Appearance and modeling of bubble artifacts in intracranial magnetic resonance-guided laser interstitial thermal therapy (MRg-LITT) temperature images

Proton resonance frequency- (PRF-) shift thermometry is used to monitor thermal therapies including magnetic resonance-guided laser interstitial thermal therapy (MRg-LITT) and MR-guided focused ultrasound (MRgFUS) [1,2]. These devices can ablate tissue in the brain and body [1,[3], [4], [5]]. PRF-shift thermometry uses gradient-recalled echo (GRE) image phase data to compute localized changes in tissue temperature in real time during the thermal therapy. Well-known sources of inaccuracies in PRF-shift thermometry include physiological (respiratory and cardiac) motion, bulk patient motion [[6], [7], [8], [9], [10]], spatial variations in the macroscopic magnetic field caused by heating of the tissue [11] and the presence of metallic ablative devices such as RF electrodes [12], as well as drift of the main magnetic field [13,14].

While the physics underpinning the above sources of inaccuracies are well-understood, clinical publications have recently reported artifacts that occur specifically during MRg-LITT heating in the brain [[15], [16], [17]]. These reports hypothesized that the artifacts may be caused by microscopic hemorrhages. These artifacts present as a dark void in the GRE magnitude images used in PRF-shift thermometry, and as a butterfly or bowtie pattern in the temperature maps displayed to the clinician [15] [18]. De Landro et al. [18] have recently demonstrated how laser heating can produce air bubbles in gelatin phantoms, which then produce dipole-like (bowtie) phase distortions in GRE temperature mapping images similar to those observed clinically, with regions of measured temperature appearing to cool rapidly below initial conditions, leading to temperature errors as large as 80 °C. Overall, there is an unresolved question of whether the artifacts observed clinically are caused by air bubbles or hemorrhages. Because phase data are directly used for calculating temperature change maps and, in some devices, thermal damage maps to inform clinicians on the extent of tissue damage, the nature of artifacts in phase data should be understood in order to appropriately inform clinicians.

Fig. 1 shows an example of such an artifact from a clinical MRg-LITT treatment. In this example, because the laser was actively operating up until the artifact was observed, the image magnitude (Fig. 1e) was somewhat attenuated due to heating but remained non-zero, and the phase map (Fig. 1f) contained a smooth heating-induced phase shift whose principal axis coincided with the laser fiber. In contrast, in the artifacted images there appeared voids with no signal at or adjacent to the heating site in the magnitude image (Fig. 1g), and the phase image (Fig. 1h) contained a bowtie distortion characteristic of magnetic field disturbances created by approximately spherical regions with magnetic susceptibility different from that of soft tissue. Notably, the artifact was also aligned with the scanner's B0 field, which is consistent with a difference in magnetic susceptibility.

The purpose of the present study is to assess whether these clinically observed temperature imaging artifacts during MRg-LITT are better explained by an air bubble or a hemorrhage formed in the tissue. We present a GRE image model of a spherical air bubble or hemorrhage centered in the imaged slice, which is based on the difference in magnetic susceptibility between air or blood and tissue. The model was compared to clinical data where the artifact was seen, to evaluate whether the artifacts were likely caused by a sphere filled with air (air bubble) versus blood (hemorrhage). The model was then used to demonstrate how the appearance of such susceptibility mismatches could cause over- and under-estimation of temperature data in sagittal/coronal, axial and oblique imaging orientations. Temperature and thermal damage maps were also reproduced with and without superimposed bubbles to illustrate how they may appear to clinicians.

留言 (0)

沒有登入
gif