Human Visual System (HVS) Inspired Automatic Brain Tumor Segmentation

Artificial Intelligence plays an important role in the field of Biomedical Technology for various applications including diagnosis, imaging, surgery, etc. Artificial intelligence helps in reducing the error in diagnosis that may happen due to human judgment. The brain-related diseases require greater precision and even small uncertainty can result in fatal loss. Traditionally, medical experts use medical imaging techniques such as CT Scans, X-Rays MRI (Magnetic Resonance Imaging), etc. to diagnose the brain tumor. However, such a process is slow and requires a large amount of manpower. In this work, we propose an automated brain tumor detection workflow using the advancements in artificial intelligence to accurately segment the brain tumor. The proposed automated brain tumor detection systems can detect the smallest aberrance in the brain with minimum error. We have used a Convolutional Neural Network (CNN) to detect brain tumor from MRI images. CNN is first trained using the data acquired from local diagnostic centers. Trained CNN is capable of detecting tumors with good accuracy. In order to further improve the results, we propose here to use contrast limiting adaptive histogram equalization (CLAHE) with CNN. The CNN based tumor segmented images were precise to 78%on other hand the CNN with CLAHE based segmented images provided as accurate results as hand marked results by radiologist. With the average precision of 91.9% and average Fmeasure of 90.5% the CNN based Btumor framework. The proposed framework overcomes the limitation of existing methods that are not efficient with low-quality MRI images. We have used publically available datasets along with MRI data collection from the local diagnostic center. The experimental results show that the proposed framework outperforms other existing methods in terms of accuracy. The initial feedback received from expert radiologist is also very encouraging