Transitioning Imaging FlowCytoBot for Harmful Algal Bloom Mitigation and Research
Monitoring programs for harmful algal blooms (HABs) typically rely on toxin analysis of seafood samples or time-consuming manual enumeration of phytoplankton, restricting the frequency of observations and the potential forearly warning. Automated submersible microscopes such as Imaging FlowCytobot (IFCB) have begun to change this situation. IFCB uses a combination of flow cytometric and video technology to capture high resolution images of suspended particles, and machine learning technology can identify potentially toxic species from the images.
The Woods Hole Oceanographic Institution (WHOI) was awarded a three year grant by IOOS’ Ocean Technology Transition Project to expand the IFCB’s potential operational use by deploying it on autonomous vehicles in the Gulf of Maine and in the Gulf of Mexico, to enable high resolution plankton studies with both long duration and spatial coverage. The current generation IFCB and its suite of image analysis software have been in development for over 10 years. Initially conceived as an instrument that would be deployed at fixed locations, the IFCB-AV project has been designed so that the strengths of the current IFCB design are best translated to a mobile deployment platform. The project also leverages an extensive effort to transition the suite of IFCB image analysis tools to a web services design, minimizing the investment necessary for new users to take advantage of these.
Expected benefits from the proposed work are improved early warning of HABs in the Gulf of Maine and the Gulf of Mexico. The project will also benefit the oceanographic research community by providing much needed capability for high resolution observations of plankton and will aid in developing a Marine Biodiversity Observation Network for US coastal waters, which will benefit many other end users responsible for managing fisheries, protected species, water quality, invasive species, and other societally relevant issues. Project outcomes include a commercially available instrument and open source software and data systems that incorporate standards to promote interoperability.
Planned activities are listed below by implementation year. Status updates will be added to this page as the project progresses.
Year 1 – $439,730 (September 2015 – August 2016)
- Extend IFCB capabilities to include deployment on autonomous vehicles (AV) including: 1) construct research prototype IFCB-AV for multiple surface vehicles and 2) Lab and Field evaluation of prototype IFCB-AV for long duration deployments. WHOI will employ an iterative process for design, construction, and evaluation that culminates in first a prototype IFCB-AV, and second a pre-commercial IFCB-AV, both available for integration with different vehicles. The impact and effectiveness of the pre-commercial design and construction phase will be ensured by direct involvement of industry partner McLane Research Labs, the licensed manufacturer of IFCB. Update: A research prototype for the IFCB-AV was constructed and evaluated in the laboratory and in Woods Hole Harbor. Extensive field tests in conjuction with the anticipated bloom in Salt Pond, MA were conducted in the early spring of 2016. Continued field testing of a prototype IFCB-AV has yielded positive results and motivated additional modifications to the existing IFCB design to enhance performance in a mobile horizontal orientation. Prototype IFCB-AV has successfully field tested in a JetYak vehicle in Woods Hole Harbor and Salt Pond in the Nauset Marsh System in Eastham, MA with acceptable results. This successful evaluation of the prototype design in conjunction with communications with industry partner McLane Research Laboratories, Inc. has resulted in the production of a pre-commercial IFCB-AV. This unit has been delivered to WHOI for testing. IFCB-AV will continue to undergo field deployments to refine sampling methods and start testing on a Liquid Robotics Wave Glider.
- Initiate development of an information support system for HAB end-users by developing use cases.While IFCB has proven capability for HAB detection, transition of the technology into operational use requires investment in user-friendly data analysis and interpretation utilities that directly meet end user needs. Importantly, in view of the large amounts of data collected during continuous monitoring, effective automated image classification (often to genus or even species level) has been demonstrated (Sosik and Olson 2007). The WHOI PIs have collaborated with computer scientists to develop prototype informatics tools that enable web-based near real time access to IFCB data and to analysis products produced by automated workflows (Sosik and Futrelle 2012) (see http://ifcb-data.whoi.edu/). WHOI will work with small interdisciplinary teams including academic and government partners as well as hardware and information systems engineers to develop use cases specific to HAB end user needs. Update: The IFCB data and analysis software has been enhanced in several ways: (1) new data product generation code was implemented with tools unencumbered by commercial licenses (Python/scipy), while continuing to support users of MATLAB; (2) new user guides were written and made publically available, along with test data sets, for IFCB data analysis and product generation including image segmentation, feature extraction, construction and evaluation of automated image classifiers, classifier application, and interpretation of classification results
(https://github.com/hsosik/ifcb-analysis/wiki); and (3) support for spatiotemporal data including capturing GPS data from IFCB-AV vehicles and temporally aligning GPS tracks with IFCB sample collection, as well as export of that data into standard formats (e.g., KML). All code is available in GitHub (http://github.com/joefutrelle/oii; https://github.com/hsosik/ifcb-analysis). In addition to software enhancements, the IFCB dashboard has been packaged into a virtual machine that can be rapidly deployed to users in the field without requiring a Linux OS or manual installation of dashboard code (http://github.com/joefutrelle/ifcb-dashboard); end user documentation is provided. Field-deployed IFCB dashboards provide visualization of data, imagery, and instrument parameters via the same user interface provided by onshore web services.
