https://orcid.org/0009-0002-9283-0062
Figures
Abstract
Cells and microorganisms are motile, yet the stationary nature of conventional microscopes impedes comprehensive, long-term behavioral and biomechanical analysis. The limitations are twofold: a narrow focus permits high-resolution imaging but sacrifices the broader context of organism behavior, while a wider focus compromises microscopic detail. This trade-off is especially problematic when investigating rapidly motile ciliates, which often have to be confined to small volumes between coverslips affecting their natural behavior. To address this challenge, we introduce Trackoscope, a 2-axis autonomous tracking microscope designed to follow swimming organisms ranging from 10μm to 2mm across a 325cm2 area (equivalent to an A5 sheet) for extended durations—ranging from hours to days—at high resolution. Utilizing Trackoscope, we captured a diverse array of behaviors, from the air-water swimming locomotion of Amoeba to bacterial hunting dynamics in Actinosphaerium, walking gait in Tardigrada, and binary fission in motile Blepharisma. Trackoscope is a cost-effective solution well-suited for diverse settings, from high school labs to resource-constrained research environments. Its capability to capture diverse behaviors in larger, more realistic ecosystems extends our understanding of the physics of living systems. The low-cost, open architecture democratizes scientific discovery, offering a dynamic window into the lives of previously inaccessible small aquatic organisms.
Citation: Soneji P, Challita EJ, Bhamla S (2024) Trackoscope: A low-cost, open, autonomous tracking microscope for long-term observations of microscale organisms. PLoS ONE 19(7): e0306700. https://doi.org/10.1371/journal.pone.0306700
Editor: Baeckkyoung Sung, KIST: Korea Institute of Science and Technology, GERMANY
Received: December 26, 2023; Accepted: June 21, 2024; Published: July 11, 2024
Copyright: © 2024 Soneji et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All the data required to replicate the results are included within the Supporting information + github + figshare: a. https://github.com/bhamla-lab/Trackoscope b. https://figshare.com/projects/Trackoscope_A_Low-Cost_Open_Autonomous_Tracking_Microscope_for_Long-Term_Observations_of_Microscale_Organisms/200934.
Funding: NIH Grant R35GM142588; NIGMS SEPA Grant R25GM142044; NSF Grants MCB-1817334; CAREER IOS-1941933; and the Open Philanthropy Project The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Beyond conventional microscopy: Enabling the study of microorganism motility
Microscopy serves as a pivotal tool for delving into the microscopic realm, facilitating the study of the inner mechanisms and behaviors of organisms. Traditional microscopy, constrained by a fixed lens, falls short in capturing the full spectrum of microorganism motility. The manual tracking of these microscale entities under a microscope presents challenges due to their diverse sizes, velocities, and the precision required for observation. A tracking microscope mitigates these issues, enabling precise and efficient monitoring of an organism’s movement, thus fostering more controlled experimentation and detailed observation of behavioral patterns and microstructures. It automates the tracking process, conserving time, reducing user error, and minimizing variability.
Several tracking solutions exist within the microscopy landscape, including motorized stage microscopes, laser tracking systems, and image-based tracking systems. Motorized stage microscopes utilize motors to maneuver the stage that holds the sample, facilitating user control over sample movement with autonomous tracking capabilities within a specified area (up to 144 cm2) or infinitely in one plane [1–3]. Laser tracking systems, employing lasers to detect and follow sample movement, adjust the microscope’s focus accordingly [4]. Image-based tracking systems, prevalent in the field, apply computer vision algorithms to follow the sample’s movement through images captured by the microscope [5–8]. While these methods mark significant strides in data collection for organism study, they are not without drawbacks. Image-based systems are bound by the resolution and field of view of the imaging device, laser systems are adept at following single particles but are less effective with complex organisms, and motorized stage microscopes, while versatile, can be costly and restricted by the size of the trackable area. Additionally many of these setups can be quite expensive, costing between 1000 and 5000 dollars [2, 3].
Despite recent advancements in accessible, affordable, and open hardware microscopy [9–15], the ability to observe moving organisms remains a significant challenge with traditional microscopes. To bridge this gap, we introduce Trackoscope, a low-cost, open tracking microscope. Trackoscope, costing around 400 dollars in parts, is accessible when compared to traditional light-field microscopes and can be assembled by a minimally trained individual. It employs image-based tracking to autonomously follow and focus on moving organisms, enabling precise tracking over an expansive field of view (approximately 325 cm2). Equipped with a 12-megapixel camera, Trackoscope captures video data amenable to machine-learning-based behavior analysis pipelines. The affordability, customizability, and ease of assembly make Trackoscope an invaluable asset in both university-level research and K-12 education, fostering the exploration and analysis of micro-organism behaviors.
