Corresponding author: Patina K. Mendez (
Academic editor: Pavel Stoev
Availability of 3D-printed laboratory equipment holds promise to improve arthropod digitization efforts. A 3D-printed specimen scanning box was designed to image fluid-based arthropod collections using a consumer-grade flatbed scanner. The design was customized to accommodate double-width microscope slides and printed in both Polylactic Acid (
Mendez PK, Lee S, Venter CE (2018) Imaging natural history museum collections from the bottom up: 3D print technology facilitates imaging of fluid-stored arthropods with flatbed scanners. ZooKeys 795: 49–65.
Specimen digitization efforts continue to be a high priority for natural history collections (
Arthropod imaging efforts are largely focused on pinned collections because challenges in imaging specimens stored in fluids is complicated by greater logistical difficulties, making their imaging a lower priority. Soft-bodied invertebrates such as larval insects, spiders, and adult aquatic insects are preferentially stored in ethanol or other preservatives so that shrinking during drying does not distort or obscure morphological structures required for identification. Vials are often stored within larger jars, and each vial often contains multiple specimens in a “lot” sometimes with free-floating dissections (e.g., genitalia) (Figure
Fluid-stored arthropods in
These curatorial activities during the imaging process results in longer handling times than pinned material when digitizing, thereby increasing digitization costs (
3D printing technology can be used to develop custom equipment for research and can take advantage of available high-quality laboratory materials. Custom laboratory and field equipment is common in the life sciences because the materials are low cost and can be designed for the study goals. Although 3D printing has been common in prototyping in many industries, its adoption in the life sciences has just recently emerged: beetle decoys made on 3D printers can be used to visually attract emerald ash borers (Domingue et al. 2015), life history behavior of insets can be monitored with custom 3D-printed equipment (
The goal of this project was to design a standardized 3D-printed piece of laboratory equipment to streamline imaging of small invertebrates stored in fluid collections of natural history museums. Specifically, we designed the system to image on a flatbed scanner using standard, large, optically-clear 75 × 50 × 1 mm glass slides mounted into a 3D-printed plastic box. We also provide a basic scanning procedure with image resolution settings and an R script to subset the batch image scans. We tested our system using fluid-stored material from the Essig Museum of Entomology. We imaged specimen lots of
To standardize the box size and reduce the need to cut custom glass, we designed the scanning box around a pair of large, double-wide microscope slides (75 × 50 × 1 mm, Fisher Scientific, Fisherbrand Plain Microscope Slides 12–550C). We selected this glass type and size because of its clarity, durability, relatively low cost (approx. $0.50/slide) and wide availability.
The imaging design approach addresses issues with imaging in liquid, specifically to eliminate reflections on the surface from overhead shots by scanning through a clear-bottomed box. This design is based on an improvised glass-bottomed VWR pipet box lid-scanning approach used by some researchers for imaging voucher materials as required for the BOLD project (
Our approach used 3D print technology because of its ability to produce precise, low-volume runs in a variety of materials. Using Autodesk Meshmixer 2.9 (
Fluid material scanning box:
To compare frame materials, we printed the design in multiple materials, colors, and finishes at Shapeways. Polylactic Acid (
To complete each box, we fixed the slide into the box with two applications of E6000 silicone adhesive. To apply each layer, we used a wide-tip syringe loaded with adhesive, carefully tracing the margins of the slide and air-dried the box for 24 hours. E6000 shrinks when it dries and we found that two applications prevented most leaks.
We tested the completed scanning boxes by filling each box approximately 1/3 full of 75% ethanol. We examined the boxes every 20 minutes over the first hour, and then hourly thereafter for 4 hours. We recorded any leaking, softening of the material, staining, fading of colors, or odors.
The scanning workflow consisted of moving the specimen to the scanning boxes, placing them on the scanner, arranging and covering the specimens, acquiring a batch image, and then cropping the images in software (Figure
To improve the quality of the scans, we arranged specimen strategically and added supports in the specimen box. For larvae, we lined up the caddisfly cases, but did not remove larvae from cases, and for adults with genitalic dissections, we often paired the body and dissections within a ring (a plumbing slip joint washer) to keep the material from floating and to serve as a standoff for the background (Figure
To image specimens efficiently, and reduce overall handling time, we imaged batches of 6 boxes (Figure
Scanning process images:
To determine the most appropriate resolution and file size to image batches, we scanned a representative batch at 600 dpi, 1200 dpi, and 2400 dpi, compared file characteristics and determined the resolvable structures at each resolution. To determine resolvable structures, we compared images of
Imaging technicians worked in teams of 2–3 for 3-hour scanning sessions with a set of 12 scanning boxes in two batches of 6. 1–2 technicians loaded batches of scanning boxes by working one vial at a time to load the label panel and the specimen panel. Fluid-stored material in the Essig Museum of Entomology is stored in thin, long shell vials (mostly 8 × 90 mm and 12 × 90 mm) stoppered with a firm piece of cotton and inserted cotton-side down in wide-mouth pint jars of ethanol. Each sample scanned required that a vial was selected from pint jar, the cotton was removed, and the contents of the vial emptied into the scanning box, and the vial rinsed with ethanol to make sure all small parts were moved to the scanning box. A batch was then passed to a technician who operated the scanning station.
