[MR thermometry for laser-induced thermotherapy at 1.5 Tesla]

Evaluation of thermometry with fast MR sequences for laser-induced interstitial laser therapy (LITT) and verification of the thermometric results with a fiber-optic thermometer.

In vitro experiments were conducted using an agarose gel mixture and pig liver lobes. MR-guided LITT was performed using a laser power between 3 and 15 watts. Thermometry was performed using longitudinal relaxation time T1 and proton resonance frequency shift (PRF) methods under acquisition of amplitude and phase shift images. PRF was measured with a fast spoiled GRE sequence. Four different sequences were used for T1 thermometry: gradient echo (GE), TrueFISP (TRUFI), Saturation Recovery Turbo-FLASH (SRTF) and Inversion Recovery Turbo-FLASH (IRTF) sequences. The temperature was controlled using a fiber-optic Luxtron device and correlated with the MR temperature. The range of applied and monitored temperatures exceeded 80 degrees Celsius.

The temperature dependence showed a good linear relationship up to 60 degrees Celsius. Calibration experiments for the T1 method delivered coefficients of determination from 0.977 to 0.997 for agarose and from 0.958 to 0.995 for the pig liver samples. The IRTF sequence had the highest temperature sensitivity (agarose 0.99, liver 1.19). During LITT the TRUE-FISP sequence exhibited a strong nonlinear relationship. R (2) of this sequence was 0.809 in the agarose experiments. The average temperature errors when heated up to 80 degrees Celsius were 3.86-11.38 degrees Celsius for Agarose gel and 5.7-12.16 degrees Celsius for the liver tissue. SRTF and IRTF sequences exhibited the most linear relationship with temperature but were more dependent on tissue differences.

The accuracy of the temperature measurement is sufficient for controlling the coagulation area of the LITT. PRF is the method of choice since it shows the best linear correlation with fiber-optic temperature. If only T1 sequences are concerned, the FLASH sequence is preferred. It is the most robust, though not the most accurate, T1 sequence.