Trackoscope: An affordable microscope with open hardware and software
Trackoscope’s low-cost motorized XY stage
Trackoscope is a modular, lead screw-driven, two-axis actuator designed for automated visual tracking of motile microorganisms. It can be controlled by a standard computer or laptop using an Arduino Uno and communicates with the computer via a USB-based serial communication (COM) port (Fig 1). The driver stack can support any motorized stages that are stepper motor-based. The Trackoscope design features motorized X and Y stages with an 18 cm x 18 cm travel range, powered by two NEMA 17 stepper motors with 400 steps/rev (precision of 0.9 degrees), which are controlled by a CNC Shield (S1 Fig). The CNC Shield allows for the increase of torque and the addition of micro-stepping, which can increase precision up to 32 times. Specifically, micro-stepping allows the same motors to move at a range of speeds, from 4,600 μm/s for fast-swimming ciliates to 145 μm/s for slow-moving amoeba and tardigrades.
(A) The Trackoscope prototype in the inverted tracking position. The platform’s footprint is 900 cm2, with a tracking area of 325 cm2. (B) Displays a CAD model of the optics assembly. The objective lenses are interchangeable, permitting the observation of a variety of organisms. The Raspberry Pi HQ Camera alongside the Raspberry Pi Zero function as a webcam, allowing this custom digital microscope to interface with any device. (C) Demonstrates the Trackoscope user interface, featuring a live video feed, real-time tracking map, manual actuator controls, and data saving options. (D) Illustrates the flowchart of the tracking algorithm and the visualization of logic. Utilizing OpenCV’s built-in tracking, the algorithm pinpoints the organism’s location based on the user’s initial selection of a region of interest. The organism’s location informs the actuator’s movements to maintain the organism within the field of view.
To improve precision and absorb vibrations from potential assembly misalignments, the X and Y axes have pulleys with a 3:1 gear ratio. The XY stage utilizes a stacked design, with the X axis driving the Y axis, which then moves a magnetically-attached optics stage. This stacked system simplifies the assembly since each axis can be constructed separately, and it also allows for modularity if movement in only one axis is needed. Based on observations that a thin liquid layer in the organism sample negates the need for Z-axis adjustments, we opted for a manual Z-stage in this design. This choice reduces cost and simplifies user modifications to the optics unit. Additionally, Trackoscope can also be inverted to improve imaging based on the organism (S2 Fig).
The footprint of Trackoscope is 900 cm2 (fits within a square foot), weighs 3.4 kg (equivalent to a gallon of milk), and the actuator itself costs $167 (Table 1).
The optical core of Trackoscope
The optics module consists of four different components: an imaging sensor, an achromatic doublet lens, a lens tube, and an objective (Fig 1b). The imaging sensor is a Raspberry Pi High-Quality Camera paired with a Raspberry Pi Zero. A bare sensor works best as it optimizes the largest possible field of view (FOV). While the Raspberry Pi High-Quality Camera (12.3 MP Sony IMX477 sensor, 1.55 um pixel size) is the most economical option at $50, higher quality cameras can also be attached to further increase imaging resolution. A higher-end $400 camera (Imaging Source DFK 37BUX273 USB 3.1) was also tested and produced images of similar quality to the Raspberry Pi camera. Beneath the camera is an achromatic doublet lens inside a lens tube, at the end of which a variety of objectives can be connected, ranging from 2X to 20X. With the distance of achromatic doublet within the lens tube, the compounded magnification of the microscope can range from 20X to 200X, depending on the camera sensor used, as the focal length of the sensor affects the initial magnification. A LED ring light around the optics provides bright field illumination, and a white backing is placed behind the sample to provide a clean background. Dark field illumination can also be used with Trackoscope by adding a 3D printed attachment to the ring light that directs the light around the sample from below.
Micro-tracking with OpenCV
To track the organisms, we employ OpenCV’s Channel and Spatial Reliability Tracker (CSRT) [16, 17]. The CSRT is favored over machine learning trackers for its broader applicability without the need for a high-performance GPU. By allowing the user to select the organism or region of interest (ROI), the tracker bypasses the necessity of pre-training with the specific organism being tracked. In the graphical user interface (GUI) (Fig 1c), designed using Tkinter, the user can view a live video feed from the microscope, observe the live trajectory of the organism, manually operate the Trackoscope, and save tracking data as a CSV file.
