Intraoperative MRI-derived volumetric ablation margins and initial correlation with local outcome after MRI-guided cryoablation of renal tumors.

PURPOSE: To assess volumetric ablation margins derived from intraoperative pre- and post-ablation MRI after magnetic resonance imaging (MRI)-guided percutaneous cryoablation of renal tumors and explore its correlation with local treatment success. METHODS: Retrospective analysis was performed on 30 patients (mean age 69y) who underwent percutaneous MRI-guided cryoablation between May 2014 and May 2020 for 32 renal tumors (size: 1.6-5.1 cm). Tumor and ice-ball volumes were segmented on intraprocedural pre- and post-ablation MR images using Software Assistant for Interventional Radiology (SAFIR) software. After MRI-MRI co-registration, the software automatically quantified the minimal treatment margin (MTM),defined as the smallest 3D distance between the tumor and ice-ball surface. Local tumor progression (LTP) after cryoablation was assessed on follow-up imaging. RESULTS: Median follow-up was 16 months (range: 1-58). Local control after cryoablation was achieved in 26 cases (81%) while LTP occurred in 6 (19%). The intended MTM of ≥5 mm was achieved in 3/32 (9%) cases. Median MTM was significantly smaller for cases with (- 7 mm; IQR:-10 to - 5) vs. without LTP (3 mm; IQR:2 to 4) (P < .001). All cases of LTP had a negative MTM. All negative treatment margins occurred in tumors > 3 cm. CONCLUSIONS: Determination of volumetric ablation margins from intraoperative MRI was feasible and may be useful in predicting local outcome after MRI-guided renal cryoablation. In our preliminary data, an intraoperative MRI-derived minimal margin extending at least 1 mm beyond the MRI-visible tumor led to local control and this was more difficult to achieve in tumors > 3 cm. Ultimately, online margin analysis may be a valuable tool to intraoperatively assess therapy success, but larger prospective studies are needed to establish a reliable threshold for clinical use.

Learning-based autonomous vascular guidewire navigation without human demonstration in the venous system of a porcine liver.

PURPOSE: The navigation of endovascular guidewires is a dexterous task where physicians and patients can benefit from automation. Machine learning-based controllers are promising to help master this task. However, human-generated training data are scarce and resource-intensive to generate. We investigate if a neural network-based controller trained without human-generated data can learn human-like behaviors. METHODS: We trained and evaluated a neural network-based controller via deep reinforcement learning in a finite element simulation to navigate the venous system of a porcine liver without human-generated data. The behavior is compared to manual expert navigation, and real-world transferability is evaluated. RESULTS: The controller achieves a success rate of 100% in simulation. The controller applies a wiggling behavior, where the guidewire tip is continuously rotated alternately clockwise and counterclockwise like the human expert applies. In the ex vivo porcine liver, the success rate drops to 30%, because either the wrong branch is probed, or the guidewire becomes entangled. CONCLUSION: In this work, we prove that a learning-based controller is capable of learning human-like guidewire navigation behavior without human-generated data, therefore, mitigating the requirement to produce resource-intensive human-generated training data. Limitations are the restriction to one vessel geometry, the neglected safeness of navigation, and the reduced transferability to the real world.

Robotic Assistance System for Cone-Beam Computed Tomography-Guided Percutaneous Needle Placement.

PURPOSE: The study aimed to evaluate a new robotic assistance system (RAS) for needle placement in combination with a multi-axis C-arm angiography system for cone-beam computed tomography (CBCT) in a phantom setting. MATERIALS AND METHODS: The RAS consisted of a tool holder, dedicated planning software, and a mobile platform with a lightweight robotic arm to enable image-guided needle placement in conjunction with CBCT imaging. A CBCT scan of the phantom was performed to calibrate the robotic arm in the scan volume and to plan the different needle trajectories. The trajectory data were sent to the robot, which then positioned the tool holder along the trajectory. A 19G needle was then manually inserted into the phantom. During the control CBCT scan, the exact needle position was evaluated and any possible deviation from the target lesion measured. RESULTS: In total, 16 needle insertions targeting eight in- and out-of-plane sites were performed. Mean angular deviation from planned trajectory to actual needle trajectory was 1.12°. Mean deviation from target point and actual needle tip position was 2.74 mm, and mean deviation depth from the target lesion to the actual needle tip position was 2.14 mm. Mean time for needle placement was 361 s. Only differences in time required for needle placement between in- and out-of-plane trajectories (337 s vs. 380 s) were statistically significant (p = 0.0214). CONCLUSION: Using this RAS for image-guided percutaneous needle placement with CBCT was precise and efficient in the phantom setting.

