Thesis: Ph. D., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2019
Cataloged from PDF version of thesis.
Includes bibliographical references.
...High-quality complex biopharmaceutical products based on cells and proteins are transforming modem medicine and advancing treatments for many health conditions. Continuous biomanufacturing is one of the top technology trends in the biopharmaceutical industry to improve biological product quality and reduce manufacturing cost. This thesis introduces novel high-throughput microfluidic cell separation and nanofluidic protein quality monitoring technologies. The novel micro/nanofluidic system enables reliable and efficient microfiltration and robust online rapid product quality monitoring during continuous biomanufacturing. This technology overcomes the limitations of the current membrane-based microfiltration and quality monitoring technologies, including filter clogging, low product recovery, manual sample preparation, and off-line analysis. The first part describes the novel cell retention device for perfusion culture based on inertial sorting.
Size-dependent hydrodynamic forces enabled membrane-less microfiltration for the separation of suspended mammalian cells. The device performance in terms of cell retention efficiency, long-term biocompatibility, and scalability was assessed. Long-term and small-scale perfusion culture using the device was subsequently demonstrated. Clog-free cell retention with high product recovery in this work can be utilized for long-term reliable and efficient perfusion culture. The next part describes the removal of small dead cells from bioreactor cultivation by high-throughput size-based cell separation using inertial sorting. The device parameters were studied to optimize removal of the dead cells, and high-throughput and high-concentration dead cell removal was demonstrated. Finally, continuous online purity monitoring of the proteins in the cell culture supernatant during perfusion culture was achieved with a novel nanofluidic filter array.
This nanofluidic device with online sample preparation system was integrated with perfusion culture using the microfluidic cell retention device. The purity of proteins in the cell culture supernatant was monitored for more than a week in a fully automated continuous manner. As a robust online sensing technology for continuous biomanufacturing, this nanofluidic filter array could replace the existing offline analytical technologies for protein purity monitoring. In summary, this thesis presents a novel micro/nanofluidic system for separation and monitoring of cells and proteins for continuous biomanufacturing. This innovative approach can contribute to long-term reliable and efficient biomanufacturing in the future.
a dissertation presented by Taehong Kwon.
Ph. D.
Ph.D. Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science
Musculoskeletal disorders are an unavoidable occupational health problem. In particular, workers who perform repetitive tasks onsite in the manufacturing industry suffer from musculoskeletal ...problems. In this paper, we propose a system that evaluates the posture of workers in the manufacturing industry with single-view 3D human pose-estimation that can estimate the posture in 3D using an RGB camera that can easily acquire the posture of a worker in a complex workplace. The proposed system builds a Duckyang-Auto Worker Health Safety Environment (DyWHSE), a manufacturing-industry-specific dataset, to estimate the wrist pose evaluated by the Rapid Limb Upper Assessment (RULA). Additionally, we evaluate the quality of the built DyWHSE dataset using the Human3.6M dataset, and the applicability of the proposed system is verified by comparing it with the evaluation results of the experts. The proposed system provides quantitative assessment guidance for working posture risk assessment, assisting the continuous posture assessment of workers.
CT-P10 was the first licensed rituximab biosimilar. This Korean post-marketing surveillance study evaluated CT-P10 safety and effectiveness in approved indications.
This prospective, open-label, ...observational, phase 4 study collected routine clinical practice data across 27 centers in the Republic of Korea. Patients received their first CT-P10 treatment, per prescribing information, for non-Hodgkin's lymphoma (NHL), chronic lymphocytic leukemia (CLL), rheumatoid arthritis (RA), granulomatosis with polyangiitis (GPA), or microscopic polyangiitis (MPA) during the surveillance period (16 November 2016-15 November 2020). Safety (including adverse events AEs and adverse drug reactions ADRs) and disease-specific clinical response (by best overall response NHL/CLL, Disease Activity Score in 28-joints RA, or Birmingham Vasculitis Activity Score for Wegener's Granulomatosis GPA/MPA) were assessed for ≤1 year (NHL/CLL) or ≤24 weeks (RA/GPA/MPA).
The safety population comprised 677 patients (604 NHL, 16 CLL, 42 RA, 7 GPA, 8 MPA). AEs/ADRs were reported for 68.4%/27.7% (NHL/CLL), 31.0%/14.3% (RA), and 86.7%/13.3% (GPA/MPA) of patients. Serious AEs and unexpected ADRs did not raise new safety signals. Pneumonia was the most frequent serious ADR overall. Positive effectiveness outcomes were observed.
Findings were consistent with the known CT-P10/reference rituximab safety profile, with high effectiveness observed in NHL/CLL and RA.
