Monitoring of interstitial laser thermotherapy with heat-sensitive colour subtraction magnetic resonance imaging: Calibration with absolute tissue temperature and correlation with predicted lesion size
Swt. Gould et al., Monitoring of interstitial laser thermotherapy with heat-sensitive colour subtraction magnetic resonance imaging: Calibration with absolute tissue temperature and correlation with predicted lesion size, LASER MED S, 14(4), 1999, pp. 250-256
Magnetic resonance imaging (MR) is the most sensitive modality for monitori
ng interstitial thermotherapy (ILT). A real-time pulse sequence that assign
s a colour spectrum to grey-scale changes could potentially increase the ac
curacy of MR-guided thermal surgery. This study aimed to calibrate this seq
uence with tissue temperature and then to determine whether it could be use
d to predict accurately the extent of tissue necrosis produced during the f
ormation of a thermal lesion.
Porcine livers were studied within a 0.5T Interventional MR Unit. A Nd:YAG
laser fibre (lambda=1064 nm) with a diffuser tip was placed within the live
r parenchyma adjacent to an MR compatible thermocouple. A template sagittal
MR image containing the fibre tip was obtained. A 3 cm region of interest
(ROI) was centred on the fibre. Thermal lesions were produced (5 W for 20 m
in) with real-time subtraction MR monitoring with colour overlay throughout
(acquisition time 4 s). At 60 s intervals the pixel intensity value, tempe
rature and colour at the laser tip were noted. Twenty burns were produced.
Pixel intensity measurements were expressed as percentages of mean pixel in
tensity within the ROI to standardise measurements. Using the colour repres
enting the temperature above which tissue necrosis would be expected to occ
ur, predicted maximum lesion size was measured from the images and compared
with histological assessment.
There was a linear relationship between temperature and percentage pixel ch
ange (r(2) = 0.84). Six discrete colours were determined, all significantly
different from each other in terms of mean percentage pixel change (p<0.01
) and mean temperature (p<0.01 except between orange and yellow, p = 0.037)
. Green had a mean temperature of 55.6 (+/- 5)degrees C, and thus predicted
necrosis. Image-predicted maximum lesion size correlated closely with hist
ology (r(2)=0.93).
The colour changes produced by this unique pulse sequence have been calibra
ted with tissue temperature in vitro. The green colour represents the tempe
rature above which necrosis would be expected to occur and can be used to a
ccurately predict lesion size. This will potentially allow greater accuracy
and safety for MR monitoring of ILT in vivo.