MoleMapper™ Enables Research
Research is hard. Research is expensive. The MoleMapper™ project, with the help of all those who participate, works to make both of those things better. Computer science is currently teaching computers to recognize cars, cats, and people in images. We, and our research partners, hope to eventually use those techniques to detect suspicious lesions in MoleMapper™ images. To get there, we have a lot of questions that require mountains of data to be able to answer.
Research often happens in waves. The initial release of MoleMapper™ (2.0) allowed OHSU and Sage Bionetworks to collaborate on proving the efficacy of using smartphone capabilities as a research tool empowering citizen-supported science. This resulted in a paper published in Nature’s Scientific Data journal titled, “The Mole Mapper Study, mobile phone skin imaging and melanoma risk data collected using ResearchKit.“ In that report, we demonstrated the efficacy of using smart phones to collect data and perform relevant scientific analysis on it.
With the advent of version 3.0 for the iPhone, we are embarking on the next wave of research. We expect to generate more and better data to be used to answer many important questions. The answers to these questions will help us understand both the future capabilities and limitations of using smart phones as potential aids to patients and the medical community seeking to provide quality healthcare.
Researchers are innately curious people and have a never ending list of questions. Here are just a few of the many things we’re eager to answer. Many of these are precursors to the big question the entire community is seeking to answer: can we detect dangerous moles while they are still easily treated?
- Are there techniques we can use to reliably detect mole sizes without using coins or other reference objects?
- Can we reliably co-register two-dimensional images of three-dimensional body regions in a way that accurately detects changes to entire “zones”?
- Are the cameras on iPhones (in particular) sufficient to capture the types of images needed to reliably perform classification tasks on?
- Do the ways typical users photograph their moles result in sufficiently clear images to train computers with?
Participating in Research
To participate in MoleMapper™ research, Principal Investigators first need to register with Sage Bionetworks as a Certified User and then sign a data use agreement with the MoleMapper™ team at OHSU. Your research has to be related to the work on identifying potentially dangerous lesions but can include a diverse array of topics from classification tasks using Deep Learning techniques to lesion segmentation and image registration challenges.