Upon selecting an organism for tracking, the user clicks “start tracking” in the GUI, prompting a window for manual organism selection using the cursor. After selection and hitting the “enter” key, tracking commences. OpenCV identifies the organism and its bounding box from the user’s initial selection, then continually updates the box position as the organism moves. The CSRT model updates the bounding box by continually adapting over the duration of the track as it considers both the appearance of the organism and its motion characteristics. This adaptability makes the CSRT particularly effective in handling challenges such as occlusion, rotation, and scale variation. The bounding box data then informs actuator movement calculations (Fig 1d). If the bounding box center moves outside the central zone, commands are issued to the Arduino to adjust the axis until the organism is re-centered in the video frame. This feedback mechanism ensures the organism remains within the microscope’s field of view (FOV). The use of a central zone instead of an exact center optimizes the smoothness of the video by reducing the chance of overcompensating movements, especially with organisms that may stop or pivot abruptly. Additionally, as the CSRT model evolves over the track, the perceived center of the organism may not always be accurate or precise; this is another factor for the incorporation of a central zone as the perceived center, while always on the organism, may not be on the exact center.
Intuitive and custom graphical user interface
The user interface provides access to tools and data useful throughout the tracking process (Fig 1c). The GUI displays a live video feed from the digital microscope, averaging 107 FPS at a resolution of 640 x 480, which is adjustable. A timestamp is also included for reference during analysis. A live trajectory map on the right side of the video feed plots points based on actuator movements during tracking, with a blue hexagon indicating the organism’s current position. Tracking locations and timestamps are saved over time in a CSV file for later download. Manual actuator controls are available in the GUI’s bottom right along with various tracking-related controls such as toggling telemetry data display or stage smoothness adjustments (S1 Video).
Performance evaluation of tracking and imaging capabilities
We evaluated the Trackoscope’s resolution capabilities, lighting, and tracking speed (Fig 2). Resolution was empirically determined using USAF 1951 Resolution Targets (Thorlabs R1DS1P), with the 4x objective at 35.08 μm (Fig 2d) and the 10x objective at 8.77 μm (Fig 2e). The resolution formula for the USAF 1951 target is RESLP = 2Group+(Element−1)/6, with the center-to-center line distance given by RESCC = 1000/RESLP. This resolution allows for the observation of detailed organism features such as cilia clusters or internal structures (Fig 2b and S2 Video). Dark field imaging allows for higher contrast when observing more translucent organisms (Fig 2c and S2 Video). Image brightness was assessed for the Raspberry Pi HQ microscope setup using the Python Image Library (PIL) to analyze grayscale pixel brightness. Brightness levels were 88.32% for the Amscope ring light and 58.21% for the AIXPI ring light. Speed was measured by timing a 1 cm travel at different micro-stepping settings (Fig 2f). Depending on the organism’s speed, an appropriate micro-stepping setting is selected, typically twice the organism’s speed.
(A) Microscope image resolution of Stentor coeruleus with a 4x objective lens. (B) Detail resolution of Stentor with a 10x objective lens. (C) Dark field image of Spirostomum ambiguum with a 4x objective lens. (D) Image of a USAF 1951 resolution target captured with a 4X objective on the Raspberry Pi HQ microscope setup, showing an enlarged view of Group 4 and an intensity profile along the indicated blue line. This confirms resolution for Group 4, Element 6, equivalent to 28.51 Line Pairs/mm or a 35.08 μm resolution. (E) Image of a USAF 1951 resolution target with a 10x objective on the Raspberry Pi HQ microscope setup, showing an enlarged view of Group 6 and an intensity profile along the indicated blue line. This confirms resolution for Group 6, Element 6, equivalent to 114.04 Line Pairs/mm or an 8.77 μm resolution. (F) Chart of platform speeds across various micro-stepping settings, with examples of trackable organisms at those speeds.
Movement pattern analysis from tracking data
We analyze the position and video tracking data to extract movement patterns, speeds, feeding behaviors, and other organism dynamics. Data points are recorded every 50 ms, capturing the timestamp, the organism’s position relative to the center of the video frame (xo/f, yo/f), and the platform’s position (xf, yf) relative to the starting point. The organism’s position P(xo, yo) is computed by adding the displacements of the platform and the organism’s position within the video frame: P(xo, yo) = P(Δxo/f + Δxp, Δyo/f + Δyp) (S3 Fig).
