Detection of Myocardial Ischemic Events from Echocardiogram using Linear Discriminant Analysis and Multilayer Perceptron
Sujata Joshi1, Mydhili K. Nair2
1Sujata Joshi*, Department of Computer Science and Engineering, Nitte Meenakshi Institute of Technology, Bangalore, India.
2Mydhili K. Nair, Department of Information Science and Engineering, M.S.Ramaiah Institute of Technology, Bangalore, India.
Manuscript received on November 14, 2019. | Revised Manuscript received on 25 November, 2019. | Manuscript published on December 10, 2019. | PP: 2875-2880 | Volume-9 Issue-2, December 2019. | Retrieval Number: B7891129219/2019©BEIESP | DOI: 10.35940/ijitee.B7891.129219
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In the recent past, Myocardial Ischemia has emerged as a major cause of death worldwide, and hence we should look at its timely diagnosis and care. The cause of Ischemia is due to clogging of blood supplying arteries leading to the heart by a substance called plaque. This leads to reduced supply of oxygen to heart without which the heart muscles begin to die leading to a myocardial infarction or heart attack. Early detection of ischemia is of critical importance because, in most of the cases, the effects of myocardial ischemia are reversible if detected early enough. The objective of this paper is to detect Myocardial Ischemic events from Echocardiography data. The echocardiography data reports used in this work is obtained from M.S. Ramaiah Narayana Heart Centre from June 2018 to June 2019. It has 344 patient data and 42 features extracted for each patient. The two events identified to be detected are Left ventricular hypertrophy(LVH) and Left ventricular diastolic dysfunction(LVDD). Firstly the data is preprocessed for removal of noise and outliers. Then we apply feature reduction techniques to retain only those features which are meaningful to the purpose of prediction of events identified. The predictive models are then developed on the reduced feature set. In this research, we have proposed the algorithms LVH-PRED-MODEL and LVDD-PRED_MODEL for detection of LVH and LVDD ischemic events. The proposed algorithm uses Linear Discriminant Analysis for feature reduction and Multilayer perceptron for training the models. The results show excellent accuracy of 88.37% for LVH-PRED-MODEL and 86.04% for LVDD-PRED-MODEL using the proposed approach. The Area Under Curve is 83.5% and 73.37% for LVH-PRED-MODEL and LVDD-PRED-MODEL events respectively. These models can be applied in real time to detect myocardial ischemic events by healthcare workers.
Keywords: Myocardial Ischemia, Data Mining, Linear Discriminant Analysis, Multilayer Perceptron, Left ventricular hypertrophy, Left Ventricular diastolic dysfunction, Heart disease
Scope of the Article: Data Mining