Analysis of Brain MRI Images Using Improved CornerNet Approach

• DOI: 10.3390/diagnostics11101856
• Publication Date: 2021-10-10T03:37:08Z
• Authors: Marriam Nawaz; Tahira Nazir; Momina Masood; Awais Mehmood; Rabbia Mahum; Muhammad Attique ...; Seifedine Kadry; Orawit Thinnukool
• Keywords: Brain tumor; Brain disease

The brain tumor is a deadly disease that caused by the abnormal growth of cells, which affects human blood cells and nerves. Timely precise detection tumors an important task to avoid complex painful treatment procedures, as it can assist doctors in surgical planning. Manual time-consuming activity highly dependent on availability area experts. Therefore, need hour design accurate automated systems for classification various types tumors. However, exact localization categorization challenging job due extensive variations their size, position, structure. To deal with challenges, we have presented novel approach, namely, DenseNet-41-based CornerNet framework. proposed solution comprises three steps. Initially, develop annotations locate region interest. In second step, custom DenseNet-41 base network introduced extract deep features from suspected samples. last one-stage detector employed classify several evaluate method, utilized two databases, Figshare Brain MRI datasets, attained average accuracy 98.8% 98.5%, respectively. Both qualitative quantitative analysis show our approach more proficient consistent detecting classifying than other latest techniques.