Multimodal biomedical image retrieval and indexing system using handcrafted with deep convolution neural network features

Abstract The advances in biomedical imaging equipment have produced a massive amount of medical images that are generated by the different modalities. Consequently, huge volume data has been and caused complex time-consuming retrieving process relevant cases. To resolve this issue, Content-Based Biomedical Image Retrieval (CBMIR) system is applied to retrieve related from earlier patients’ databases. However, previous handcrafted features methods CBMIR model shown poor performance many multimodal In paper, we focus on designing technique using Deep Learning (DL) models. We present new Multimodal Classification (M-BMIRC) for classifying proposed M-BMIRC involves three dissimilar processes as following: feature extraction, similarity measurement, classification. It uses an ensemble Zernike Moments (ZM) deep Convolutional Neural Networks (DCNN) extraction process. Additionally, Hausdorff Distance based measure employed identify resemblance between queried image exist database. Moreover, classification gets executed retrieval Probabilistic Network (PNN) model, which allocates class labels tested images. Finally, experimental studies conducted two benchmark datasets results ensure superior terms measures include Average Precision Rate (APR), Recall (ARR), F-score, accuracy, Computation Time (CT).