequences must be acquired with the same , TE and TR.

RESULTS
Figure 1 shows a representative MR image of a liver specimen 15 min after the laser was turned on at 2 W. The optical fiber is in the plane of the image, and enters the sample from the bottom. Reduction of signal intensity during laser irradiation reflects heat propagation in the area surrounding the optical fiber tip.

In order to conduct a quantitative analysis of the MR images, the signal intensity values along the direction orthogonal  to the optical fiber tip have been evaluated. Figure 2 shows the bell-shaped curves representing decrease in signal intensity at four different times of irradiation. The signal continuously decreased as time of irradiation increased, showing an asymmetry with respect to the central axis and had a relatively great variability at the far ends. In addition, after 10 min of irradiation signal intensity reached a practically saturated minimum.
Figure 3 shows T1 values calculated as previously reported. The curves are symmetric, and the effect of the laser-induced heating is evident. T1 gradually increased from about 115 ms before the start of laser irradiation, to a maximum of about 335, 1300, 1740 and 2500 ms after 5, 10, 12 and 15 min laser irradiation, respectively.

Figure 4 shows a representative lesion induced by irradiation at 2 W for 15 min in a bovine liver specimen. Around the point of insertion of the fiber there is a band of tissue charring followed by an area of coagulative necrosis. The intensity of the tissue damage gradually declines in the most distant areas. The maximum extension of the coagulative necrosis measures 1,8 mm from the vaporized core of 1,7 mm diameter. Thus, the full diameter of the lesion is about 5,3 mm.

DISCUSSION
As previously reported (3,5), MRI must satisfy several requisites before ILT can be carried out in clinics on a safe basis. Briefly, imaging contrast should be strictly related to temperature changes, and should allow quantitative analysis of tissue damage; spatial resolution should be sufficient to determine the heated volume precisely; temporal resolution should be faster than heating diffusion time.
In this work we have shown that FLASH sequence imaging method is sensitive to temperature changes. In addition, temporal resolution allows measurements within short enough time intervals to permit continuous monitoring of thermal effect. Nevertheless, several findings deserve further considerations. Disomogenity in tissue composition (i.e., proton density) introduces disomogenity in signal intensity. Fluctuations in the contrast may be misleading in evaluating the treated volume, especially when they are close to the irradiated site. Imaging of T1, as calculated in our procedure, might greatly reduce those effects. Moreover, percentage variation of T1 during irradiation is greater than that shown by signal intensity, thus further improving quantitative analysis.
As regards to the relationship between lesion image acquired during irradiation and the actual induced lesion, our results suggest that FLASH sequence is able to predict the geometry of the treated volume. Histology of the sample irradiated for 15 min shows a lesion of about 5,3 mm in diameter. This value corresponds, looking at the figure 3 where data are referred to the same treated area shown in figure 4, to the diameter of a lesion image characterized by a T1 of about 350 ms. This relaxation time can be easly discriminated with respect to a background of 100-200 ms, and may be inferred as an indications of an induced lesion.
In conclusion, this work demonstrates that FLASH sequence satisfactorily images laser induced damage, and suggests a possible method to achieve dynamic control of ILT in real-time.

REFERENCES
1. Dickinson RJ, Hall AS, Hind AJ, Young IR. J Comput Assist Tomogr; 10: 468-472.
2. Jolesz FA, Bleier AR, Jakab P, et al. Radiology; 168: 249 - 253.
3. Bleier AR, Jolesz FA, Cohen MS, Weisskoff RM, et al. Magn Reson Med; 21: 132-137.
4. Cohen MS, Weisskoff RM, Magn Reson Imaging; 9: 1-37.
5. Tracz RA, Wyman DR, Little PB, et al. Lasers Surg Med; 12: 165-173.