Classification of Skin Disease using Image Processing

Project Title

Classification of Skin Disease using Image Processing

Project Area of Specialization Artificial IntelligenceProject Summary

Introduction

The aim of our project is to help dermatologist to detect and classify skin diseases like Melanoma, Nevus and Seborrheic keratosis from dermascopic images. Disease like melanoma, which spread throughout the body, if it is detected earlier, it can be cure 97% but if it not detected in mean time the probability of its cure is drop to 14%. Moreover, the detection of these types of disease take more than one month in testing out project will provide immediate result at good accuracy.

Project Objectives

Objectives

Efficient extraction and classification of Pigmented Skin lesion such as that of Melanoma, Melanocytic
nevus and Seborrheic keratosis.

Project Implementation Method

Process

• Image Acquisition

The doctor will first upload the dermascopic image into the system.

• Pre-Processing

For Preprocessing, we have to set the contrast, sharpness, white balance, saturation and brightness of the image. Moreover we also want to remove the hair particles and noise from the image.

• Segmentation

We then segment the lesion from background and extract the lesion from image.

• Feature Extraction

The next step is to extract the features from the image like, Assymmtery, Area, Color, Diameter and texture of the lesion.

• Classification

In last we will then classify the result based from the features detect using Neural Networks.

Benefits of the Project

Benefits

• Many patients will be saved from Biopsy saving cost, time and pain cause by procedure.
• Helping Dermatologist to decide to operate the disease or use medication after seeing the result.
• Help dermatologist whether the skin lesion is malignant or benign.

Technical Details of Final Deliverable

• We will use Image Processing to detect and extract the lesion. For that we have used Otsu Thresholding for automatic and custom ROI + active contour for manual.
• Will then extract features like Area, Diameter, GLCM for texture of ROI, color features and use Artificial Neural Networks for the classification of disease.

Final Deliverable of the Project Software SystemType of Industry Medical Technologies Artificial Intelligence(AI)Sustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	21500
Asus PH-GTX1050-2G Phoenix GeForce® GTX 1050 2GB GDDR5 Video Graphics	Equipment	1	21500	21500