A learning robot for cognitive camera control in minimally invasive surgery.

BACKGROUND: We demonstrate the first self-learning, context-sensitive, autonomous camera-guiding robot applicable to minimally invasive surgery. The majority of surgical robots nowadays are telemanipulators without autonomous capabilities. Autonomous systems have been developed for laparoscopic camera guidance, however following simple rules and not adapting their behavior to specific tasks, procedures, or surgeons. METHODS: The herein presented methodology allows different robot kinematics to perceive their environment, interpret it according to a knowledge base and perform context-aware actions. For training, twenty operations were conducted with human camera guidance by a single surgeon. Subsequently, we experimentally evaluated the cognitive robotic camera control. A VIKY EP system and a KUKA LWR 4 robot were trained on data from manual camera guidance after completion of the surgeon's learning curve. Second, only data from VIKY EP were used to train the LWR and finally data from training with the LWR were used to re-train the LWR. RESULTS: The duration of each operation decreased with the robot's increasing experience from 1704 s ± 244 s to 1406 s ± 112 s, and 1197 s. Camera guidance quality (good/neutral/poor) improved from 38.6/53.4/7.9 to 49.4/46.3/4.1% and 56.2/41.0/2.8%. CONCLUSIONS: The cognitive camera robot improved its performance with experience, laying the foundation for a new generation of cognitive surgical robots that adapt to a surgeon's needs.

Robust near real-time estimation of physiological parameters from megapixel multispectral images with inverse Monte Carlo and random forest regression.

PURPOSE: Multispectral imaging can provide reflectance measurements at multiple spectral bands for each image pixel. These measurements can be used for estimation of important physiological parameters, such as oxygenation, which can provide indicators for the success of surgical treatment or the presence of abnormal tissue. The goal of this work was to develop a method to estimate physiological parameters in an accurate and rapid manner suited for modern high-resolution laparoscopic images. METHODS: While previous methods for oxygenation estimation are based on either simple linear methods or complex model-based approaches exclusively suited for off-line processing, we propose a new approach that combines the high accuracy of model-based approaches with the speed and robustness of modern machine learning methods. Our concept is based on training random forest regressors using reflectance spectra generated with Monte Carlo simulations. RESULTS: According to extensive in silico and in vivo experiments, the method features higher accuracy and robustness than state-of-the-art online methods and is orders of magnitude faster than other nonlinear regression based methods. CONCLUSION: Our current implementation allows for near real-time oxygenation estimation from megapixel multispectral images and is thus well suited for online tissue analysis.