Instrument localisation for endovascular aneurysm repair: Comparison of two methods based on tracking systems or using imaging.

BACKGROUND: In endovascular aneuysm repair (EVAR) procedures, medical instruments are currently navigated with a two-dimensional imaging based guidance requiring X-rays and contrast agent. METHODS: Novel approaches for obtaining the three-dimensional instrument positions are introduced. Firstly, a method based on fibre optical shape sensing, one electromagnetic sensor and a preoperative computed tomography (CT) scan is described. Secondly, an approach based on image processing using one 2D fluoroscopic image and a preoperative CT scan is introduced. RESULTS: For the tracking based method, average errors from 1.81 to 3.13 mm and maximum errors from 3.21 to 5.46 mm were measured. For the image-based approach, average errors from 3.07 to 6.02 mm and maximum errors from 8.05 to 15.75 mm were measured. CONCLUSION: The tracking based method is promising for usage in EVAR procedures. For the image-based approach are applications in smaller vessels more suitable, since its errors increase with the vessel diameter.

Evaluation of a numerical simulation for cryoablation - comparison with bench data, clinical kidney and lung cases.

PURPOSE: The accuracy of a numerical simulation of cryoablation ice balls was evaluated in gel phantom data as well as clinical kidney and lung cases. MATERIALS AND METHODS: To evaluate the accuracy, 64 experimental single-needle cryoablations and 12 multi-needle cryoablations in gel phantoms were re-simulated with the corresponding freeze-thaw-freeze cycles. The simulated temperatures were compared over time with the measurements of thermocouples. For single needles, temperature values were compared at each thermocouple location. For multiple needles, Euclidean distances between simulated and measured isotherms (10 °C, 0 °C, -20 °C, -40 °C) were computed. Furthermore, surface and volume of simulated 0 °C isotherms were compared to cryoablation-induced ice balls in 14 kidney and 13 lung patients. For this purpose, needle positions and relevant anatomical structures defining material parameters (kidney/lung, tumor) were reconstructed from pre-ablation CT images and fused with postablation CT images (from which ice balls were extracted by manual delineation). RESULTS: The single-needle gel phantom cases showed less than 5 °C prediction error on average. Over all multiple needle experiments in gel, the mean and maximum isotherm distance were less than 2.3 mm and 4.1 mm, respectively. Average Dice coefficients of 0.82/0.63 (kidney/lung) and mean surface distances of 2.59/3.12 mm quantify the prediction performance of the numerical simulation. However, maximum surface distances of 10.57/10.8 mm indicate that locally larger errors have to be expected. CONCLUSION: A very good agreement of the numerical simulations for gel experiments was measured and a satisfactory agreement of the numerical simulations with measured ice balls in patient data was shown.

Three-dimensional guidance including shape sensing of a stentgraft system for endovascular aneurysm repair.

PURPOSE: During endovascular aneurysm repair (EVAR) procedures, medical instruments are guided with two-dimensional (2D) fluoroscopy and conventional digital subtraction angiography. However, this requires X-ray exposure and contrast agent is used, and the depth information is missing. To overcome these drawbacks, a three-dimensional (3D) guidance approach based on tracking systems is introduced and evaluated. METHODS: A multicore fiber with fiber Bragg gratings for shape sensing and three electromagnetic (EM) sensors for locating the shape were integrated into a stentgraft system. A model for obtaining the located shape of the first 38 cm of the stentgraft system with two EM sensors is introduced and compared with a method based on three EM sensors. Both methods were evaluated with a vessel phantom containing a 3D-printed vessel made of silicone and agar-agar simulating the surrounding tissue. RESULTS: The evaluation of the guidance methods resulted in average errors from 1.35 to 2.43 mm and maximum errors from 3.04 to 6.30 mm using three EM sensors, and average errors from 1.57 to 2.64 mm and maximum errors from 2.79 to 6.27 mm using two EM sensors. Moreover, the videos made from the continuous measurements showed that a real-time guidance is possible with both approaches. CONCLUSION: The results showed that an accurate real-time guidance with two and three EM sensors is possible and that two EM sensors are already sufficient. Thus, the introduced 3D guidance method is promising to use it as navigation tool in EVAR procedures. Future work will focus on developing a method with less EM sensors and a detailed latency evaluation of the guidance method.

