International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 07 | July 2024
p-ISSN: 2395-0072
www.irjet.net
Design and Implementation of a Web-Based Deep Learning System for Melanoma Detection Utilizing the Inception Model Md. Ashikur Rahman1, Ishtiak Ahmed2, Md. Abdullah Al Humayun3 1 Department of Petroleum and Mining Engineering, Chittagong University of Engineering and Technology,
Chattogram, Bangladesh.
2Mechanical and Production Engineering, Ahsanullah University of Science and Technology, Dhaka, Bangladesh. 3Deptartment of Electrical and Electronic Engineering, Eastern University, Dhaka, Bangladesh
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract
-
occurrence vary geographically; in Western countries, melanomas typically arise in sun-exposed areas such as the chest, forehead, and limbs. Conversely, in Asia, melanomas more commonly develop on the hands and feet, which are less exposed to the sun, as well as on mucous membranes, including the lining of the mouth, throat, gastrointestinal tract, and the vaginal tract in women [4].
Melanoma represents a particularly aggressive form of skin cancer, necessitating prompt detection to mitigate mortality rates and minimize the invasiveness of treatment. Contemporary advances in computer-aided diagnosis (CAD) have leveraged sophisticated imaging techniques to enhance early-stage diagnosis of skin malignancies. This research aims to construct a web-based platform for the automated analysis of dermoscopic images to ascertain early melanoma presence. The implementation employs the Inception V3 model, a Convolutional Neural Network (CNN) architecture renowned for its efficacy in image classification tasks, specifically in medical imaging domains such as dermoscopy. This model facilitates the processes of image acquisition, preprocessing, segmentation, feature extraction, and classification. The web application, developed using Python, HTML, and CSS, showcases a streamlined interface for clinical application. Empirical evaluations reveal that the model achieves an accuracy range of 90-93%, underscoring its potential utility in clinical settings. This platform empowers users to swiftly identify skin abnormalities, thereby enhancing early diagnosis and preventive care, significantly contributing to advancements in dermatological oncology.
Excessive exposure to ultraviolet (UV) radiation from sunlight or artificial tanning is a major risk factor for the development of melanoma. Reducing UV exposure can significantly lower the risk of melanoma, achievable through measures such as consistent use of sunscreen, wearing long-sleeved clothing, and using hats to block the sun. Additionally, genetic mutations in cancer-causing genes, particularly BRAF and NRAS, CDKN2A, play a crucial role in melanoma development [2][4]. In the past decade, malignant melanoma has become one of the most dangerous cancers, spreading rapidly worldwide. Each year, more than one million cases of nonmelanoma skin cancer and over 250,000 cases of melanoma are reported [2]. In the United States, melanoma was ranked fifth for expected new cancer cases in both males and females in 2019 Error! Reference source not found.. The incidence of melanoma in the U.S. has exhibited a concerning upward trend, significantly contributing to cancer-related mortality over the past decades. To address this, a web-based deep learning system has been developed for melanoma detection, employing the Inception model to achieve faster operation and higher accuracy. In response, a sophisticated webbased deep learning system using the Inception model has been developed, offering rapid and highly accurate melanoma detection with innovative approach aims to enhance early diagnosis.
Key Words: Convolutional Neural Network (CNN), Computer-aided diagnosis (CAD), Dermoscopic images, Empirical evaluations, Image classification, Inception V3 model, Melanoma
1. INTRODUCTION Melanoma is recognized as the deadliest form of skin cancer, with significant global impact. It has been reported recently that new melanoma cases reported in 2020 was 325,000 may increase up to 510,000 by 2040. Likewise, deaths due to melanoma may rise by approximately 68%, from 57,000 in 2020 to 96,000 in 2040[1].
1.1 Background
Early detection of melanoma can lead to a complete cure through simple excision of the malignant tissue. However, late-stage detection often results in metastasis, leading to a poor prognosis [2]. Melanoma originates in melanocytes, the skin cells responsible for producing pigments that determine skin color. The patterns of melanoma
© 2024, IRJET
|
Impact Factor value: 8.226
Before Skin cancer, particularly melanoma, represents a pressing global health concern. A new study by scientists from the International Agency for Research on Cancer (IARC) and partners predicts that the number of new cases of cutaneous melanoma per year will increase by more
|
ISO 9001:2008 Certified Journal
|
Page 1212