Optimal Deep Neural Network model based multimodality fused medical image classification

• Published in: Physical communication Vol. 41; p. 101119
• Main Authors: Subbiah Parvathy, Velmurugan, Pothiraj, Sivakumar, Sampson, Jenyfal
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.08.2020
• Subjects: Classification; Deep learning; Deep Neural Network; Image fusion; Medical image; Optimization; Transform; Deep learning; Deep Neural Network; Classification; Medical image; Transform; Image fusion; Optimization
• ISSN: 1874-4907; 1876-3219
• DOI: 10.1016/j.phycom.2020.101119

Medical image fusion and classification has been utilized to get valuable significant data multimodality medical image data. The fundamental commitment of this paper is to extricate the compelling features from the fused medical image that is Multimodality Medical images (MMI) for the classification model. At first, MMI are considered for the fusion procedure by shearlet Transform and optimization model to get fused images. The medical images of fusion has demonstrated to be valuable for propelling the clinical dependability of utilizing medical imaging for medical diagnostics and investigation, this procedure is considered as the principal phase. In send phase, few features are extracted from the Fused multimodality Medical Image (FMMI). For example, texture and XOR pattern models for classification, the point of recognition as given fused image as begin or malignant by Deep Neural Network (DNN) with Optimization. The significant purpose of optimization is to improve the classification accuracy. To optimize the DNN structure, we proposed the Discrete Gravitational Search Algorithm (DGSA), and it depends on increasing speed and force of DNN weights. When the optimal structure finds, the test image comprises the input of the structure for classification. The experimental results stated that the proposed model shows effective performance with the fusion factor of 6.52 and spatial frequency of 26.8.