WikiJournal Preprints/Design and validation of an open access, 3D printed dermatoscope

Introduction
Distributed digital manufacturing of free and open-source medical hardware is gaining momentum. The increased availability of 3D printing technology, coupled with the growing number of enthusiasts and small enterprises utilizing these technologies, has led to local small-scale production of various medical devices. Some of these productions have been proposed to address crises such as war zones, disaster-stricken areas, and more recently, shortages caused by the COVID-19 pandemic. In these situations, simple, low-tech, and easy-to-build medical items like stethoscopes or otoscopes can be crucial in providing adequate medical care to those in need. In developing countries, shortages of medical equipment may be more pervasive than occasional, resulting in limited access and deep inequalities.

Hobbyists and Makers worldwide possess the technology and skills to design and assemble complex and reliable prototypes for medical use. While their reverse-engineering capabilities have shown increasing success in building advanced devices such as robots or moving prosthetic arms in recent years, there is a need for the earlier development of validated models for various basic medical equipment pieces.

The aim of this paper is to describe the prototyping process and evaluate the performance of an open-source, 3D-printed, low-cost dermatoscope compared to a commercial solution. We specifically chose this medical device for several reasons. Its design has undergone minimal evolution, and it can be easily reproduced using commonly available electronics and magnifying systems for Makers. The cost of commercially available, medical-grade devices remains high, ranging from a few hundred to several thousand euros. This poses a barrier for general practitioners and emergency doctors, resulting in only a minority of them adopting this tool in their practice. By openly publishing this prototype, we hope to enable others to replicate and potentially improve it.

Material and methods
Our dermatoscope was designed with the goal of maintaining quality standards similar to current professional devices while significantly reducing production costs. The entire design was created using a free online CAD tool called TinkerCAD (www.tinkercad.com, Autodesk Inc., San Rafael, California, USA), and it was manufactured using a Prusa i3 pro B FDM 3D printer (Geeetech Ltd, Shenzhen, China) and an SLA Mars Pro (Elegoo Inc., Shenzhen, China). Detailed building instructions can be found in Appendix A.

A dermatoscope consists of two main parts: a head that houses the lighting and magnification systems, and a battery compartment for handling the device. We made the decision to design the head and handle separately. For the head, we utilized the same battery compartment as a low-cost 3D-printed otoscope that we had previously developed. This allows for multiple heads to be used with the same handle, easily transitioning from an otoscope to a dermatoscope. In case of irreparable damage, the heads can be replaced. The development of a shared platform for multiple devices could lead to a reduction in overall production costs, a concept also employed in the car industry.

To facilitate component retrieval, the power for the dermatoscope is supplied by two common AA batteries, connected using ball pen springs and a layer of tin.

For the lighting system, we chose to use 8 white LEDs, which are typically already available in a Maker's toolbox. To ensure water resistance, the LEDs need to be placed in a transparent enclosure. While FDM clear filaments are an option, we opted for stereolithography resin to guarantee waterproofing of the head.

In our design, we incorporated an adjustable optical block consisting of plastic Fresnel lenses, which are affordable and easily obtainable from e-commerce websites. To facilitate skin contact and enable the epiluminescence effect after applying oil, we added a plexiglass layer. The internal face of the plexiglass features an engraved millimeter scale for easy measurement of detected skin lesions. By utilizing these components, we were able to keep the costs around 5 euros (Table 1).



An alternative to Fresnel lenses is to use a printable optical lens system. Theoretically, one could take advantage of the high-resolution stereolithography printers provide to produce the optical block of an entire photographic camera. However, the post-printing process required to achieve sufficient visual quality is complex and time-consuming, and it relies on specific and expensive optical resin. Using stereolithography with standard (non-optical quality) clear resin to manufacture the entire device increases the overall cost, as the resin is sold at around 30€/L. On the other hand, for fused deposition modeling, 1 kg of ABS filament can be purchased online for around 15€. The estimated average power consumption is 0.05 kWh for FDM printing and 0.03 kWh for stereolithography.

Results
Medical literature does not provide standardized parameters for comparing the performance of dermatoscopes. Typically, these devices are described in terms of technical features such as magnifying power and light intensity. In our study, we compared our low-cost prototype to an entry-level commercial solution (dermatoscope Gima 2000, G.I.M.A. S.p.A., Milan, Italy) using these parameters, as well as evaluating the field of view and color quality.

The 3D printed dermatoscope functions in the same way as traditional dermatoscopes. Healthcare professionals simply need to turn on the light using the switch and gently but firmly press the dermatoscope head against the patient's skin. When the epiluminescence effect is desired, oil should be applied since the plexiglass interface is not polarized. Adjusting the helix provides focus adjustment when needed, and removing the lens holder allows for changing the magnification level from 6x to 3x.

Our analysis revealed some differences between the prototype and the certified dermatoscope. Fresnel lenses can only provide limited magnification power without distorting the image. The prototype was able to achieve 6x magnification, while the glass lenses of the commercial dermatoscope provided 10x magnification without chromatic aberrations. However, it's important to note that magnification alone does not guarantee clarity of vision. For this reason, we tested both devices using an imaging-quality millimeter scale designed for comparing photographic lenses. Both our prototype and the Gima dermatoscope were able to clearly visualize the gaps between the smallest test lines, which were placed at a 0.2mm distance. This, combined with a comparable field of view (prototype: 25 mm; medical device: 26mm), resulted in a smaller but clear image that was sufficient for evaluating the details of most skin lesions, which are typically at least 10 times larger than the measured minimum resolution.

The total amount of light emitted by the instruments was measured using a professional exposimeter (Bowens flash meter III, Sekonic Electronics Inc, Japan). Our prototype emitted 5000 lux, significantly higher than the Gima dermatoscope, which only produced 1400 lux. The commercial dermatoscope was equipped with a dimmer to adjust the amount of light according to the characteristics of the detected skin lesions.

