News related to Science, Technology, Environment, Agriculture and Medicine in India

Automatic identification of melanoma from dermoscopic images


Not all melanoma cases are this evident. Image via Wikimedia Commons

In recent years, skin melanoma appears to be on the increase. The rate of survival is high if skin melanoma is detected in an early stage. The lesions that appear on the skin in melanoma are of irregular shape with fuzzy borders. Variations in skin color, hair and reflections from skin while taking dermoscopic images make analysing the dermoscopic images difficult. Techniques to automatically recognise the lesions that have been developed so far either give inaccurate results or are computationally costly. 

Recently, M N Giri Prasad and T Sreelatha from JNTUA, Ananthapuram collaborated with M.V.Subramanyam, Santhiram Engineering College, Nandyal to develop a precise segmentation method for diagnosing dermoscopic images and to distinguish lesions due to melanoma of skin.  

Melanoma lesions in dermoscopic images have edges that are diffuse – There is a colour gradient from the inside to the outside. This feature need to be taken into account when the contour of the lesion is identified. If that is done, then the regions of interest within the dermoscopic images can easily be distinguished 

The team developed a gradient and feature adaptive contour model for efficient segmentation. The model analyses the image based on the Gaussian statistical technique, by measuring an equal number of pixels above and below the mean value of the gradient. This helps create an outline of melanoma lesions.  

The confirmation of the technique was done using  the dermoscopic images from PH2, a database created by a hospital in Portugal. The database contains both carcinogenic and non-carcinogenic dermoscopic images identified by experts in the field and is, therefore, a good resource for testing the accuracy of algorithms for melanoma identification. The researchers say that the gradient and feature adaptive contour model achieved 98% accuracy and 99% specificity in the tests.

The researchers compared the efficiency  and accuracy of their segmentation technique with existing methods and found that it is better than existing methods.

The automatic identification method will help doctors who do not have adequate experience in interpreting dermoscopic images Diagnose skin melanoma easily and initiate treatments early enough to control cancerous growth.

Medical Systems, 43 (7): 190 (2019);
DOI: 10.1007/s10916-019-1334-1

CFTRI, Mysore

STEAMindiaReports: providing relevant science news tips for media houses

Categorised in: Andhra Pradesh, Diagnostics, Medicine

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Follow Us

%d bloggers like this: