International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 06 | Jun 2025
p-ISSN: 2395-0072
www.irjet.net
Intelligent Algorithms for Detection and Classification Tasks of Dermatological Malignancies Prof Aziz Makandar1, Mrs. Ayisha Soudagar2* 1Professor and 2*Research Scholar 1&2 Department of Computer Science Karnataka State Akkamahadevi Women’s University Vijayapura, India
---------------------------------------------------------------------***--------------------------------------------------------------------cancer mortality, making it one of the most deadly types of Abstract - The rising rate of skin cancer cases necessitates the disease [2]. The annual incidence of newly diagnosed malignant melanoma cases increased by 31% throughout the preceding ten years (2012–2022) [3].
sophisticated early detection methodologies. Modern machine learning techniques, especially for image analysis, have emerged as potent tools in this endeavour. These models are designed to process substantial datasets, including high-resolution images of coetaneous lesions, to identify subtle patterns potentially indicative of malignancy. By harnessing machine learning, medical experts have the potential to enhance their ability to detect skin cancer earlier, facilitating timely interventions and potentially improving patient outcomes. This research focuses on applying the support vector machine (SVM) technique, implemented via MATLAB, for skin cancer classification. SVMs are well-suited for this task due to their capacity to manage high-dimensional data and delineate complex decision boundaries. The SVM algorithm's efficacy is evaluated using key performance metrics, including accuracy, sensitivity, and specificity. These metrics assess the algorithm's ability to accurately classify cutaneous lesions, minimize false negatives, and reduce false positives. Through the development of such machine learning approaches, researchers aim to create more reliable and efficient tools for skin cancer detection. This endeavour has offers the capability to redefine how early diagnosis and treatment are approached in skin cancer
The estimated five-year chances of survival in those suffering from early-detected melanoma is around 99 percent, although early detection is essential for the possibility of successful therapy. When the disease spreads to distant organs, the survival rate drops to 30%, and when it affects the lymph nodes, it drops to 68% [3]. Given these numbers, it is critical to accelerate the diagnosis of melanoma and skin-related malignancies. In conclusion, early diagnosis is vital for effective treatment and better consequences of skin malignancy. In an effort to preserve lives and lessen the financial and medical burdens on patients, automated strategies for reliably diagnosing cancer must be developed because specialists are not always available. Melanoma has a very varied look, and skin cancers can be hard to differentiate from normal benign skin abnormality. The morbidity and mortality linked to skin cancer may be decreased by using artificial intelligence to help detect the disease early [6]. In addition to reducing effort, AI-based solutions may improve the Identification of skin abnormalities [7, 8]. One of the main drivers of the fourth industrial revolution is artificial intelligence (AI), a branch of computer science that uses computers and software to mimic intelligent human behaviour using a range of technologies [9]. A subset of artificial intelligence known as machine learning, or ML, repeatedly learns from data using statistical models and algorithms. This enables computers to perform certain jobs and forecast the properties of upcoming samples. As a result, the complex algorithms are made to do procedures that would likely otherwise be difficult for humans to understand. Convolution neural networks (CNNs), a kind of machine learning that replicates the way biological neurons behave, are the primary architecture for pattern detection in medical image analysis [10]. Millions of people worldwide are affected by skin cancer, a prevalent and sometimes deadly illness. Improving prognosis and lowering death rates depend on early recognition and accurate differentiation of skin lesions. The majority of Traditional approaches to skin cancer diagnosis rely on the visual inspection of dermatologists, which can be labour-intensive and subjective [11]. Because machine learning (ML) techniques automate the process and
Key Words: accuracy, sensitivity, classification of skin cancer, specificity, support vector machine, skin lesions.
1. INTRODUCTION One major worldwide health concern is cancer. According to global predictions, cancer will be responsible for about 10.0 million deaths in 2020 (9.9 million excluding nonmelanoma skin cancer). Lung cancer, prostate cancer, and breast cancer in women are the most often diagnosed cancers. The major causes of death associated with cancer are stomach, liver, and lung malignancies [1]. Skin cancer is common in Caucasians, including non-melanoma skin cancer (NMSC) and malignant melanoma. [2] The incidence is rising. The US Skin Cancer Foundation claims that, Skin cancer affects more Americans, each year than by any other type of disease. In recent decades, Skin malignancies have grown more prevalent common, posing serious threats to public health [1]. Melanoma accounts for 80% of skin
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 800