Digital photography is now available on most smartphones with widespread access globally and has been more commonly used in clinical photography in recent years. This is especially true during the recent COVID-19 pandemic, resulting in dramatically increased utilization of telemedicine. Although most clinical photographs are now obtained with smartphones, dermatologists and plastic surgeons seek a more sophisticated measurement reference standard (e.g. over a ruler or a coin), especially with evolving efforts in enterprise imaging (EI). 

 

Skin lesions and anatomic deformities often occur in planes different from a flat 2D ruler, leading to inaccurate size assessment estimations that could be mitigated with a 3D reference tool to optimize off-plane measurement precision. 

 

We propose applying interior orientation photogrammetry (method of similar triangles) and parallax to clinical photography for consistent measurements across time and among patients with various pathology. This should also be helpful for growing publicly available databases for machine learning training and testing. 

 

Knowing camera (receptor) distance and perspective to skin lesion or anatomic region (source) with a 3D reference standard, should allow for more precise quantification (e.g. pixels/mm) in 3D. We designed this cuboid reference matrix for a proposed clinical photography standard to assist in 3D space mapping and temporal registration in clinical photography as a scaling reference. Tsai’s grid for camera calibration was considered, however, we believe our cuboid approach would be more generalizable and an improvement over 2D rulers often (but not universally) included in clinical photography. 

 

Each perspective of the cuboid reference is unique depending on source-receptor distance and divergence angle [Do HM, Folio LR, et al] making 3D information potentially computer vision machine readable, similar to a 3D QR code sometimes used in virtual multi-modality (ultrasound, MRI, augmented robotic vision) surgical environments.

 

The design is simple and intuitive with each side a different integer measurement as follows: 2.0 cm x 3.0 cm x 4.0 cm, with each side also having a different number of windows, 2, 3 and 4, respectively (to maximize unique perspectives.  To further differentiate perspectives, we included three fiducial markers on one side (one square, two spherical) to further assure distinguishing angles, minimizing chances of perspective ambiguity as seen in the Necker Cube [Donaldson, J. et.al.] If a human can perceive two opposing views the same, a machine likely would as well. 

 

Analogous to 2D QR codes that leverage unique optical patterns, our proposed 3D cuboid reference could serve as a measurement standard for dermatology, plastic surgery and whole-body imaging. Since camera angle and distance can be calculated based on unique patterns derived from the 3D cuboid, more precise lesion measurements and improved determination of anatomic relationships may be possible. 

 

Different patterns visually indicate they would be machine-readable, and distance quantification from the center should be possible with geometric algorithms, allowing for more precise measurements and/or registration of lesions using the method of similar triangles. If you are interested in helping determine perspectives quantitatively, we established a 3D coordinate system with the square fiduciary marker as 0,0,0 (see attached drawing).  

 

We are making our cuboid design open-source/publically available that can readily be made on a wide range of printers, affording an opportunity to crowdsource testing and design optimization. We envision patients followed by dermatology may use a standard such as this cuboid when taking photographs of their skin lesions, allowing for temporal registration and relative off-plane measurements that are not possible with 2D rulers to effectively monitor therapeutic response.  

 

Statement of Impact: A 3D reference standard should provide more precise measurements of skin lesions in clinical photographs; especially when compared over time or when registering images digitally.

 

References:

Huy M Do, Andrew J Henderson, Veronica M Rotemberg, Raisa Z. Freidlin, Les R Folio. "3D QR code-like cuboid as a proposed measurement standard for visible light enterprise imaging in dermatology, plastic surgery and whole-body photography" SIIM Research Abstract Presentation. June 2020 (pending).

 

Projection perspective: https://www.cse.unr.edu/~bebis/CS791E/Notes/PerspectiveProjection.pdf 

Donaldson, J. and Macpherson, F. (July 2017), "Necker Cube" in F. Macpherson (ed.), The Illusions Index. Retrieved from https://www.illusionsindex.org/ir/necker-cube

 

If you find this useful or have an application or design ideas we would like to hear your inputs. We also anticipate a challenge to determine perspective differences more quantitatively, eventually by machine learning.

 Feel free to contact the lead author at les.folio@nih.gov if you find this useful, or if you have questions or suggestions. 

 

To use the cuboid photo reference, just include in photos where you think size will be important (all clinical photography, some architectural and manufacturing). Knowing the dimensions (2x3x4 cm) allows for extrapolation of size to the area of interest. 

Thank you.