- Initiate partnerships to promote access and sustainability of IFCB technology, including coordination with IFCB manufacturers, vehicle manufacturers, state agency HAB application users, and NERACOOS and GCOOS to ensure data access. Update: The PI has been building on existing collaborative relationships among a variety of researchers, IFCB developers at WHOI, and McLane Research Laboratories (MRL; licensed manufacturer of IFCB) to develop hardware and software improvements and transition them into commercial production. In addition, we have provided a variety of end user support including assisting users via the online IFCB user group (hosted by MRL), developed online documentation for the IFCB data dashboard and analysis software, and provide hardware and software training for new IFCB users.
Year 2 (September 2016 – August 2017):
- Continue development of an information support system for HAB end-users by developing prototype solutions supporting the end user needs identified in Year 1, as well as accompanying design and evaluation documentation to allow for sustained development and enhancement. Design, prototype, and evaluation steps will be repeated as various candidate technical implementations evolve. The team will thereby converge on solutions that meet high priority end user goals, while leveraging as much existing technology and technical efficiency as possible. Update: Current focus is on information delivery to HAB management end users. Developing protocols for data product delivery to IOOS data providers (GCOOS and NERACOOS), and designing an end user interface. A new prototype for issuing HAB alerts through GCOOS is now under evaluation.
- Demonstrate IFCB capabilities on a wave glider in the Gulf of Maine. To date, IFCB deployments to study HABs in the Gulf of Maine region have been restricted to the Nauset estuary, an inshore embayment on Cape Cod that suffers annual blooms of A. fundyense. A major focus of year 2 activities for this project will be to demonstrate IFCB-AV capability for targeting larger extent coastal blooms within the Gulf of Maine. To do this, the project will leverage on-going efforts to deploy ESP technology in the region and coordinated use of ships to sample blooms that ESP detects. These cruises will make it possible to deploy the Wave Glider with integrated IFCB-AV for multi-day surveys in the vicinity of ESP moorings. In this scenario, IFCB-AV will provide mobile, high frequency (greater than twice per hour) observations of phytoplankton diversity and abundance and ESP will verify (or negate) the presence of toxic cells by direct measurement of biotoxins. Update: Wave Glider planned to be depolyed for demonstration in spring 2017.
- Continue to maintain and build partnerships to promote access and sustainability of IFCB technology,including coordination with IFCB manufacturers, vehicle manufacturers, state agency HAB application users, and NERACOOS and GCOOS to ensure data access. Update: Collaborating with: 1) the IFCB manufacturer for broad distribution with focus on IFCB hardware and acquisition, including streamlining new user hardware/software training and revisions to manuals and protocol documents; 2) Vehicle manufacturers for expertise; 3) State agency users with HAB applications, including participation in a joint technology developer; and 4) NERACOOS / GCOOS for data access, including information system components being (a) codeveloped by the IFCB teams at WHOI and TAMU and the GCOOS Data Portal team and (b) being evaluated by team members from each sector (academic researchers, IOOS RA offices, and state agency representatives).
Year 3 (September 2017 – August 2018)
- Complete development information support system. In addition to delivering specific end user products, the project will adopt technical approaches to facilitate broad adoption of the information system for a variety of uses and settings, including meeting diverse hardware requirements, making code available under open-source licensing, reducing complexity of installation and configuration, adopting standard data and metadata formats as appropriate, and enabling operation in low-bandwidth deployments. Enhancements to the current informatics system will include production of and web-based access to new kinds of IFCB data products such as maps, as well as new technical capabilities such as improved interoperability with standard data formats and IOOS protocols.
- Demonstrate IFCB capabilities on an AutoNaut in the Gulf of Mexico. IFCB has already proven to be extremely valuable, both as a research tool and in providing early warning for TWPD and TX-DSHS personnel (e.g., Campbell et al.2010; Campbell et al. 2013). A major focus of Year 3 activities for this project will be to demonstrate IFCB-AV capability for detecting offshore blooms sooner and tracing their trajectory along the Texas Coast. To do this, on-going efforts by PI Knap to develop autonomous observing capabilities with the AutoNaut surface vehicle will be leveraged. Autonaut deployments with integrated IFCB-AV will begin in the late summer (Aug/Sept 2017), the most frequent period of Karenia brevis bloom initiation, for a 2-4 month deployment. The Autoaut will permit expanded spatial coverage of offshore regions, which are currently not monitored for HABs. Update: AutoNaut on track to be deployed in Year 3 of award.
- Continue to maintain and build partnerships to promote access and sustainability of IFCB technology, including coordination with IFCB manufacturers, vehicle manufacturers, state agency HAB application users, and NERACOOS and GCOOS to ensure data access.
For more information about this project contact Dr. Heidi Sosik.
More information on Harmful Algal Blooms can be found at: