6533b81ffe1ef96bd1277273
RESEARCH PRODUCT
Color constancy in dermatoscopy with smartphone
Boštjan LikarFranjo PernušBlaž CugmasBlaž Cugmassubject
:MEDICINE [Research Subject Categories]Computer scienceDermatoscopesComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION02 engineering and technologysmartphone01 natural sciences0202 electrical engineering electronic engineering information engineeringmedicineComputer visionDermatoscopyColor constancymedicine.diagnostic_testbusiness.industry010401 analytical chemistry020206 networking & telecommunicationscomputer.file_formatJPEGcolor reproduction accuracy0104 chemical sciencesdermatologymedicine.anatomical_structure:NATURAL SCIENCES::Physics::Atomic and molecular physics [Research Subject Categories]Lab color spaceRGB color modelHuman eyetelemedicineArtificial intelligencecolor constancybusinessSkin lesioncomputerdescription
The recent spread of cheap dermatoscopes for smartphones can empower patients to acquire images of skin lesions on their own and send them to dermatologists. Since images are acquired by different smartphone cameras under unique illumination conditions, the variability in colors is expected. Therefore, the mobile dermatoscopic systems should be calibrated in order to ensure the color constancy in skin images. In this study, we have tested a dermatoscope DermLite DL1 basic, attached to Samsung Galaxy S4 smartphone. Under the controlled conditions, jpeg images of standard color patches were acquired and a model between an unknown device-dependent RGB and a device independent Lab color space has been built. Results showed that median and the best color error was 7.77 and 3.94, respectively. Results are in the range of a human eye detection capability (color error ≈ 4) and video and printing industry standards (color error is expected to be between 5 and 6). It can be concluded that a calibrated smartphone dermatoscope can provide sufficient color constancy and can serve as an interesting opportunity to bring dermatologists closer to the patients.
| year | journal | country | edition | language |
|---|---|---|---|---|
| 2017-12-07 | Biophotonics—Riga 2017 |