Ultrasound simulation with deformable and patient-specific scatterer maps

Endobronchial Ultrasound Image Simulation for Image-Guided Bronchoscopy

BACKGROUND/OBJECTIVE

Accurate disease diagnosis and staging are essential for patients suspected of having lung cancer. The state-of-the-art minimally invasive tools used by physicians to perform these operations are bronchoscopy, for navigating the lung airways, and endobronchial ultrasound (EBUS), for localizing suspect extraluminal cancer lesions. While new image-guided systems enable accurate bronchoscope navigation close to a lesion, no means exists for guiding the final EBUS localization of an extraluminal lesion. We propose an EBUS simulation method to assist with EBUS localization.

METHODS

The method draws on a patient's chest computed-tomography (CT) scan to model the ultrasound signal propagation through the tissue media. The method, which is suitable for simulating EBUS images for both radial-probe and convex-probe EBUS devices, entails three steps: 1) image preprocessing, which generates a 2D CT equivalent of the EBUS scan plane; 2) EBUS scan-line computation, which models ultrasound transmission to map the CT plane into a preliminary simulated EBUS image; and 3) image post-processing, which increases realism by introducing simulated EBUS imaging effects and artifacts.

RESULTS

Results show that the method produces simulated EBUS images that strongly resemble images generated live by a real device and compares favorably to an existing ultrasound simulation method. It also produces images at a rate greater than real time (i.e., 53 frames/sec). We also demonstrate a successful integration of the method into an image-guided EBUS bronchoscopy system.

CONCLUSION/SIGNIFICANCE

The method is effective and practical for procedure planning/preview and follow-on live guidance of EBUS bronchoscopy.

Learning ultrasound rendering from cross-sectional model slices for simulated training

PURPOSE

Given the high level of expertise required for navigation and interpretation of ultrasound images, computational simulations can facilitate the training of such skills in virtual reality. With ray-tracing based simulations, realistic ultrasound images can be generated. However, due to computational constraints for interactivity, image quality typically needs to be compromised.

METHODS

We propose herein to bypass any rendering and simulation process at interactive time, by conducting such simulations during a non-time-critical offline stage and then learning image translation from cross-sectional model slices to such simulated frames. We use a generative adversarial framework with a dedicated generator architecture and input feeding scheme, which both substantially improve image quality without increase in network parameters. Integral attenuation maps derived from cross-sectional model slices, texture-friendly strided convolutions, providing stochastic noise and input maps to intermediate layers in order to preserve locality are all shown herein to greatly facilitate such translation task.

RESULTS

Given several quality metrics, the proposed method with only tissue maps as input is shown to provide comparable or superior results to a state-of-the-art that uses additional images of low-quality ultrasound renderings. An extensive ablation study shows the need and benefits from the individual contributions utilized in this work, based on qualitative examples and quantitative ultrasound similarity metrics. To that end, a local histogram statistics based error metric is proposed and demonstrated for visualization of local dissimilarities between ultrasound images.

CONCLUSION

A deep-learning based direct transformation from interactive tissue slices to likeness of high quality renderings allow to obviate any complex rendering process in real-time, which could enable extremely realistic ultrasound simulations on consumer-hardware by moving the time-intensive processes to a one-time, offline, preprocessing data preparation stage that can be performed on dedicated high-end hardware.