This displacement data facilitates the calculation of organism speed and identification of features such as feeding-swimming transitions and gait-switching frequencies. During tracking, we also utilize Open Broadcast Studio [18] to record the tracking window, providing visual data for subsequent analysis of shape changes and gait through visual tracking.
Microscopic tracking of varied organism behaviors
We demonstrate the capabilities of Trackoscope for tracking microscopy by analyzing the movement of organisms varying in size, speed, and behavior. This tracking not only showcases the range of Trackoscope’s functionality but also provides insights into unique organism behaviors.
High-speed ciliate tracking.
We tracked fast-moving ciliates such as Bursaria truncatella (800 μm), Blepharisma japonicum (500 μm), Spirostomum ambiguum (600 μm), and Stentor coeruleus (600–900 μm). The long-duration tracks of Bursaria truncatella and Stentor coeruleus test Trackoscope’s ability to maintain focus on rapidly moving ciliates, which typically escape the view of traditional microscopes instantaneously. For instance, Bursaria truncatella, shaped like a scoop, achieves top speeds of 1775 μm/s (11.8 body lengths/second) multiple times within an 18-minute track, covering a distance of 45 centimeters (Fig 3a). The mutualistic endosymbiotic relationship between Bursaria truncatella and green algae (Chlorella) triggers faster movements under light [19], a condition enhanced by Trackoscope’s light microscopy setup, allowing Bursaria truncatella to reach peak speeds. Manually tracking such swift organisms without Trackoscope would be nearly impossible for extended periods.
(A) Bursaria truncatella tracked over 18 minutes. (B) Spirostomum ambiguum tracked over 12 minutes. (C) Blepharisma tracked over 10 minutes. (D) The violin plot shows the speed distributions of Spirostomum ambiguum, Bursaria truncatella, Blepharisma, and Actinosphaerium, illustrating the broad spectrum of speeds Trackoscope can handle.
Additionally, we monitored Blepharisma for up to 1.5 hours, observing speeds up to 500 μm/s (2 body lengths/second) (Fig 3c). We successfully tracked the entire process of asexual reproduction by binary fission in Blepharisma, including cytokinesis over 78 minutes (S3 Video) (Fig 4d).
(A) Stentor coeruleus assumes different geometries as it feeds (trumpet shape) and swims (spherical shape) over 25 minutes. While in the feeding position, different anatomical features such as the holdfast and oral pouch are also visible (S3 Video). (B) Actinosphaerium hunting Paramecium over 5.5 hours (S3 Video). Features such as axopods and the development of contractile vacuoles for digesting are visible throughout the track. (C) Deeplabcut analysis of Tardigrada as it slips on a petri dish over five seconds. The velocities and positions of the front two legs and mass center of the Tardigrada over the five seconds. (D) Binary fission of motile Blepharisma over 1.5 hours (S3 Video).
Spirostomum ambiguum exhibited ciliary movement along with multiple contractions, maintaining an average speed of 161 microns/second and demonstrating the most consistent speed distribution among the observed organisms (Fig 3b and 3d).
Tracking of Stentor coeruleus provided a window into predator-prey interactions and morphological changes during different locomotive states (Fig 4a). As it moves, Stentor coeruleus adopts a more spherical shape, and when feeding, it anchors itself with a holdfast and assumes its characteristic trumpet shape [20]. The Trackoscope’s high-resolution video capture enables clear visualization of feeding behaviors using the cilia clusters in its oral pouch (S2 Video). These observations are informed by the long-term tracking of Stentor coeruleus that Trackoscope enables. This extended tracking empowers researchers to observe subtle behavioral patterns, shedding light on potential mechanisms governing information processing, learning, and regeneration in these simple-celled organisms.
Observing slow-mover behaviors.
For slower unicellular organisms, we track Actinosphaerium eichhorni (500 μm) rolling and drifting with water currents and Amoeba proteus (450 μm) crawling. Actinosphaerium, with its sea urchin-like shape and numerous extending axopodia, reaches a maximum of 12 μm/min (0.02 body lengths/minute) as it advances slowly [21]. Over 5.5 hours of tracking, we observe the capture and digestion of 8 Paramecium caudatum (Fig 4b). The SI video also reveals how Actinosphaerium employs its axopodia to seize Paramecium and contractile vacuoles for encapsulation and digestion [22].
