Abstract

Skin diseases are frequent and quite perennial in the world, and in some cases, these lead to cancer. These are curable if detected earlier and treated appropriately. An automated image-based detection system consisting of four main modules: image enhancement, region of interest segmentation, feature extraction, and detection can facilitate early identification of these diseases. Diverse image-based methods incorporating machine learning techniques are developed to diagnose different types of skin diseases. This article focuses on the review of the tools and techniques used in the diagnosis of 28 common skin diseases. Furthermore, it has discussed the available image databases and the evaluation metrics for the performance analysis of various diagnosis systems. This is vital for figuring out the implementation framework as well as the efficacy of the diagnosis methods for the neophyte. Based on the performance accuracy, the state-of-the-art method for the diagnosis of a particular disease is figured out. It also highlights challenges and shows future research directions.

Highlights

  • The skin is the largest organ of the human body

  • The fifth method uses thresholding-based segmentation followed by an ABCD feature extraction rule and the extracted features are fed into the KNN to classify the skin disease (Kumar, 2016)

  • The sixth method uses a region growing segmentation followed by the gray level co-occurrence matrix (GLCM) feature extraction algorithm, and a hybrid classification algorithm (SVM and KNN) is used to classify the melanoma (Sumithra, 2015)

Read more

Summary

INTRODUCTION

The skin is the largest organ of the human body. For an adult, the skin surface measures approximately 16,000 cm and represents about 8% of the body weight. There are several types of skin diseases with their particular characteristics, such as dermatitis eksfoliatif generalisata, impetigo, pityriasis rosea, erisipelas, nekrolisis toksika epidermal, eczema, psoriasis, acne, warts, vitiligo, tinea corporis, scabies, hives, rosacea, and shingles, boiling, cell, cold sores, corns, calluses, dyshidrotic, malluscum contagiosum, neurofibromatosis, skin tags, melanoma, rash, malignant melanoma - squamous cell carcinoma (SCC), basal cell carcinoma (BCC), genetic diseases - genetic skin disorders, sickle cell disease anemia, leprosy, viral infection, seborrhoeic dermatitis, lichen planus, pink pityriasis, chronic dermatitis, pityriasisrubrapilaris, herpes, seborrheic kurtosis (SK), nevus, bullae, splitz, venous malformations, and scleroderma, etc. Automatic skin disease detection systems use skin images or dermoscopy images. Detection of skin cancer helps to take remedial measures for the complete elimination of the disease from the body; otherwise, the skin will be severely affected by cancer and will not be curable.

THEORETICAL BACKGROUND
System Performance Metric
CLASSIFICATION OF SKIN DISEASE DETECTION SYSTEM
Impetigo
Pityriasis rosea
Erisipelas and nekrolisis toksika
Plaque psoriasis and chronic eczema
Nail psoriasis and warts
Psoriasis
10. Vitiligo and Tinea corporis
15. Splitz nevus and Venous malformations
19. Melanoma
20. Skin cancer
SUMMARY OF THE STATE-OF-THE-ART SKIN DISEASE DETECTION SYSTEM
CHALLENGES AND RECENT THREATS IN SKIN DISEASE DETECTION
CONCLUSION AND FUTURE RESEARCH
Findings
Limitations
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call