A probabilistic thermal dose model for the estimation of necrosis in MR-guided tumor ablations.

BACKGROUND: Monitoring minimally invasive thermo ablation procedures using magnetic resonance (MR) thermometry allows therapy of tumors even close to critical anatomical structures. Unfortunately, intraoperative monitoring remains challenging due to the necessary accuracy and real-time capability. One reason for this is the statistical error introduced by MR measurement, which causes the prediction of ablation zones to become inaccurate. PURPOSE: In this work, we derive a probabilistic model for the prediction of ablation zones during thermal ablation procedures based on the thermal damage model CEM43 . By integrating the statistical error caused by MR measurement into the conventional prediction, we hope to reduce the amount of falsely classified voxels. METHODS: The probabilistic CEM43 model is empirically evaluated using a polyacrilamide gel phantom and three in-vivo pig livers. RESULTS: The results show a higher accuracy in three out of four data sets, with a relative difference in Sørensen-Dice coefficient from - 3.04 % $-3.04\%$ to 3.97% compared to the conventional model. Furthermore, the ablation zones predicted by the probabilistic model show a false positive rate with a relative decrease of 11.89%-30.04% compared to the conventional model. CONCLUSION: The presented probabilistic thermal dose model might help to prevent false classification of voxels within ablation zones. This could potentially result in an increased success rate for MR-guided thermal ablation procedures. Future work may address additional error sources and a follow-up study in a more realistic clinical context.

Adaptive simulation of 3D thermometry maps for interventional MR-guided tumor ablation using Pennes' bioheat equation and isotherms.

Minimally-invasive thermal ablation procedures have become clinically accepted treatment options for tumors and metastases. Continuous and reliable monitoring of volumetric heat distribution promises to be an important condition for successful outcomes. In this work, an adaptive bioheat transfer simulation of 3D thermometry maps is presented. Pennes' equation model is updated according to temperature maps generated by uniformly distributed 2D MR phase images rotated around the main axis of the applicator. The volumetric heat diffusion and the resulting shape of the ablation zone can be modelled accurately without introducing a specific heat source term. Filtering the temperature maps by extracting isotherms reduces artefacts and noise, compresses information of the measured data and adds physical a priori knowledge. The inverse heat transfer for estimating values of the simulated tissue and heating parameters is done by reducing the sum squared error between these isotherms and the 3D simulation. The approach is evaluated on data sets consisting of 13 ex vivo bio protein phantoms, including six perfusion phantoms with simulated heat sink effects. Results show an overall average Dice score of 0.89 ± 0.04 (SEM < 0.01). The optimization of the parameters takes 1.05 ± 0.26 s for each acquired image. Future steps should consider the local optimization of the simulation parameters instead of a global one to better detect heat sinks without a priori knowledge. In addition, the use of a proper Kalman filter might increase robustness and accuracy if combined with our method.

Comparison study of reconstruction algorithms for volumetric necrosis maps from 2D multi-slice GRE thermometry images.

Cancer is a disease which requires a significant amount of careful medical attention. For minimally-invasive thermal ablation procedures, the monitoring of heat distribution is one of the biggest challenges. In this work, three approaches for volumetric heat map reconstruction (Delauney triangulation, minimum volume enclosing ellipsoids (MVEE) and splines) are presented based on uniformly distributed 2D MRI phase images rotated around the applicator's main axis. We compare them with our previous temperature interpolation method with respect to accuracy, robustness and adaptability. All approaches are evaluated during MWA treatment on the same data sets consisting of 13 ex vivo bio protein phantoms, including six phantoms with simulated heat sink effects. Regarding accuracy, the DSC similarity results show a strong trend towards the MVEE ([Formula: see text]) and the splines ([Formula: see text]) method compared to the Delauney triangulation ([Formula: see text]) or the temperature interpolation ([Formula: see text]). Robustness is increased for all three approaches and the adaptability shows a significant trend towards the initial interpolation method and the splines. To overcome local inhomogeneities in the acquired data, the use of adaptive simulations should be considered in the future. In addition, the transfer to in vivo animal experiments should be considered to test for clinical applicability.

A deep learning approach for 2D ultrasound and 3D CT/MR image registration in liver tumor ablation.

BACKGROUND AND OBJECTIVE: Liver tumor ablation is often guided by ultrasound (US). Due to poor image quality, intraoperative US is fused with preoperative computed tomography or magnetic tomography (CT/MR) images to provide visual guidance. As of today, the underlying 2D US to 3D CT/MR registration problem remains a very challenging task. METHODS: We propose a novel pipeline to address this registration problem. Contrary to previous work, we do not formulate the problem as a regression task, which - for the given registration problem - achieves a low performance regarding accuracy and robustness due to the limited US soft-tissue contrast and the inter-patient variability on liver vessels. Instead, we first estimate the US probe angle roughly by using a classification network. Given this coarse initialization, we then improve the registration by formulation of the problem as a segmentation task, estimating the US plane in the 3D CT/MR through segmentation. RESULTS: We benchmark our approach on 1035 clinical images from 52 patients, yielding average registration errors of 11.6° and 4.7 mm, which outperforms the state of the art SVR method[1]. CONCLUSION: Our results show the efficiency of the proposed registration pipeline, which has potential to improve the robustness and accuracy of intraoperative patient registration.