By tracking Amoeba proteus for 1.5 hours, Trackoscope discloses new insights that augment existing research. We track the dual locomotion of Amoeba proteus, crawling on a confined glass slide and swimming at the air-water interface (Fig 5) (S4 Video). Swimming proves five times faster than crawling (21 μm/s versus 4 μm/s), since swimming amoebas do not form pseudopodia for surface adherence, and water flow with surface tension enhances their movement. This observation is similar to existing literature [23, 24]. On average, swimming amoebas exhibit two pseudopodia, while crawling amoebas display five, indicating a significant difference in locomotion strategies. This behavior may assist pathogenic amoebas like Naegleria fowleri in traversing from water into the nasal passages and eventually the brain [25].
(A) Typical shape of Amoeba proteus swimming in a worm-like shape and visual of swimming setup. Track of Amoeba proteus swimming detailing speed and distance traveled over 1 hour. (B) Typical shape of Amoeba proteus crawling with multiple pseudopodia and visual of crawling setup. Track of Amoeba proteus crawling detailing speed and distance traveled over 1 hour. (C) Speed comparison of Amoeba proteus between swimming and crawling. Swimming has a higher average speed of 21 um/s while crawling has an average speed of 4 um/s. (D) Shape-changing tendencies based on the locomotion method through counting pseudopodia. While crawling, Amoeba proteus generate pseudopodia constantly while when swimming, only short pseudopodia are formed occasionally. (E) Pattern of a swimming Amoeba proteus capturing Paramecium over 1.5 hours (S4 Video). For this track, it was observed that every time the Amoeba proteus fed, it changed the general direction it was traveling in.
Observing complex organisms.
Trackoscope can be used to observe behaviors in complex, multicellular organisms as well; as examples, we track Tardigrada (Hypsibius exemplaris) (150 μm) and brine shrimp nauplii (Artemia salina) (600–1000 μm).
Long-duration tracking of Tardigrada in a plastic petri dish submerged in water demonstrates them taking periodic breaks across two instances for a total of 3.5 hours before moving on and continuing to explore during the 6.5 hours of tracking (S5 Video). We analyze the Tardigrada walking using DeeplabCut, a markerless neural-network-based gait tracking software [26]. Enclosed in a plastic petri dish, the Tardigrada’s legs slip on the smooth surface while attempting a tripod walking gait (S6 Video), revealing an average speed of 40 μm/s (Fig 4c). These observed stepping distances agree with prior research [27], underscoring Trackoscope’s potential for facilitating new analyses of locomotion with automated pose estimation methods. The tracking also captures movement strategies such as navigating on debris.
Tracking of brine shrimp nauplii emphasizes their chaotic and random movements. In the nauplii stage, brine shrimp are still adjusting to movement with front jointed antennae and their motion is not smooth as their phyllopods (swimming appendages) have not developed yet [28]. In the track, we observe circular lurching motion [29] as the nauplii moves at an average speed of 515 μm/s with a top speed of 1200 μm/s (1.2 body lengths/second) (S6 Video). As the nauplii attempts to move linearly, it tends to move in a clockwise circle suggesting that its left antenna is stronger than the right; when coordination is achieved, fast linear motion is observed.
Trackoscope’s role in enabling microorganism research in STEM
Trackoscope, with its economical design and tracking capabilities, empowers under-resourced laboratories and educational institutions to investigate a wide array of microorganisms, both slow and swift. This technology surpasses traditional microscopy by documenting detailed micro-structures and comprehensive behavioral patterns. The example organism tracks demonstrate Trackoscope’s capability to monitor organisms at velocities from 0.1 μm/s to 2 mm/s. With a vast tracking area of 325 cm2, Trackoscope uniquely facilitates the observation of natural organism behavior, minimizing the chance of organisms colliding with container walls and altering their paths. This “pond” environment could further enable experiments examining predator-prey dynamics.
Trackoscope’s customizable design for wider use
The Trackoscope design prioritizes both mass production and user customization. By utilizing laser-cut components for the base and motor mounts, the number of 3D-printed parts is minimized to eight, facilitating faster production. Additionally, the design allows for scaling the laser-cut parts to accommodate desired travel distances while maintaining compatibility with the majority of components.
The simple, modular approach to Trackoscope’s design also facilitates component modifications, enabling users to tailor the design to their specific needs. For example, acrylic components can be substituted with MDF (medium density fiberboard) to reduce costs, or the ring light and optics can be enhanced for improved imaging.