There was also a difference in color temperature. Our white LEDs emitted a cool light (7600K), whereas the halogen lamp of the certified dermatoscope consistently emitted a warmer light (4700K). Although LED dermatoscopes have become the industry standard, it remains unclear whether the disparities in color temperature can interfere with diagnosis. It is a common experience for doctors to have to adapt to the technical specifications of a new instrument, and to address this issue, attempts have been made to standardize imaging.



Color comparison
To compare the performance of the two devices on skin lesions of different colors, we utilized a color scale chart - specifically von Luschan's skin chromatic scale - which has a high correlation with skin color evaluation conducted by a reflectance spectrophotometer. We printed a copy of the scale and captured digital images of its 36 items using both devices at their brightest lamp setting. Subsequently, the images were analyzed using Photoshop CS6 (Adobe Inc., San Jose, California, USA), which provided a 3-point RGB component profile for each image, enabling us to examine each color individually. We compared these data with the color profiles obtained directly from the chart and computed the Pearson's correlation coefficient for each color and device. The results demonstrated a strong correlation for every comparison, with the prototype exhibiting a higher correlation than the Gima dermatoscope. For example, in the "green" channel, the correlation coefficient (r) was 0.87 when comparing the certified medical device to the chart and 0.97 when using our prototype. Similar patterns were observed for the other channels: the "red" correlation coefficients were 0.73 and 0.96, and the "blue" coefficients were 0.80 and 0.89, favoring the prototype. It is important to note that correlation should not be interpreted as agreement.

To better visualize the differences between the two measurement instruments, we employed a Bland-Altman plot. This graphical tool is typically used to compare laboratory diagnostic tests, plotting the means of each pair of measurements on the x-axis against the percentage of differences between the measurements on the y-axis. In all the graphs, an increase in variability of the differences was observed as the magnitude of the measurement increased. This was likely caused by an elevated level of reflections produced by the brighter items on von Luschan's scale. The trend line consistently showed a positive slope: for lower (darker) values, both devices produced brighter colors compared to the chart values, while for the brightest values, the lines converged towards zero, indicating a more accurate color representation. The majority of measurements fell within the confidence intervals, suggesting that both devices yielded accurate results. However, this was not the case for the "red" channel. The data from this component did not follow a normal distribution, which is a requirement for the Bland-Altman plot. We identified the cause as the brightness setting chosen for the measurements, as depicted in the graphs where many values reached 255, the maximum value. To overcome this issue, we plotted the Bland-Altman graph using the 2.5 and 97.5 percentiles. These graphs closely resembled those of other colors, with linearly increasing measurements.

Overall, we can confidently state that both devices provided an accurate representation of the colors included in the skin chart, albeit with some differences related to the color temperature of the lighting system used.

Discussion
By utilizing 3D printing and leveraging Makers’ prototyping expertise, we successfully designed and fabricated a functional dermatoscope at a cost of approximately 5€. Currently, branded dermatoscopes are expensive, which can pose a barrier for clinicians, particularly in non-specialized contexts. Given the well-established working principles of this device, it was relatively easy to implement the design using modern electronics. The cost of our instrument was 40 to 80 times lower compared to professional options. Our project implemented a modular platform, enabling interchangeability between different specialized heads, further reducing costs for small-scale production. While the performance of our dermatoscope was similar to entry-level commercial solutions, it did have some limitations. The magnification power was lower compared to glass optics, as the Fresnel lenses only achieved a 6x image enlargement instead of the professional solution's 10x. However, our dermatoscope was capable of detecting details as small as 0.2mm, making it suitable for most skin lesions. Color profiles were similar between devices, but some differences emerged at extreme brightness levels, deviating from the reference table. This can be addressed by incorporating a dimmer, which is already present in the certified device. In our case, Makers could easily refine our design by, for example, adding a potentiometer.

To ensure safety, every possible measure to prevent harm to patients should be taken. For instance, the materials used should be washable and antibacterial. FDM 3D prints are not waterproof or sterilizable, and stereolithography resin is toxic in its liquid form. However, the dermatoscope only makes contact with the patient's skin through the central part of its head, where an easily cleanable and non-toxic glass/plexiglass square is positioned. When electronics are involved, the final product should be waterproof to prevent potential short circuits. Only a harmless, low voltage (3V) current is used. Therefore, we can assume that the safety criteria are met.

When considering emergency contexts, dermoscopy may not immediately come to mind. So far, the focus of developing 3D printed instruments has been on items such as stethoscopes and prosthetic arms. However, one of the lessons learned from the COVID-19 pandemic is that ordinary and chronic diseases may be neglected throughout the duration of an emergency, compromising patient management and increasing mortality rates. Increased availability of these instruments could prove beneficial in situations where the duration of the emergency is unknown.

Conclusions
The proliferation of 3D printed instruments has the potential to reduce inequalities by providing more accessible medical diagnostic tools. With this technology, general practitioners and primary care doctors can readily obtain medical equipment that is typically reserved for specialists. While general practitioners may not possess the specialized training required for complex diagnoses, they can use these instruments as a preliminary assessment tool, enabling them to differentiate between common diseases and those that require specialist attention. Without such instruments, primary care doctors might unnecessarily seek specialist consultations, leading to increased costs for patients and healthcare systems. Further research should focus on designing and validating additional equipment. The dermatoscope we have developed appears to be adequate for initial clinical evaluations. However, the overall quality of self-manufactured devices may vary depending on the available materials and assembly proficiency. Whenever possible, it is advisable to prioritize medical-grade certified instruments over self-made solutions.

Competing interests
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Ethics statement
No human or animal subjects were involved in this research.