Cardiovascular MRI image analysis by using the bio inspired (sand piper optimized) fully deep convolutional network (Bio-FDCN) architecture for an automated detection of cardiac disorders

•An automated detection method based on SPO-FDCN for detecting the cardiovascular disease.•CVD-MRI is used to extract features from different data sets using LV and LVEDV and LVESV, LVM, RVEDV.•The main aim of this work is to increase the complexity of the calculation with the help of time cost and SPO.•Bio-FCN architecture, and deep learning convolution filters are used to estimate the CVD.•The experimental results demonstrate that efficiency of the proposed system is more efficient compared with existing methods. This manuscript proposes an automated detection method depends on sand piper optimized fully deep convolutional neural network (FDCN-SPO) for detecting the cardiovascular disease. In this detection process, the cardiovascular magnetic resonance imaging (CVD-MRI) is used, to extract that shape features from Kaggle cardio-vascular disease dataset using left ventricular (LV) volume, left ventricle end-diastolic volume (LVEDV), left ventricle end-systolic volume (LVESV), left ventricle mass (LVM), right ventricular (RV), right ventricular with end-diastolic volume (RVEDV) and right ventricular with end-systolic volume (RVESV). Here, the fully deep convolution neural network (Bio-FCN), and deep learning convolution filters are used to estimate the cardio vascular disease (CVD). The major intention of this proposed operation “to decrease the complexity of the calculation, cost function and increase the accuracy of fully deep convolution neural network, which was optimized using the proposed Sand Piper Optimization (SPO) algorithm”. The proposed FDCN-SPO algorithm shows the optimal accuracy and computational performance for mass of myocardial, thickness of wall, left and right ventricular volume, and ejection fraction (EF). The proposed fully deep convolutional neural network optimized with Sand Piper Optimization (FDCN-SPO) method shows 97.63% accuracy, 96.50% sensitivity, 98.01% specificity, 95.39% F-measure, 94.29% precision and 93.80% MCC value. The experimental outcomes illustrate the proposed FDCN-SPO scheme is more efficient than existing process like Random Forest Classifier with principle component analysis (RFC-PCA), Neural Network with partial least squares regression (NN-PLSR), recurrent neural network using principle component analysis and partial least squares regression (RNN-PCA & PLSR), Extreme Learning Model with Normal Sinus Rhythm (ELM-NSR).