Magnetic resonance thermometry in the target volume versus intraluminal probe thermometry for hyperthermia treatment monitoring

Abstract

Background and purpose: Hyperthermia, the elevation of target temperature to 39-44 °C, is monitored using temperature probes. However, these provide limited spatial information, sampling only a few discrete locations. Magnetic resonance (MR) thermometry currently offers an option for three-dimensional (3D) temperature monitoring during hyperthermia. This study compares and correlates temperatures measured by intraluminal probes with MR-based temperatures in (1) the anatomical region containing the intraluminal probes and (2) the hyperthermia target volume (HTV), located at a distance from the probes and representing the primary region of clinical interest.

Methods: Thirteen locally advanced cervical cancer (LACC) patients treated with radiotherapy and hyperthermia were included. Hyperthermia was monitored using intraluminal probes and MR thermometry. MR-based temperatures were compared to intraluminal probe temperatures. Repeated measures correlation was applied to correlate probe and MR-based temperatures in the HTV across all data and on a patient-specific basis.

Results: MR-based temperatures at probe locations showed good agreement with probe measurements (median absolute error ≤ 0.7 °C). In the HTV, MR-based temperatures deviated by a median absolute error of 0.5 °C from probe temperatures. Repeated measures correlations (r_rm) between MR and probe-based HTV temperatures ranged from 0.74 to 0.79 across all data and 0.64-0.96 on a patient-specific basis.

Conclusions: MR thermometry demonstrated promising performance for retrospective evaluation of temperature distributions in the HTV. While its current reliability for real-time treatment guidance remains limited, our results support further development towards broader clinical implementation in hyperthermia.

Keywords: Hyperthermia; Locally advanced cervical cancer; Magnetic resonance thermometry; Magnetic resonance-guided hyperthermia; Probe thermometry; Repeated measures correlation; Target volume temperature.

Fig. 1

Axial (left) and sagittal (middle) and coronal (right) views of the T1-weighted MR image. Fat reference tubes and subcutaneous fat are delineated in orange and yellow, respectively. The delineation of the HTV is shown in red, while the locations of the thermal probes are shown in magenta, yellow and blue, for bladder, rectal and vaginal probes, respectively. (HTV: hyperthermia target volume). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2

Comparison of probe and MR temperatures over the entire treatment duration, encompassing both the heating-up and steady-state phases. Data are presented separately for each probe location – bladder, rectum and vagina – as well as aggregated across all probe locations (“all probes”). (a) Comparison of median values for the entire patient cohort, showing temperatures measured by intraluminal probes (grey), MR thermometry at the probe locations (dark blue) and MR thermometry within the HTV (light blue). (b) Point-by-point comparison of median MR temperatures and median intraluminal probe temperatures at corresponding time points for each patient, , shown per probe location (grey) and for the HTV (light blue). (c) Comparison of median absolute errors between median MR temperatures and intraluminal probe temperatures, shown per probe location (grey) and for the HTV (grey). The temperature difference in the HTV is calculated relative to the median temperature measured by the probes across all locations for each time point. (MR: magnetic resonance; HTV: hyperthermia target volume; T_MR: temperature measured with MR; T_probe: temperature measured with intraluminal probes). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

Point-by-point comparison of MR temperatures and the corresponding intraluminal probe temperatures at the same time points for each patient and treatment over the entire treatment. (a) Difference between mean MR temperatures at the probe locations and mean probe temperatures. (b) Difference between mean MR temperatures in the HTV and mean probe temperatures. (MR: magnetic resonance; HTV: hyperthermia target volume; $\overline{\mathrm{T}}$_{MR probes}: mean MR temperature over the probe locations; $\overline{\mathrm{T}}$_probes: mean intraluminal probe temperatures over the probe locations; $\overline{\mathrm{T}}$_{MR HTV}: mean MR temperature in the HTV).

Fig. 4

Repeated measures correlation between vaginal probe temperatures and MR temperatures in the HTV, for each patient for the entire treatment duration. The lines represent the fit for each hyperthermia treatment session per patient. For patients 9, 10 and 13, who completed only one MRgHT session, standard correlation is displayed. For the remaining patients, who underwent multiple MRgHT sessions, repeated measures correlation is shown. Each plot includes patient number, number of treatments, correlation coefficient (r_rm or r), 95 % confidence intervals (CI) and p-values. (MR: magnetic resonance; HTV: hyperthermia target volume; n_treat: number of treatments; r_rm: repeated measures correlation; r: correlation coefficient; CI: 95% confidence interval; p: p-value).