The scanning technician moved the boxes to the scanner, arranged and covered the specimens, previewed the scans and then completed the high-resolution scans. While the scanning technician scanned the batch, the loading technicians either loaded a new batch if it was the first run, or unloaded the batch that came off the scanner, and then loaded the new batch. Unloading the scanning boxes required technicians to carefully return the material and labels (including the new 2D DataMatrix code label) to the shell vials, completely search the scanning boxes for small parts, refill ethanol, re-stopper the vials with cotton and return them to a pint jar. As part of this process, we often curated the material by replacing ethanol, replacing chipped shell vials, and re-copying faded labels. But, we avoided replacing all ethanol in every sample to reduce the chance of losing small parts of specimens. The estimates for the handling time include these activities.
To process the 6-box batch image into a single file for each image, we bulk-processed the 6-panel images using an R script (available from:
We later manually renamed files to the Essig Museum naming convention (e.g., EMEC41255
Of the three materials tested,
3D materials and response to 4-hour ethanol test.
Material | Material Information | Performance in EtOH test |
---|---|---|
|
Fused Deposition Modeling Technology. Heat-extruded, hard plastic, unpolished. Lines visible between layers. | No odors, softening, leaking, or creep of ethanol. No color change to model or ethanol. |
|
Inkjet-like process, requires support material when printing, unfinished or polished. Unfinished noticeably rougher (layer lines) than polished nylon. | No odors or softening. Creep of ethanol up and over side. Extreme leeching of color into ethanol in black and orange models. |
Scanning time and file size increased with increasing dpi setting on the scanner (Table
Cropped scan and scan resolution comparison:
Scanning details and resolvable structures of
Scanning resolution (dpi) | Scanning time | Dimensions (pixels) | JPG file size | TIFF file size | Resolvable Structures in adults |
---|---|---|---|---|---|
600 | < 15s | 5091 × 3714 | 3.4 MB | 83.2 MB | Tibial spurs, wing venation major veins, abdominal shape to determine sex, large male genitalic structures. |
1200 | 1m 10s | 10183 × 11428 | 10.2 MB | 333 MB | The same structures @ 600 dpi, but greater detail on shape and count of tibial spurs, detail in wing pigment, shape detail on male genitalic structures, shape of maxillary and labial palp segments sometimes possible to discern. |
2400 | 3m 40s | 20366 × 22856 | 34 MB | 1.29 GB | The same structures available @ 1200 dpi, but also segments of antennae, individual setae on the head and abdomen, and additional detail on male genitalic structures. Tibial spines, ocelli, and setae on wings was also discernable for some taxa. |
For larvae, materials and structure of the case were identifiable (Figure
Scan quality decreased when scanner glass and exterior surfaces of slide glass were not cleaned and dried between sessions resulting in condensation on the slide surface. Yellowed ethanol also made for poor scans. When specimens were large (e.g., cased limnephilids), scans were darker than shallower scans. Dark-bodied specimens often lacked detail.
Example scans of different species and life stages of Essig Museum of Entomology
When technicians worked in teams of two or three, they scanned 3–4 batches/hour (12–20 mins per batch) resulting in a rate of ~7 mins/vial/worker from the unloading step through rehousing the material, including routine curatorial activities. Although increasing the team to three members increased the speed of loading and unloading, the scanning step was often the time limiting step in the process because the specimen had to be arranged and previewed before scanning.
Large file sizes increased image processing times often exceeding the available memory on our scanning computer and we had to perform most image processing on a computer with higher specifications (Razer RZ09-01302E21, 2.60 GHz Intel Core i7-4720HQ, 16GB Ram, Windows 10). Processing took ~1.5 mins to cut and rotate the 6 individual scanning box images per batch resulting in individual files of 5MB each (~3.5–7.5 MB). Alignment issues occurred occasionally when the marquee was misaligned in the bulk scan as a result of technician error. These errors were sometimes correctable by realigning and hand-cropping in image editing software.
Using 3D print technology in research workflows is an opportunity to reduce costs for customized laboratory equipment as well as provide standardized and reproducible components available anywhere that it can be printed. For imaging arthropod natural history collections, low-cost
Valued for its wide availability on consumer-level 3D printers,
We hypothesize that lactic acid resulting from the breakdown of
Nylon (polyamide) serves as a viable alternative because of its wide availability and chemical resistance for the non-dyed models. However, we did not have access to printers that printed in nylon other than through web vendors, limiting our ability to inexpensively prototype and test, making it the more expensive of the two materials. Nylon scanning boxes remained stable after 1.5 years of use in the laboratory.
