Abstract
This paper presents SAMSON, a Spectral Absorption-fluorescence Microscopy System for ON-site-imaging of algae within a water sample. Designed to be portable and low-cost for on-site use, the optical sub-system of SAMSON consists of a mixture of low-cost optics and electronics, designed specifically to capture both fluorescent and absorption responses from a water sample. The graphical user interface (GUI) sub-system of SAMSON was designed to enable flexible visualisation of algae in the water sample in real-time, with the ability to perform fine-grained exposure control and illumination wavelength selection. We demonstrate SAMSON's capabilities by equipping the system with two fluorescent illumination sources and seven absorption illumination sources to enable the capture of multispectral data from six different algae species (three from the Cyanophyta phylum (blue-green algae) and three from the Chlorophyta phylum (green algae)). The key benefit of SAMSON is the ability to perform rapid acquisition of fluorescence and absorption data at different wavelengths and magnification levels, thus opening the door for machine learning methods to automatically identify and enumerate different algae in water samples using this rich wealth of data.
Introduction
Harmful Algae Blooms (HABs) represent a critical environmental and public health challenge. These blooms produce toxins that contaminate drinking water supplies and cause severe damage to aquatic ecosystems. Currently, monitoring HABs is a bottlenecked process requiring water samples to be transported to specialized laboratories for manual identification by expert taxonomists.
This traditional approach suffers from significant latency, high costs, and inherent subjectivity, with human identification accuracy typically ranging between 67% and 83%. Furthermore, the global shortage of expert taxonomists makes scaling these manual efforts difficult. SAMSON (Spectral Absorption-fluorescence Microscopy System for ON-site-imaging) was developed to address these issues by enabling rapid, digital, and consistent data collection directly at the water source.
Methods
The SAMSON system is built on a modular, low-cost architecture divided into two primary sub-systems:
Optical Sub-system
Unlike conventional microscopes, SAMSON utilizes a multispectral approach to capture biological "fingerprints." The system incorporates nine LEDs: seven for absorption (ranging from 465nm blue to 660nm deep-red) and two for fluorescence (385nm and 405nm ultraviolet). A 20x objective lens, paired with a high-pass filter, allows the sensor to capture both the light absorbed by the algae and the specific wavelengths emitted when excited by UV light.
GUI and Software Triggering
To capture microorganisms moving in water without motion blur, the researchers developed a custom Graphical User Interface (GUI) and a back-end software triggering system. This system synchronizes the LEDs with a 4.1 MP camera sensor, allowing it to cycle through all nine wavelengths rapidly. This process creates a consistent "data cube" for every frame, ensuring that multispectral data is perfectly aligned for each organism.
Experimental Evaluation
The system was evaluated using six algae species from the Canadian Phycological Culture Centre (CPCC), representing two major phyla:
- Cyanophyta (Blue-green algae): Microcystis aeruginosa, Anabaena flos-aquae, and Pseudanabaena tremula.
- Chlorophyta (Green algae): Scenedesmus obliquus, Scenedesmus quadricauda, and Ankistrodesmus falcatus.
These species include several known toxin producers, making them high-priority targets for environmental monitoring and automated classification.
Key Results
The SAMSON system demonstrated high-performance imaging capabilities suitable for field deployment:
- Spatial Resolution: The system achieved a resolution of 0.26 µm/pixel, capturing fine morphological details necessary for species differentiation.
- Acquisition Speed: By optimizing LED brightness and exposure, the system achieved an effective frame rate of approximately 30 frames-per-second while cycling through all nine wavelengths.
- Spectral Differentiation: The resulting "imaging cubes" revealed that different species possess unique visual profiles across the spectral range. Notably, blue-green algae (Cyanophyta) exhibited significantly higher fluorescence at 385nm and 405nm compared to green algae (Chlorophyta), providing a clear spectral signature for identification.
Discussion and Future Work
SAMSON successfully demonstrates that high-quality multispectral data can be collected using a portable, low-cost instrument. By replacing the traditional optical eyepiece with a high-resolution digital sensor, the system facilitates real-time collaboration and longitudinal digital storage.
The current study focuses on the instrumentation and data acquisition phase. The next step involves implementing machine learning algorithms to automate the identification and counting of algae species using the rich spectral data provided by SAMSON. Future research will also focus on testing the system with complex, turbid real-world samples and developing an autofocus mechanism to enable fully autonomous monitoring stations.
Summary updated