A framework for image-based, automated, multilevel analysis of the cytoskeletal morphology

Aim: The aim of this thesis is to create an integrated bioinformatic framework for the quantitative assessment of cytoskeleton (CSK) morphology. In this research, we demonstrated the validity of this approach by applying computational biology in two cases: Cytospace images and neurite images. I used these 2 dataset of images as reference dataset to build the computational biology framework and to verify its functionality. To accomplish this, I have analyzed Cytospace optical and confocal microscope images by Image J and MATLAB programming. I have analyzed the neurite number and length of PC 12 cells by using the following software and tools: (a) custom made tool MATLAB, (b) Cell profiler, (c) Image J, and (d) Filopodyan. Outside this goal of CSK analysis I have also analysed the microcalcification and parenchyma images through my framework to verify the capability of the framework to operate on a different scenario. Methods: We have developed various algorithms and protocols for the analysis of nuclei, tubulin and microtubules. Each component of the cytoskeleton plays a very essential role in understanding the behaviour of the cell. Moreover, the confocal images were processed properly to extract the information regarding the cytoskelton. Finally, we also studied PC 12 cells and selected / designed useful algorithms to analyze changes in the number and length of neurites. Results: In the first part of this thesis we presented an in-silico model constructed by using measurable parameters obtained from microscope images of cells. We use image analysis software Image J to identify cell shape parameters – including surface area, roundness, fractal dimension, such as entropy and coherency. In the second part of thesis, we also analysed area, perimeter, major and minor axis, circularity, the solidity of the nuclei and tubulin by MATLAB programming. In the third part of the thesis, we anlayzed the dynamics of neurites by developing MATLAB scripts, using also other analysis tools such as Image J, Neuron J. Conclusion: I have build a computational framework to analyse in quantitative manner images from optical and confocal microscopy. During the first two years of PhD course I select, develop and integrated bioinformatics protocols and algorithms to define optimized operational pipelines. I used two datasets of image as reference to tuning the computational pipeline. In the last period of phd course I can apply my optimised framework on microscopy images for detection and analysis of neurites. By means of the application of the framework was possible reduce considerable time of biologist to analyse the images.