The design flexibility also extends to single-build customization. We built a version of Trackoscope that features a larger tracking area (625cm2, size of an A4 sheet) and is constructed from hand-cut poplar wood and 3D-printed brackets instead of laser-cut acrylic (S4 Fig). In this prototype we also included a motorized Z-axis which can enable tracking and automated control in the Z-axis. This version highlights the customizability of the Trackoscope platform through its design design for restricted tool access and added movement capabilities via a motorized Z-axis.
Future directions for Trackoscope enhancement
Looking to the future, Trackoscope could incorporate enhanced flexibility in tracking. One of the prototypes we constructed features a motorized Z-axis stage without Z-axis tracking due to excessive processing demands and slow performance. Introducing Z-axis tracking would enable longer observation periods, more consistently focused videos, and the collection of a third dimension of data for comprehensive analysis. Additionally, the use of economical linear rods introduces slight jerks at higher magnifications due to the actuation method. A modest investment in a more refined actuation system or reduction in platform size could mitigate this issue. Finally, a more tactile control of Trackoscope with a physical joystick can allow Trackoscope to be used as a standalone device without automated tracking capabilities and a connection to a computer running Python. This direct control will make it possible for a person to manually follow an organism while using the Trackoscope as a webcam.
Materials and methods
Construction of Trackoscope’s hardware framework
We designed the base parts using materials such as wood or acrylic, ensuring a straightforward construction process. To minimize costs and enhance design flexibility, we utilized 3D printing for custom parts. The system also incorporates Thorlabs’ cage plates and lens tube systems, enabling users to customize the base optics design by selecting from Thorlabs’ extensive range of compatible parts. We constructed the Trackoscope prototype, depicted in Fig 1 and S2 Fig, from laser cut quarter-inch acrylic. To demonstrate the customizability of Trackoscope, we also built a prototype with a larger tracking area, 625cm2 (size of an A4 sheet), from hand-cut lightweight poplar wood that also included a Z-axis (S4 Fig). All brackets are 3D printed using PLA filament at a 0.2 mm layer height. For a comprehensive list of materials, refer to S1 Table in S1 File. The assembly time for Trackoscope is approximately ninety minutes, as demonstrated in S7 Video and detailed in the assembly instructions in S1 File.
Software architecture and user interface
We developed the firmware for the Arduino Uno using Arduino IDE. The host computer’s software, written in Python, leverages libraries such as OpenCV, Matplotlib, imutils, and pyserial. We crafted the graphical user interface (GUI) with Python’s Tkinter library. For converting the Raspberry Pi Zero into a webcam, we installed a camera firmware directly onto the SD card [30]. More information can be found on GitRepo (https://github.com/bhamla-lab/Trackoscope).
Execution of organismal tracking experiments
We conducted most tracking experiments in 70 mm diameter Petri dishes, with exceptions for the Amoeba slide track and the Blepharisma and Bursaria tracks, which we carried out in an 8x8 cm square Petri dish. We sourced organisms from Carolina Biological Supply and maintained them at room temperature (24°C). Before tracking experiments, we diluted the cultures in spring water. We performed the tracking experiments on a desktop computer (Precision 3630 Tower with an Intel i7–9700 CPU, Nvidia RTX 2070 GPU, and 16 GB of RAM), which achieved tracking at 120 Hz. We also conducted tracking tests on a laptop (Dell XPS 13 with an Intel i7–10710U CPU, Integrated Intel UHD Graphics, and 16 GB of RAM), reaching a tracking FPS of 60 Hz.
Supporting information
S1 Fig. Trackoscope wiring diagram.
Wiring diagram for Trackoscope detailing the connections between various components.
https://doi.org/10.1371/journal.pone.0306700.s001
(TIF)
S2 Fig. The Trackoscope prototype.
(A) CAD model of Trackoscope in the raised sample imaging setup. A white or contrasting covering is typically placed over the sample to create a clean background in the image. (B) Trackoscope is designed to be mass-produced and is constructed primarily out of laser-cut parts (acrylic or MDF) and minimal 3D printed components. It also uses standard metric nuts and bolts to join components together.
https://doi.org/10.1371/journal.pone.0306700.s002
(TIF)
S3 Fig. Trackoscope movement analysis calculation.
(A) (Δxo/f, Δyo/f) are calculated by taking the organism’s location in the video frame at the start (is not always (0, 0)) and finding the displacement within the frame, , with Cpixels to μm depending on the magnification. (B) (Δxp, Δyp) are calculated by adding up all platform displacements throughout the track; for instance, a single data point would be calculated with
where
is the velocity of the axis on the platform.
https://doi.org/10.1371/journal.pone.0306700.s003
(TIF)
S4 Fig. Customized Trackoscope prototype.