Fiber optical shape sensing of flexible instruments for endovascular navigation.

PURPOSE: Endovascular aortic repair procedures are currently conducted with 2D fluoroscopy imaging. Tracking systems based on fiber Bragg gratings are an emerging technology for the navigation of minimally invasive instruments which can reduce the X-ray exposure and the used contrast agent. Shape sensing of flexible structures is challenging and includes many calculations steps which are prone to different errors. To reduce this errors, we present an optimized shape sensing model. METHODS: We analyzed for every step of the shape sensing process, which errors can occur, how the error affects the shape and how it can be compensated or minimized. Experiments were done with one multicore fiber system with 38 cm sensing length, and the effects of different methods and parameters were analyzed. Furthermore, we compared 3D shape reconstructions with the segmented shape of the corresponding CT scans of the fiber to evaluate the accuracy of our optimized shape sensing model. Finally, we tested our model in a realistic endovascular scenario by using a 3D printed vessel system created from patient data. RESULTS: Depending on the complexity of the shape, we reached an average error of 0.35-1.15 mm and maximal error of 0.75-7.53 mm over the whole 38 cm sensing length. In the endovascular scenario, we obtained an average and maximal error of 1.13 mm and 2.11 mm, respectively. CONCLUSION: The accuracies of the 3D shape sensing model are promising, and we plan to combine the shape sensing based on fiber Bragg gratings with the position and orientation of an electromagnetic tracking to obtain the located catheter shape.

Segmentation of stochastic images with a stochastic random walker method.

We present an extension of the random walker segmentation to images with uncertain gray values. Such gray-value uncertainty may result from noise or other imaging artifacts or more general from measurement errors in the image acquisition process. The purpose is to quantify the influence of the gray-value uncertainty onto the result when using random walker segmentation. In random walker segmentation, a weighted graph is built from the image, where the edge weights depend on the image gradient between the pixels. For given seed regions, the probability is evaluated for a random walk on this graph starting at a pixel to end in one of the seed regions. Here, we extend this method to images with uncertain gray values. To this end, we consider the pixel values to be random variables (RVs), thus introducing the notion of stochastic images. We end up with stochastic weights for the graph in random walker segmentation and a stochastic partial differential equation (PDE) that has to be solved. We discretize the RVs and the stochastic PDE by the method of generalized polynomial chaos, combining the recent developments in numerical methods for the discretization of stochastic PDEs and an interactive segmentation algorithm. The resulting algorithm allows for the detection of regions where the segmentation result is highly influenced by the uncertain pixel values. Thus, it gives a reliability estimate for the resulting segmentation, and it furthermore allows determining the probability density function of the segmented object volume.

Radiofrequency ablation planning beyond simulation.

It is a challenging task to plan a radiofrequency (RF) ablation therapy to achieve the best outcome of the treatment and avoid recurrences at the same time. A patient specific simulation in advance that takes the cooling effect of blood vessels into account is a helpful tool for radiologists, but this needs a very high accuracy and thus high computational costs. In this work, we present various methods, which improve and extend the planning of an RF ablation procedure. First, we discuss two extensions of the simulation model to obtain a higher accuracy, including the vaporization of the water in the tissue and identifying the model parameters and to analyze their uncertainty. Furthermore, we discuss an extension of the planning procedure namely the optimization of the probe placement, which optimizes the overlap of the tumor area with the estimated coagulation in order to avoid recurrences. Since the optimization is constrained by the model, we have to take into account the uncertainties in the model parameters for the optimization as well. Finally, applications of our methods to a real RF ablation case are presented.