Researchers from Missouri University of Science and Technology have received a patent for “Automatic Detection of Critical Dermoscopy Features for Malignant Melanoma Diagnosis.” The system relies on taking close-up photos of suspected lesions using a specialty built device. These are then processed on a computer that grades the presence of various image characteristics and comes up with a total score that suggests potential melanoma.
Moss and his fellow researchers use photographs taken with a device called a dermatoscope. The scope contains a magnifier, a light source and a transparent plate, which is placed on skin coated with mineral oil, alcohol or a clear gel. The instrument illuminates the skin at a low angle in all directions, which makes the skin’s upper layers more transparent, thus making deeper pigment patterns and structures visible.
S&T’s program extracts a number of features (22 in one study) from the lesion image and uses them to assist in the diagnosis of malignant melanoma.
“In one study, our method achieved a diagnostic accuracy rate of more than 94 percent on a set of 724 lesion images,” says Moss. “Without the use of technology, only 82-87 percent of lesions are correctly diagnosed.”
Link: Skin cancer detection made quicker, easier by S&T researchers…
Image credit: B.A. Rupert, Missouri S&T