In this study, the photosensitive carbon nanotube (CNT) paste has been prepared by a 3-roll milling process of the mixture having CNT, organic binder, terpineol, and additives such as Ag powder, frit ...glass, and photosensitive polymer. The composition of the mixture was varied to control the viscosity. The field emission display (FED) panel has been fabricated with a photolithography process. TGA experiment shows that most of the organic components of the paste are decomposed at 400°C. The current density of 330.7μA/cm2 was obtained with applying electric field of 7.5V/μm. When the surface of the CNTs was treated by adhesive tape, the field enhancement factor could be improved due to the enlargement of emitting area.
Since the introduction of DeepMimic Peng et al. 2018, subsequent research has focused on expanding the repertoire of simulated motions across various scenarios. In this study, we propose an ...alternative approach for this goal, a deep reinforcement learning method based on the simulation of a single-rigid-body character. Using the centroidal dynamics model (CDM) to express the full-body character as a single rigid body (SRB) and training a policy to track a reference motion, we can obtain a policy that is capable of adapting to various unobserved environmental changes and controller transitions without requiring any additional learning. Due to the reduced dimension of state and action space, the learning process is sample-efficient. The final full-body motion is kinematically generated in a physically plausible way, based on the state of the simulated SRB character. The SRB simulation is formulated as a quadratic programming (QP) problem, and the policy outputs an action that allows the SRB character to follow the reference motion. We demonstrate that our policy, efficiently trained within 30 minutes on an ultraportable laptop, has the ability to cope with environments that have not been experienced during learning, such as running on uneven terrain or pushing a box, and transitions between learned policies, without any additional learning.
Since the introduction of DeepMimic Peng et al. 2018a, subsequent research has focused on expanding the repertoire of simulated motions across various scenarios. In this study, we propose an ...alternative approach for this goal, a deep reinforcement learning method based on the simulation of a single-rigid-body character. Using the centroidal dynamics model (CDM) to express the full-body character as a single rigid body (SRB) and training a policy to track a reference motion, we can obtain a policy that is capable of adapting to various unobserved environmental changes and controller transitions without requiring any additional learning. Due to the reduced dimension of state and action space, the learning process is sample-efficient. The final full-body motion is kinematically generated in a physically plausible way, based on the state of the simulated SRB character. The SRB simulation is formulated as a quadratic programming (QP) problem, and the policy outputs an action that allows the SRB character to follow the reference motion. We demonstrate that our policy, efficiently trained within 30 minutes on an ultraportable laptop, has the ability to cope with environments that have not been experienced during learning, such as running on uneven terrain or pushing a box, and transitions between learned policies, without any additional learning.
Dog Noseprint Identification Algorithm Cho, Sungmin; Paeng, Jinwook; Kim, Taehong ...
2021 International Conference on Information Networking (ICOIN),
2021-Jan.-13
Conference Proceeding
This paper proposes a dog noseprint identification system based on Gabor filter and feature matching. Given images of dog noseprints, the system determines the region of interest, pre-processes the ...images using adaptive thresholding, extracts features, and performs feature matching to identify dogs. To extract features, we first apply the Gabor filter with 60 directions to images. Then we employ the scale invariant feature transform (SIFT) feature extractor to obtain keypoints that are invariant to image rotation and scaling. The extracted keypoints are compared with reference key-points of a dog noseprint that needs to be identified. To improve the matching accuracy, we present several matching algorithms. Experiments show that the SIFT based identification system method surpasses other methods in terms of accuracy, while the ORB based on system outperforms other methods in terms of speed.
IP-based Wireless Sensor Network (IP-WSN) is one of the essential elements enabling Internet of Things (IoT). However, IP-WSN imposes great challenges due to low processing resources and strict ...energy constraints of sensors. Various routing protocols for IP-WSN have been proposed considering low-cost communication with resource-constraint and typical traffic patterns like multipoint-to-point, but most of routing protocols causes problems incurring inefficient route or/and consuming many processing resources. Especially, as far as point-to-point (P2P) traffic is concerned, those routing protocols incur triangular detour routes and require many processing resources at intermediate nodes where P2P traffic is an important traffic pattern. To address these challenges, we propose the Stateless P2P Routing protocol (SPR) based on shortcut tree routing algorithm which is our previous study. SPR can deliver a packet to the node having the smallest remaining hop count among neighbors without additional control overhead, instead of always delivering a packet to a parent or children along tree routes. SPR also provides a nearly stateless routing in that SPR determines a route through hierarchical address structure and one hop neighbor information without having to store global routing state. We implement SPR in our IP-WSN platform named SNAIL and conduct a simulation and a measurement to verify the performance of SPR. The simulation results show SPR provides improved hop count compared to HiLow and RPL. It also provides reduced memory usage and the number of control packets compared to RPL. Additionally, the measurement results show SPR provides decreased round trip time and increased packet delivery ratio compared to HiLow and RPL.