(A) The single-build customized version of Trackoscope featuring a motorized Z-axis and a raised sample that is observed from below. This prototype has a tracking area of 625cm2 (size of an A4 sheet) and is built using limited tools (hand saw, 3D-printer, and screw-drivers). (B) The physical custom digital microscope system used in both prototypes.
https://doi.org/10.1371/journal.pone.0306700.s004
(TIF)
S1 Video. Demonstration of Trackoscope.
Trackoscope demonstration featuring the user interface and tracking of Blepharisma.
https://doi.org/10.1371/journal.pone.0306700.s005
(MP4)
S2 Video. Video quality.
Video quality of Trackoscope viewing Blepharisma with a 10x objective lens with visible cilia movement. Stentor with a 10x objective lens with visible cilia movement and visible organelles. Two Tardigrada interacting with another under a 10x objective lens. And Spirostomum captured using a 4x objective lens with dark field illumination.
https://doi.org/10.1371/journal.pone.0306700.s006
(MP4)
S3 Video. Unicellular organism tracks.
Video of tracking binary fission of Blepharisma over 75 minutes, Actinosphaerium hunting Paramecium over 5.5 hours, Bursaria swimming, Stentor feeding and changing shape, and bright field and dark field illumination comparisons of Spirostomum swimming.
https://doi.org/10.1371/journal.pone.0306700.s007
(MP4)
S4 Video. Amoeba proteus locomotion.
Comparison of Amoeba proteus as it swims at the air-water interface and crawls on a glass slide.
https://doi.org/10.1371/journal.pone.0306700.s008
(MP4)
S5 Video. Tardigrada locomotion.
6.5-hour track of a Tardigrada crawling around a petri dish and interacting with other Tardigrada and plant material.
https://doi.org/10.1371/journal.pone.0306700.s009
(MP4)
S6 Video. Multicellular Organism Tracks.
A Deeplabcut track of Tardigrada highlighting the tripod gait and limb recognition and a track of brine shrimp nauplii.
https://doi.org/10.1371/journal.pone.0306700.s010
(MP4)
S7 Video. Demonstration of Trackoscope assembly.
Assembly tutorial for Trackoscope.
https://doi.org/10.1371/journal.pone.0306700.s011
(MP4)
S1 File. Detailed parts list and assembly instructions (PDF).
More information can be found on GitRepo (https://github.com/bhamla-lab/Trackoscope).
https://doi.org/10.1371/journal.pone.0306700.s012
(PDF)
Acknowledgments
We thank all members of the Bhamla Lab for their feedback; Johnathan O’Neil for help setting up Deeplabcut; Alina Soifer for assistance.
References
- 1. Merces GOT, Kennedy C, Lenoci B, Reynaud EG, Burke N, Pickering M. The incubot: A 3D printer-based microscope for long-term live cell imaging within a tissue culture incubator. HardwareX. 2021;9:e00189. pmid:35492043
- 2. Li H, Krishnamurthy D, Li E, Vyas P, Akireddy N, Chai C, et al. Squid: Simplifying Quantitative Imaging Platform Development and Deployment; 2020.
- 3. Krishnamurthy D, Li H, Benoit du Rey F, Cambournac P, Larson AG, Li E, et al. Scale-free vertical tracking microscopy. Nat Methods. 2020;17(10):1040–1051. pmid:32807956
- 4. Hou S, Exell J, Welsher K. Real-time 3D single molecule tracking. Nat Commun. 2020;11(1):3607. pmid:32680983
- 5. Sridhar VH, Roche DG, Gingins S. Tracktor: Image-based automated tracking of animal movement and behaviour. Methods Ecol Evol. 2019;10(6):815–820.
- 6. Johnson C, Exell J, Lin Y, Aguilar J, Welsher KD. Capturing the start point of the virus–cell interaction with high-speed 3D single-virus tracking. Nat Methods. 2022;19(12):1642–1652. pmid:36357694
- 7. Silvestri L, Müllenbroich MC, Costantini I, Di Giovanna AP, Mazzamuto G, Franceschini A, et al. Universal autofocus for quantitative volumetric microscopy of whole mouse brains. Nat Methods. 2021;18(8):953–958. pmid:34312564
- 8. Dewan MAA, Ahmad MO, Swamy MNS. Tracking biological cells in time-lapse microscopy: an adaptive technique combining motion and topological features. IEEE Trans Biomed Eng. 2011;58(6):1637–1647. pmid:21278009
- 9. Sharkey JP, Foo DCW, Kabla A, Baumberg JJ, Bowman RW. A one-piece 3D printed flexure translation stage for open-source microscopy. Rev Sci Instrum. 2016;87(2):025104. pmid:26931888
- 10. Vos BE, Blesa EB, Betz T. Designing a High-Resolution, LEGO-Based Microscope for an Educational Setting. The Biophysicist. 2021;2(3):29–40.
