Indian Journal of Science and Technology
DOI: 10.17485/ijst/2018/v11i1/120361
Year: 2018, Volume: 11, Issue: 1, Pages: 1-10
Original Article
Rupal Snehkunj1*, Ashish N. Jani2 and Nalin N. Jani3
1Department of Computer Science, Shree Ramkrishna Institute of Computer Education and Applied Sciences, Surat – 395001, Gujarat, India; [email protected]
2Department of Computer Engineering, P. P. Savani University, Kosamba – 394125, Gujarat, India; [email protected]
3Department of Computer Science, Kadi University, Gandhinagar – 382023, Gujarat, India; [email protected]
*Author for correspondence
Rupal Snehkunj,
Department of Computer Science, Shree Ramkrishna Institute of Computer Education and Applied Sciences, Surat – 395001, Gujarat, India; [email protected]
Objectives: The research initiative in this paper focused on brain image feature extraction and its organized storage filtered of from abnormalities located on the brain Magnetic Resonance (MR) and Computer Tomography (CT) scan images which are preprocessed.
Methods/Statistical Analysis: For this study, abnormalities such as brain tumor & brain hemorrhage are taken into consideration as they share many common characteristics which can be diagnose using same implementation methodology. The MRI & CT brain images were studied so as to explore various phases such as brain image extraction, brain image transformation and brain image progression on it.
Findings: This work integrates the phases in a computer based system which facilitates the use of the processes in an integrated, distinctive and sequenced manner with ease and comfort in its uses. The brain image extraction and brain image transformation phase inculcates merging of patient’s MRI or CT Dicom image slices into single image, noise reduction by three different methods, noise selection based on Peak Signal to Noise Ratio (PSNR) and Mean Squared Error (MSE) error metrics, skull removal and lastly image enhancement. The outcome of this stage is inputted to brain image progression phase where image is characterize into T1-weighted, T2-weighted and PD-weighted for segmentation where uncommon areas are fragmented using T1-w, T2-w and PD-w brightness and intensity values. Finally based on segmented results, the features are extracted and selected for empowering classification capacity and detection accuracy.
Application: Experiments are conducted on more than 200 brain MRI/CT image datasets and promising results were reported.
Keywords: Brain Abnormalities, Brain Image Progression, Feature Extraction, Feature Selection
Subscribe now for latest articles and news.