The efficiency of imaging through a scanning approach was similar to other fluid-based approaches yet required double the handling time of pinned material. Volunteers working in teams of 2–3 processed vials at a rate of ~8 vials/worker/hour (~7 mins per vial + <1min/image of digital processing) from unloading and positioning specimen to scanning and rehousing. This rate is similar to imaging samples from ichthyology and herpetology museums, where 41% of the museums reported 6–10 mins/sample and 22% reported 11–15 mins/sample (
Compared to whole drawer imaging approaches such as SatScan (
Scanning at 2400 dpi created high-quality scans appropriate to resolve some taxonomically-informative structures in many groups of
Scans have the potential to be used beyond identifications and museum record-keeping. Wing measurements derived from imaging pinned specimens with a camera lens are not reliable for morphometric analyses, especially for larger specimens where error increases if the wing is not aligned with the measurement plane or if there is distortion from the lens used to acquire the image (
Although 2400 dpi produced the highest quality scans, the final selection of the scanning settings should be determined based on the desired utility of the scan and how the image files will be stored. For natural history museums, if the scans are primarily a snapshot of the material that provides visual verbatim label information and the count and condition of loan material, a lower resolution and smaller file size may be more desirable. However, acquiring lower resolution scans may not measurably speed up the imaging process because the handling time per vial is so high. If wider utility is desired, such as for identification, teaching, or morphometrics, higher resolution scans are worthwhile. We approached this project from the perspective of wanting to determine the highest quality scans we could get though the scanning method. In our case, the size of the Essig
The scanning box applications are not limited to museum imaging and this method may serve as a tool that can be used more widely to archive aquatic ecology specimens. For example, samples taken for biological monitoring are not often stored long term, representing a lost opportunity to permanently archive life history condition (e.g., body size, life stage) of aquatic taxa which may ultimately be useful in long-term ecological studies. Images may also be able to provide easily examinable records when physical material is unavailable or as part of a workflow for QA/QC by remote reviewers or experts. Other fluid-based bulk samples such as those from malaise or pitfall traps can be imaged and sent to experts to share samples. This method is appropriate for image-based vouchering to accompany DNA sequences such as those archived by BOLD (
The design size for this fluid-based application for natural history arthropod collections was based on the wide variability in label sizes for fluid-based collections. Compared to pinned collections, which usually have much smaller labels (e.g., 1.5 cm x 0.75 cm) to fit under the insect body to maximize the number of individuals that fit in a museum drawer, fluid collections have no such constraint. The size of vials used for storage in collections range from wide scintillation vials with caps, to stoppered vials, to very thin shell vials. Many labels tend to be larger because of the need to write by hand with pencil and archival pigment inks to prevent loss of label information in ethanol. The double wide museum slide best accommodated the label and specimen lots of collection material, yet a standard slide sized scanning box may be more appropriate for other collections.
The scanning boxes are a 3D printable design, and with software the design can be modified to change the dimensions of the scanning box for alternative imaging purposes. Smaller scannable areas, such as those designed around a standard microscope slide or even microscope cover slips, may be viable for digital archiving, especially for projects such as BOLD vouchering where label information is not required and the frame needs to only accommodate one specimen; the 3D-printed boxes may be single and mobile, or have a fused layout for batch imaging could match indexing of DNA samples to reduce laboratory errors. The number of images captured would be dependent on the printing size capabilities of the 3D printer and the overall scan area. Larger area scans, such as for large benthic samples or ichthyology, are also possible but there may be challenges in finding high-quality glass that is thin enough not to reduce image quality and inexpensive enough to be used in this application.
The value of developing a standardized method and equipment for imaging fluid-based museum material should not be underestimated. In natural history collections, freshwater aquatic species, with the exception of odonates, are some of the least documented with digital images, mainly restricted to images of pinned adult forms. Most museums have not embarked on digitizing fluid material, and this method represents a clear opportunity for a low-cost approach that can be integrated into existing workflows. The scanning box and laboratory workflow performed well with few leaks and used a simple approach for technicians. Compared to other systems, the components are inexpensive, available, and replaceable. Boxes can be printed in-house or ordered with the cost of 12 boxes (2 batches) for well under $400 USD and can be used with existing scanners and computers. Because the design is fairly simple and designed to make use of available laboratory materials, it can be made to accommodate smaller slides or customized to the imaging needs of the museum or for other research needs.
We thank Milo Fedane, Jessica Garcia-Reyes, Gregory Gladkov, Chloe Kapanen, Stella Li, Soledad Soto, Sonja Stott, and Michelle Yang for scanning and organizing material. Kipling Will and Charles Griswold encouraged pursuit of the project. We thank Vincent H. Resh for laboratory facilities, Peter Oboyski for access to the Essig Museum