Liver tumors are increasingly treated with radioembolization. Here, we present first evidence of catheter design effect on particle-fluid dynamics and downstream branch targeting during microsphere administrations. A total of 7 experiments were performed in a bench-top model of the hepatic arterial vasculature with recreated hemodynamics. Fluorescent microspheres and clinically used holmium microspheres were administered with a standard microcatheter (SMC) and an anti-reflux catheter (ARC) positioned at the same level along the longitudinal vessel axis. Catheter-related particle flow dynamics were analyzed by reviewing video recordings of UV-light illuminated fluorescent microsphere administrations. Downstream branch distribution was analyzed by quantification of collected microspheres in separate filters for two first-order branches. Mean deviation from a perfectly homogenous distribution (DHD) was used to compare the distribution homogeneity between catheter types. The SMC administrations demonstrated a random off-centered catheter position (in 71 % of experiments), and a laminar particle flow pattern with an inhomogeneous downstream branch distribution, dependent on catheter position and injection force. The ARC administrations demonstrated a fixed centro-luminal catheter position, and a turbulent particle flow pattern with a more consistent and homogenous downstream branch distribution. Quantitative analyses confirmed a significantly more homogeneous distribution with the ARC; the mean DHD was 40.85 % (IQR 22.76 %) for the SMC and 15.54 % (IQR 6.46 %) for the ARC (p = 0.047). Catheter type has a significant impact on microsphere administrations in an in-vitro hepatic arterial model. A within-patient randomized controlled trial has been initiated to investigate clinical catheter-related effects during radioembolization treatment.

With the worldwide prevalence of cardiovascular diseases, much attention has been focused on simulating the characteristics of the human heart to better understand and treat cardiac disorders. The purpose of this study is to build a finite element model of the left atrium (LA) that incorporates detailed anatomical features and realistic material characteristics to investigate the interaction of heart tissue and surgical instruments. This model is used to facilitate the design of an endoscopically deployable atrial retractor for use in minimally invasive, robotically assisted mitral valve repair. Magnetic resonance imaging (MRI) scans of a pressurized explanted porcine heart were taken to provide a 3D solid model of the heart geometry, while uniaxial tensile tests of porcine left atrial tissue were conducted to obtain realistic material properties for noncontractile cardiac tissue. A finite element model of the LA was constructed using ANSYS ™ Release 9.0 software and the MRI data. The Mooney–Rivlin hyperelastic material model was chosen to characterize the passive left atrial tissue; material constants were derived from tensile test data. Finite element analysis (FEA) models of a CardioVations Port Access™ retractor and a prototype endoscopic retractor were constructed to simulate interaction between each instrument and the LA. These contact simulations were used to compare the quality of retraction between the two instruments and to optimize the design of the prototype retractor. Model accuracy was verified by comparing simulated cardiac wall deflections to those measured by MRI. FEA simulations revealed that peak forces of approximately $2.85N$ and $2.46N$ were required to retract the LA using the Port Access™ and prototype retractors, respectively. These forces varied nonlinearly with retractor blade displacement. Dilation of the atrial walls and rigid body motion of the chamber were approximately the same for both retractors. Finite element analysis is shown to be an effective tool for analyzing instrument/tissue interactions and for designing surgical instruments. The benefits of this approach to medical device design are significant when compared to the alternatives: constructing prototypes and evaluating them via animal or clinical trials.

Invasive aquatic plant species can expand rapidly throughout water bodies and cause severely adverse economic and ecological impacts. While mechanical, chemical, and biological methods exist for the identification and treatment of these invasive species, they are manually intensive, inefficient, costly, and can cause collateral ecological damage. To address current deficiencies in aquatic weed management, this paper details the development of a small fleet of fully autonomous boats capable of subsurface hydroacoustic imaging (to scan aquatic vegetation), machine learning (for automated weed identification), and herbicide deployment (for vegetation control). These capabilities aim to minimize manual labor and provide more efficient, safe (reduced chemical exposure to personnel), and timely weed management. Geotagged hydroacoustic imagery of three aquatic plant varieties (Hydrilla, Cabomba, and Coontail) was collected and used to create a software pipeline for subsurface aquatic weed classification and distribution mapping. Employing deep learning, the novel software achieved a classification accuracy of 99.06% after training.

Recent advances in deep learning, including the development of AlexNet, Residual Network (ResNet), and transfer learning, offer unprecedented classification accuracy in the field of machine vision. A developing application of deep learning is the automated identification and management of aquatic invasive plants. Classification of submersed aquatic vegetation (SAV) presents a unique challenge, namely, the lack of a single source of sensor data that can produce robust, interpretable images across a variable range of depth, turbidity, and lighting conditions. This paper focuses on the development of a multi-sensor (RGB and hydroacoustic) classification system for SAV that is robust to environmental conditions and combines the strengths of each sensing modality. The detection of invasive Hydrilla verticillata (hydrilla) is the primary goal. Over 5000 aerial RGB and hydroacoustic images were generated from two Florida lakes via an unmanned aerial vehicle and boat-mounted sonar unit, and tagged for neural network training and evaluation. Classes included “HYDR”, containing hydrilla; “NONE”, lacking SAV, and “OTHER”, containing SAV other than hydrilla. Using a transfer learning approach, deep neural networks with the ResNet architecture were individually trained on the RGB and hydroacoustic datasets. Multiple data fusion methodologies were evaluated to ensemble the outputs of these neural networks for optimal classification accuracy. A method incorporating logic and a Monte Carlo dropout approach yielded the best overall classification accuracy (84%), with recall and precision of 84.5% and 77.5%, respectively, for the hydrilla class. The training and ensembling approaches were repeated for a DenseNet model with identical training and testing datasets. The overall classification accuracy was similar between the ResNet and DenseNet models when averaged across all approaches (1.9% higher accuracy for the ResNet vs. the DenseNet).

The development of direct-drive spherical motors offers a potential solution to the limitations of conventional multiple degree-of-freedom (DOF) actuators, which typically utilize single-DOF joints (rotational and/or prismatic), arranged in series or parallel and powered by multiple single-DOF actuators. These configurations can be accompanied by kinematic singularities, backlash, limited power density and efficiency, and computationally expensive inverse kinematics. This paper details the design, fabrication and experimental testing of permanent magnet synchronous spherical motors (PMSSM) for multi-DOF servo-actuation. Its stator-pole arrangement is based on a Goldberg polyhedron, with each pole comprised of hexagonal or pentagonal inner and outer plates. The stator geometry and winding configurations are optimized using electromagnetic finite element analysis. A custom-made controller board includes a microcontroller, servo drivers, a wireless serial interface, and a USB PC interface. Angular orientation is sensed using an inertial measurement unit in wireless communication with the microcontroller. A PID controller is implemented and demonstrated for time-varying reference trajectories.