In this work, we presented the design of the NTDscope, a portable, multi-contrast microscope that enables diagnosis of neglected tropical diseases at the point-of-care. We demonstrated its imaging capabilities for microfilariae in peripheral blood, Schistosoma eggs in urine samples, and STH eggs in processed stool samples. We also showed that the device is capable of reading molecular assays by imaging both lateral flow assays and fluorescence-based CRISPR-Cas assays. Sample preparation is simplified for these applications by the use of disposable plastic capillaries, including rectangular straight channel capillaries for defined blood volumes, tapered capillaries for filtering eggs from urine and floated stool samples, and as well as a prototype capillary for thick blood smears that is compatible with the NTDscope.

As new diagnostic applications are developed for the NTDscope, new capillary designs that fit the carriage dimension and focal plane can be designed. The original straight channel capillary was designed to easily load a peripheral blood sample for quantification of microfilariae, but it can also be loaded with other liquids for other uses. Previous applications have included imaging of parasitic eggs isolated from a flotation solution using the Fill-FLOTAC or Fecalyzer devices, as well as imaging of diatoms in marine samples for educational purposes. The tapered capillary that was designed to isolate Schistosoma eggs from urine has also proven useful for other applications that require size-exclusive filtering and concentration, such as separating two different species of STH eggs.

The prototype thick smear capillary demonstrates potential for integrating common clinical laboratory assays into the device, including wet mounts for protozoa and fungi, three-part blood counts, and parasitic egg viability assays [52, 53]. The ability to capture images of live parasites has the potential to provide new information about infections and response to treatment. For example, NTDscope videos of Loa loa microfilariae motion inherently capture only live parasites rather than dead microfilariae that could appear in conventional thick smear microscopy. In the case of low parasitemia, additional FOVs and multiple capillaries from a single patient can be analyzed, which could improve sensitivity for treatment stopping decisions in near-elimination settings.

The diagnostic power of the NTDscope is enhanced by integrating machine learning (ML) for automated analysis of patient samples. Importantly, use of ML can significantly reduce the time required for sample analysis, including manual counting of parasites that can take a minimum of 10 minutes per slide for well-seasoned technicians [54] and can have high variability [55, 56], reducing fatigue as well as increasing throughput. Use of the Android OS environment allows existing ML models designed for mobile applications to be easily run on the device. Furthermore, the portability of the NTDscope simplifies data collection for NTDs, increasing the amount of available training data for ML models, which is particularly important for capturing patient samples in NTD-endemic regions where microscopes with digital cameras are unavailable. Beyond classification of parasites in static images, ML models have the potential to differentiate microfilarial species based on movement signatures, which we have observed to be different for M. perstans, B. malayi, and Loa loa microfilariae (Fig 2). As more field data is collected, ML algorithms could be developed to distinguish between multiple filarial species. For some diagnostic uses cases, visualizing an ML-highlighted parasite on the screen of the NTDscope could help to build confidence in the tool for both providers and patients.

The multi-contrast imaging capabilities of the NTDScope could prove useful for future diagnostic applications that involve species differentiation, such as for distinguishing soil transmitted helminth species and S. mansoni eggs in stool samples. Darkfield imaging has proved beneficial when used for ML-based identification of S. haematobium eggs, where models trained on darkfield images performed better than those trained on brightfield images [34]. Darkfield illumination is particularly helpful when distinguishing eggs from other debris found in samples, which is typically achieved through staining, requiring extra steps and assay components. Given that many parasitic eggs and worms exhibit autofluorescence [57–60], the NTDscope could further facilitate speciation with its fluorescence capabilities.

The fluorescence capabilities of the NTDscope provide the potential to replace bulky laboratory equipment, such as plate readers for molecular assays. Fluorescence-based molecular assays, such as those based on CRISPR-Cas enzymes, are of growing interest for infectious disease diagnosis as they avoid the need for conventional PCR machines and temperature cycling (amplification-free CRISPR-Cas reactions are isothermal). CRISPR-Cas diagnostic assays have been demonstrated for the detection of neglected diseases including Schistosomiasis [61, 62], Trypanosomiasis [63], Leishmaniasis [64], Malaria, [65], Zika, Dengue [66], and others [67–69]. These reactions can be read using portable devices [51, 70], and could be adapted to be read using the NTDscope. We also showed that the NTDscope can be used to read the results of lateral flow assays. While multiple portable LFA readers exist [71–76], the potential to read test strips with a magnified imaging system could improve sensitivity in cases of light infection, help digitize results for future use, and open the possibility of combining multiple tests onto a single test strip, reducing the quantity of reagents needed [77–82].

Presently, the main limitation of the NTDscope lies in device availability. The NTDscope presented here is a prototype built to demonstrate the feasibility of a fully-integrated diagnostic device for multiple NTDs, and 80 devices were built in a limited manufacturing run. Collaborators in Cameroon, Gabon, Côte d’Ivoire, and Bangladesh continue to work with the device to establish efficient workflows for expanded use, as well as explore additional diagnostic and disease mapping applications. With additional feedback from ongoing field work, future versions of the device have the potential to be manufactured or assembled in the countries where they are used. The NTDscope represents an important step toward increasing access to disease testing in regions with limited healthcare infrastructure through portable diagnostic technology, a direction that may also benefit regions beyond those with NTDs.