Patient-specific placental vessel segmentation with limited data.

A major obstacle in applying machine learning for medical fields is the disparity between the data distribution of the training images and the data encountered in clinics. This phenomenon can be explained by inconsistent acquisition techniques and large variations across the patient spectrum. The result is poor translation of the trained models to the clinic, which limits their implementation in medical practice. Patient-specific trained networks could provide a potential solution. Although patient-specific approaches are usually infeasible because of the expenses associated with on-the-fly labeling, the use of generative adversarial networks enables this approach. This study proposes a patient-specific approach based on generative adversarial networks. In the presented training pipeline, the user trains a patient-specific segmentation network with extremely limited data which is supplemented with artificial samples generated by generative adversarial models. This approach is demonstrated in endoscopic video data captured during fetoscopic laser coagulation, a procedure used for treating twin-to-twin transfusion syndrome by ablating the placental blood vessels. Compared to a standard deep learning segmentation approach, the pipeline was able to achieve an intersection over union score of 0.60 using only 20 annotated images compared to 100 images using a standard approach. Furthermore, training with 20 annotated images without the use of the pipeline achieves an intersection over union score of 0.30, which, therefore, corresponds to a 100% increase in performance when incorporating the pipeline. A pipeline using GANs was used to generate artificial data which supplements the real data, this allows patient-specific training of a segmentation network. We show that artificial images generated using GANs significantly improve performance in vessel segmentation and that training patient-specific models can be a viable solution to bring automated vessel segmentation to the clinic.

A Human-Scale Clinically Ready Electromagnetic Navigation System for Magnetically Responsive Biomaterials and Medical Devices.

Magnetic navigation systems are used to precisely manipulate magnetically responsive materials enabling the realization of new minimally invasive procedures using magnetic medical devices. Their widespread applicability has been constrained by high infrastructure demands and costs. The study reports on a portable electromagnetic navigation system, the Navion, which is capable of generating a large magnetic field over a large workspace. The system is easy to install in hospital operating rooms and transportable through health care facilities, aiding in the widespread adoption of magnetically responsive medical devices. First, the design and implementation approach for the system are introduced and its performance is characterized. Next, in vitro navigation of different microrobot structures is demonstrated using magnetic field gradients and rotating magnetic fields. Spherical permanent magnets, electroplated cylindrical microrobots, microparticle swarms, and magnetic composite bacteria-inspired helical structures are investigated. The navigation of magnetic catheters is also demonstrated in two challenging endovascular tasks: 1) an angiography procedure and 2) deep navigation within the circle of Willis. Catheter navigation is demonstrated in a porcine model in vivo to perform an angiography under magnetic guidance.

CANDYBOTS: A New Generation of 3D-Printed Sugar-Based Transient Small-Scale Robots.

Sugars are ubiquitous in food, and are among the main sources of energy for almost all forms of life. Sugars can also form structural building blocks such as cellulose in plants. Because of their inherent degradability and biocompatibility characteristics, sugars are compelling materials for transient devices. Here, an additive manufacturing approach for the production of magnetic sugar-based composites is introduced. First, it is shown that sugar-based 3D architectures can be 3D printed by selective laser sintering. This method enables not only the caramelization chemistry but also the mechanical properties of the sugar architectures to be adjusted by varying the laser energy. It is also demonstrated that mixtures of sugar and magnetic particles can be processed as 3D composites. As a proof of concept, a sugar-based millimeter-scale helical swimmer, which is capable of corkscrew motion in a solution with a viscosity comparable to those of biological fluids, is fabricated. The millirobot quickly dissolves in water, while being manipulated through magnetic fields. The present fabrication method can pave the way to a new generation of transient sugar-based small-scale robots for minimally invasive procedures. Due to their rapid dissolution, sugars can be used as an intermediate step for transporting swarms of particles to specific target locations.

Magnetic cilia carpets with programmable metachronal waves.

Metachronal waves commonly exist in natural cilia carpets. These emergent phenomena, which originate from phase differences between neighbouring self-beating cilia, are essential for biological transport processes including locomotion, liquid pumping, feeding, and cell delivery. However, studies of such complex active systems are limited, particularly from the experimental side. Here we report magnetically actuated, soft, artificial cilia carpets. By stretching and folding onto curved templates, programmable magnetization patterns can be encoded into artificial cilia carpets, which exhibit metachronal waves in dynamic magnetic fields. We have tested both the transport capabilities in a fluid environment and the locomotion capabilities on a solid surface. This robotic system provides a highly customizable experimental platform that not only assists in understanding fundamental rules of natural cilia carpets, but also paves a path to cilia-inspired soft robots for future biomedical applications.

Inkjet printed superparamagnetic polymer composite hemispheres with programmed magnetic anisotropy.

We present the fabrication and characterization of large arrays of inkjet-printed superparamagnetic polymer composite (SPMPC) hemispherical microstructures. SPMPCs are appealing for applications in microsystems and nanorobotics due to the added functionality of polymers and the significant magnetic attributes of embedded nanostructures. SPMPC-based microarchitectures can be used to perform different functions wirelessly in various media (e.g. water, solvents) using external magnetic fields: handling and assembling small objects, delivering drugs or biomass, or sensing specific physical or chemical changes. In this work superparamagnetic magnetite nanoparticles are dispersed in SU-8 to form magnetic hemispheres. Magnetically anisotropic hemispheres as well as standard SPMPC hemispheres are fabricated. Magnetic anisotropy is programmed by applying a magnetic field during curing. The distribution of nanoparticles inside the polymer matrix and magnetic characteristics of the SPMPC are investigated. Magnetic manipulation of hemispheres is demonstrated at liquid-liquid interfaces. Different assembly strategies to form lines or geometric shapes from hemispheres as well as their independent dynamic control are demonstrated. Finally, a two-interface assembly strategy is demonstrated to assemble hemispheres into complete spheres for advanced self-assembly tasks.