Mathematical Sciences Research Seminar Series

As part of the School of Science and Technology Mathematical Sciences Research Seminar Series, Dr Lama Hamadeh, NTU presents: Power Spectrum and Machine Learning Analysis for Dried Blood Droplets Statistical and Quantitative Discrimination.

Abstract

One of the most interesting and everyday natural phenomenon is the formation of different patterns after the evaporation of colloidal droplets deposited on a solid surface. Analysing dried patterns resulted from biological liquids, such as blood, has recently gained a lot of attention, experimentally and theoretically, due to their wide applications in biomedicine, forensic science and many others. This work analyses a large database of dried blood droplet images with different five experimental conditions objectively and quantitatively by developing sophisticated image processing analysis routines and optimising a multivariate statistical machine learning algorithm. An analysis of various measured physiological parameters is also investigated.  We have seen that when our machine learning algorithm, which is an optimisation to a statistical model that combines Principal Component Analysis (PCA) as an unsupervised learning method and Linear Discriminant Analysis (LDA) as a supervised learning method, is applied on the logarithmic power spectrum of the images, it can provide a very good prediction accuracy, reaches to 95%, to discriminate the blood conditions, provided we undertake an average over all the images taken per volunteer per condition.

All Welcome

For any enquiries please contact Dr David Chappell.