The Sydney Morning Herald is reporting that an “Australian-developed instrument designed to automatically diagnose melanomas on patients in the doctor’s surgery is more accurate than general practitioners, a study suggests.” The study published at the Archives of Dermatology looked at the SolarScan, a device by Sydney-based Polartechnics Ltd.
Some details on the workings of the device, from the company itself:
SolarScan® captures and analyses calibrated images of pigmented skin lesions that are stored for subsequent lesion monitoring or confirmation.
The SolarScan® camera has an integrated light source and employs a three CCD video camera for enhanced colour resolution. Each image is calibrated to a common reference standard to ensure image reproducibility…
The SolarScan® camera quickly “photographs” and digitises the lesion image for processing. Characteristic features are extracted and compared to an empirical database based on more than 1,800 benign and malignant lesions collected by the Sydney Melanoma Unit, and other clinical sites in Australia, the United States, and Germany. The results of the lesion analysis are displayed as four index graphs on the screen.
1. Skin lesion – image capture subsystem
Oil or gel is applied to the lesion before imaging to eliminate surface reflections and to obtain a standard dermatoscopic image. Simply pressing the trigger on the camera handpiece captures the image.
2. Image processing
Each image undergoes real-time calibration to normalise all images to the same reference, and to correct for possible changes in illumination. The camera and light source have been specified, designed, and constructed to exacting standards to ensure stable calibration.
3. Feature extraction
Image preprocessing masks out hairs and air bubbles, and identifies the boundaries of the lesion.
4. Comparison against known feature database
SolarScan® software extracts critical features of each lesion using over 100 discriminants. These parameters include aspects of colour segmentation, lesion patterns and geometry.
5. Lesion evaluation
Lesion characteristics are compared to statistical data from the clinical lesion database.
6. Display of results
Information is clearly presented on the display screen to assist in clinical evaluation.
It seems to us that the device still has Class 1 FDA listing which permits sales of this instrument in the US for lesion monitoring and diagnostic assistance.
The product page…
(hat tip: Doc Around The Clock)
