In vivo thermal ablation monitoring using ultrasound echo decorrelation imaging

Abstract

Previous work indicated that ultrasound echo decorrelation imaging can track and quantify changes in echo signals to predict thermal damage during in vitro radiofrequency ablation (RFA). In the in vivo studies reported here, the feasibility of using echo decorrelation imaging as a treatment monitoring tool was assessed. RFA was performed on normal swine liver (N = 5), and ultrasound ablation using image-ablate arrays was performed on rabbit liver implanted with VX2 tumors (N = 2). Echo decorrelation and integrated backscatter were computed from Hilbert transformed pulse-echo data acquired during RFA and ultrasound ablation treatments. Receiver operating characteristic (ROC) curves were employed to assess the ability of echo decorrelation imaging and integrated backscatter to predict ablation. Area under the ROC curves (AUROC) was determined for RFA and ultrasound ablation using echo decorrelation imaging. Ablation was predicted more accurately using echo decorrelation imaging (AUROC = 0.832 and 0.776 for RFA and ultrasound ablation, respectively) than using integrated backscatter (AUROC = 0.734 and 0.494).

Keywords: Bulk ultrasound ablation; Echo decorrelation imaging; Radiofrequency ablation; Therapy monitoring.

Fig. 1

Experimental configurations. (a) Radiofrequency ablation of porcine liver. The ultrasound probe is shown resting on the liver, with the thermocouple and RF probe inserted into the liver in the ultrasound image. (b) Ultrasound ablation of rabbit liver, with the probe assembly resting on the liver.

Fig. 2

Representative Pearson product moment correlation coefficient of the RF echo data within the ROI for treatment number 3.

Fig. 3

Hybrid echo decorrelation and integrated backscatter images for all RFA treatments. (a) Echo decorrelation at end of treatment. (b) Integrated backscatter at end of treatment. (c) Tissue sections corresponding to the ultrasound image plane, with the ablated region enclosed by dashed black lines. Results for treatment numbers 1 through 5 are shown in rows (1) through (5), respectively.

Fig. 4

Hybrid echo decorrelation images for RFA treatment 1. (a) After 40 s pretreatment. (b) After 20 s treatment. (c) After 25 s treatment. (d) After 30 s treatment. (e) After 35 s treatment. (f) After 40 s treatment.

Fig. 5

Echo decorrelation images for all ultrasound treatments. (a) End of pre-treatment imaging. (b) End of the ultrasound ablation treatment. (c) Tissue sections corresponding to the ultrasound image plane, with the ablated region enclosed by dashed black lines. Results for treatment numbers 4 and 6 are shown in rows (1) and (2), respectively.

Fig. 6

Integrated backscatter images for all ultrasound treatments. (a) End of pre-treatment imaging. (b) End of the ultrasound treatment. (c) Tissue sections corresponding to the ultrasound image plane, with the ablated region enclosed by dashed black lines. Results for treatment numbers 4 and 6 are shown in rows (1) and (2), respectively.

Fig. 7

Bar graphs showing the mean and standard deviation of cumulative echo decorrelation in ablated and unablated regions. (a) Decorrelation at the end of RFA treatment, with and without motion gating. (b) Decorrelation for ultrasound ablation, at the end of pre-treatment imaging and at the end of ultrasound ablation treatment.

Fig. 8

Receiver operating characteristic curves for accuracy of ablation prediction. (a) RFA. (b) Ultrasound ablation.

Fig. 9

Scatter plot of spatial maximum, temporal running-average echo decorrelation map vs. simultaneous thermocouple-measured tissue temperature for all time points from 5 RFA trials.