- 11. Aidukas T, Eckert R, Harvey AR, Waller L, Konda PC. Low-cost, sub-micron resolution, wide-field computational microscopy using opensource hardware. Sci Rep. 2019;9(1):1–12. pmid:31092867
- 12. Tristan-Landin SB, Gonzalez-Suarez AM, Jimenez-Valdes RJ, Garcia-Cordero JL. Facile assembly of an affordable miniature multicolor fluorescence microscope made of 3D-printed parts enables detection of single cells. PLoS One. 2019;14(10):e0215114. pmid:31600202
- 13. Courtney A, Alvey LM, Merces GOT, Pickering M. The Flexiscope: a Low Cost, Flexible, Convertible, and Modular Microscope with Automated Scanning and Micromanipulation; 2019.
- 14. Steiner RA, Rooney TO. PiAutoStage: An open-source (3D) printed tool for the automatic collection of high-resolution microscope imagery. Geochem Geophys Geosyst. 2021;22(5).
- 15. Byagathvalli G, Challita EJ, Bhamla MS. Frugal Science Powered by Curiosity. Industrial & Engineering Chemistry Research. 2021;60(44):15874–15884. pmid:35757375
- 16. Lukežič A, Vojíř T, Čehovin Zajc L, Matas J, Kristan M. Discriminative Correlation Filter Tracker with Channel and Spatial Reliability. Int J Comput Vis. 2018;126(7):671–688.
- 17. Bradski G. The openCV library. Dr Dobb’s Journal: Software Tools for the Professional. 2000;.
- 18.
Open Broadcaster Software;. http://obsproject.com.
- 19. Mitarai A, Nakaoka Y. Photosensitive signal transduction induces membrane hyperpolarization in Paramecium bursaria. Photochem Photobiol. 2005;81(6):1424–1429. pmid:16033323
- 20. Marshall WF. Regeneration in Stentor coeruleus. Front Cell Dev Biol. 2021;9:753625. pmid:34660609
- 21. Watters C. Studies on the motility of the Heliozoa. I. The locomotion of Actinosphaerium eichhorni and Actinophrys sp. J Cell Sci. 1968;3(2):231–244. pmid:5661998
- 22. Suzaki T, Shigenaka Y, Watanabe S, Toyohara A. Food capture and ingestion in the large heliozoan, Echinosphaerium nucleofilum. J Cell Sci. 1980;42:61–79. pmid:7400244
- 23. Barry NP, Bretscher MS. Dictyostelium amoebae and neutrophils can swim. Proceedings of the National Academy of Sciences. 2010;107(25):11376–11380. pmid:20534502
- 24. Howe JD, Barry NP, Bretscher MS. How do amoebae swim and crawl? PLoS One. 2013;8(9):e74382. pmid:24040237
- 25. Cabanes PA, Wallet F, Pringuez E, Pernin P. Assessing the risk of primary amoebic meningoencephalitis from swimming in the presence of environmental Naegleria fowleri. Appl Environ Microbiol. 2001;67(7):2927–2931. pmid:11425704
- 26. Mathis A, Mamidanna P, Cury KM, Abe T, Murthy VN, Mathis MW, et al. DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat Neurosci. 2018;21(9):1281–1289. pmid:30127430
- 27. Nirody JA, Duran LA, Johnston D, Cohen DJ. Tardigrades exhibit robust interlimb coordination across walking speeds and terrains. Proceedings of the National Academy of Sciences. 2021;118(35):e2107289118. pmid:34446560
- 28.
Abatzopoulos TJ C J S P Beardmore JA. Artemia: Basic and Applied Biology. Springer Netherlands; 2002.
- 29. Anufriieva Elena V S NV. The swimming behavior of Artemia (Anostraca): new experimental and observational data. Zoology. 2014;117(6):415–421. pmid:25438935
- 30.
showmewebcam: Raspberry Pi + High Quality Camera = High-quality USB Webcam!;. https://github.com/showmewebcam/showmewebcam.