Performance analysis of melanoma classifier using electrical modeling technique
- PDF / 2,788,145 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 11 Downloads / 239 Views
ORIGINAL ARTICLE
Performance analysis of melanoma classifier using electrical modeling technique Tanusree Roy 1
&
Pranabesh Bhattacharjee 1
Received: 19 January 2020 / Accepted: 27 July 2020 # International Federation for Medical and Biological Engineering 2020
Abstract An efficient and novel modeling approach is proposed in this paper for identifying proteins or genes involved in melanoma skin cancer. Two types of classifiers are modeled, based on the chemical structure and hydropathy property of amino acids. These classifiers are further implemented using NI LabVIEW–based hardware kit to observe the real-time response for proper diagnosis. The phase responses, pole-zero diagrams, and transient responses are examined to screen out the genes related to melanoma from healthy genes. The performance of the proposed classifier is measured using various performance measurement metrics in terms of accuracy, sensitivity, specificity, etc. The classifier is experimented along with a color code scheme on skin genes and illustrates the superiority in comparison with traditional methods by achieving 94% of classification accuracy with 96% of sensitivity. Keywords Electrical modeling . Gene . Real-time classifier . Simulation . Skin cancer
1 Introduction Over the past few years, various kinds of network model– based approaches are used for modeling complex biological networks, i.e., DNA structure, RNA structure, and protein networks. [1–5, 63, 64]. For example, the protein structure is modeled as an electrical RLC (M) circuit model for analysis, visualization, shape synthesis, and pattern matching [1]. E. coli bacteria are represented using the digital and Markov method for cloning the λ Switch [2]. The DNA/RNA strings are modeled using resistor-capacitor (RC) ladder networks [3]. A Gene–Protein–miRNA electronic oscillator is developed by an electrical modeling technique [4]. An equivalent electrical network is also developed for the DNA molecule as well [5]. Similarly, various research works are in progress nowadays to develop new technologies for the detection of early-stage cancer [6–14], which motivates to design a simple, cost-effective, reliable system model. As a result, some significant
* Tanusree Roy [email protected] 1
Department of Electrical and Electronics Engineering, University of Engineering and Management, Kolkata 700135, India
research works take place like amino acid–based sensor models are developed for cancer detection [9, 10, 62, 65, 66]. In this paper, we address the issue related to the prediction of cancerous genes of the human skin, as melanoma is one of the most common human malignancies and its occurrence is higher than other cancers [15]. Due to repeated exposure to UV radiation, cancer risk in human skin is enhanced [16]. Pollutants, chemicals, and occupational exposures are also responsible for skin cancers [17]. Basal cell carcinoma of the skin is the most common type of human cancer, where mutations in some specific genes increase the risk of cancer growth [18]. Detecting this n
Data Loading...
