Diagnosis of Brain Lesions, Glioma, Multiple-Sclerosis and Metastases from MRI: An efficient classifier-aided method using Refractive Index as a surrogate Biological Marker
Keywords:Medical Imaging, Biomedical Engineering, Machine Learning, Brain Lesions, Glioma, Spectroscopy,
We introduce a highly accurate method of diagnosing the various pathological conditions that might exist in a subject’s brain, like edema, multiple sclerosis (Tumefactive, Relapsing-remitting, secondary-progressive, primary-progressive and progressive-relapsing), glioma, glioblastoma and metastases. These show up on conventional MRI scans, but it is often difficult to identify the exact type of the pathology from the grayscale image. We employ the use of Support Vector Machines (SVM) to work on the MR Spectroscopy [6, 12] data and correctly identify the condition-especially in seemingly vague cases where radiologists cannot rule out high uncertainty in their conclusion. The SVM trains on data sets collected for different patients and optimizes its hyperplanes based on eight input variables – T2, CHO, ADC, CR, CHO/NAA, CR/NAA, LIP/LAC, MI, CH/CR, T2 periphery  and Refractive index. Refractive index is an additional parameter which we include to get better boundary lines and accuracy, as shown in our prior works . We test this SVM on a set of 19 patients’ data and achieve 100% accuracy in predictions. The training and testing is carried out in MATLAB.
(1) Schwab, K.E., Gailloud P, Wyse G, Tamargo R.J., Limitations of magnetic resonance imaging and magnetic resonance angiography in the diagnosis of intracranial aneurysms. Neurosurgery, Volume 63, Issue 1, 1 July 2008
(2) Support Vector Machines - http://www.statsoft.com/Textbook/Support-Vector-Machines
(3) Mingxia Liu, Daoqiang Zhang, Joint Binary Classifier Learning for ECOC-Based Multi-Class Classification. IEEE Transactions on Pattern Analysis and Machine Intelligence ( Volume: 38, Issue: 11, Nov. 1 2016 )
(4) Pardalos P.M. (2008) Hyperplane Arrangements in Optimization. In: Floudas C., Pardalos P. (eds) Encyclopedia of Optimization. Springer, Boston, MA
(5) Tapan K Biswas, R Bandyopadhyay, A Dutta, Validating The Discriminating Efficacy Of MR T2 Relaxation Value Of Different Brain Lesions And Comparison With Other Differentiating Factors: Use Of Artificial Neural Network And Principal Component Analysis. The Internet Journal of Radiology (Volume 20, No. 1)
(6) Radiopedia Introduction to MR Spectroscopy, https://radiopaedia.org/articles/mr-spectroscopy-1.
(7) Mathworks Guide, Support Vector Machines for Binary Classification.
(8) Machine Learning, Wikipedia, https://en.wikipedia.org/wiki/Machine_learning.
(9) Nor M., Noor Rahman, J. Adnan, Intracranial Bleed Post Stereotactic
Biopsy: Lessons Learned, The Internet Journal of Neurosurgery (Volume 8, Number 1)
(10) Tapan K Biswas, S R Choudhury, A Ganguly, R Bandyopadhyay, A Dutta, Refractive Index As Surrogate Biological Marker Of Tumefactive And Other Form Of Multiple Sclerosis And Its Superiority Over Other Methods. The Internet Journal of Radiology (Volume 19, Number 1).
(11) Blake A. Johnson, Avoiding diagnostic pitfalls in neuroimaging. Applied Radiology-The Journal of Practical Medical Imaging and Management 2016;45(3):24-29. March 02, 2016
(12) Wikipedia, Nuclear Magnetic Resonance Spectroscopy, https://en.wikipedia.org/wiki/Nuclear_magnetic_resonance_spectroscopy
(13) Kasai, M., Yasuda, Y., Mizoguchi, H., Soga, K., Kaneko, K., & Takemura, H. (2017). In vivo tumor wavelength band selection using Hierarchical clustering and PCA with NIR-Hyperspectral Data. Journal of Biomedical Engineering and Medical Imaging, 4(1), 01.
(14) Limam, O. (2016). MRI Segmentation based on Multiobjective Fuzzy Clustering. Journal of Biomedical Engineering and Medical Imaging, 3(2), 07.
(15) Abd El kader, I., Zhang, S., & Xu, G. (2017). Improved Fuzzy C-Means Algorithm for Brain Tumor Identification Analysis Using Magnetic Resonance Brain Images. Journal of Biomedical Engineering and Medical Imaging, 4(3), 15.