A deep learning assisted image-guided framework for differentiation among tumors and hemorrhages in head imaging. (October 2022)

• Record Type: Journal Article
• Authors: Malik, Payal; Vidyarthi, Ankit
• Is Part Of: Computers & electrical engineering. Volume 103(2022)
• Publication Date: 2022-10
• Subjects: Brain imaging -- Deep learning -- Image-guided intervention -- Medical image classification -- Multimodal data engineering; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2022.108330
• Languages: English
• ISSNs: 0045-7906

Abstract

Abstract: Image-guided intervention refers to the analysis of multimodal data using Computer-Aided Diagnostic (CAD) systems. The inclusion of the Deep Learning algorithms in the CAD systems normally boosts the performance of the medical dataset, especially in binary classification problems. However, the performance of the deep frameworks in the area having overlapping characteristics like Brain Tumors and Brain Hemorrhages is still an open research area. This paper introduces a new deep framework named HemNet to distinguish the brain images having Tumors or Hemorrhage. The proposed framework uses the multimodal data included with the cascading of the Gated Recurrent Unit (GRU) to classify the generated image sequences. The experimentation results of the proposed framework help to distinguish the overlapping slices more accurately which were also clinically validated. The gain in the accuracy is 96.76% while the average gain in the accuracy is 96.48% when evaluated on the different epoch sizes. Graphical abstract: Highlights: Proposed a deep learning framework for multimodal data analysis. Proposed a backbone model called HemNet for differentiating Brain Tumors and Hemorrhages. Utilized Gated Recurrence Unit (GRU) to analyze the multi-modal head CT image sequences. Experimentation resulted as accuracy is 96.76% while the average accuracy is 96.48%.