Document Type



Doctor of Philosophy (PhD)


Biomedical Engineering

First Advisor's Name

Dr. Wei-Chiang Lin

First Advisor's Committee Title

Committee Chair

Second Advisor's Name

Dr. Jorge Riera

Second Advisor's Committee Title

Committee Member

Third Advisor's Name

Dr. Ranu Jung

Third Advisor's Committee Title

Committee Member

Fourth Advisor's Name

Dr. Timothy Allen

Fourth Advisor's Committee Title

Committee Member


Brain, Cortex, Motor, Somatosensory, Cortical Column, Neuron, Astrocyte, Vasculature, Nuclei, Fractional Volume Analysis, Deconvolution, Nuclei Classification

Date of Defense



Characterization of the complex cortical structure of the brain at a cellular level is a fundamental goal of neuroscience which can provide a better understanding of both normal function as well as disease state progression. Many challenges exist however when carrying out this form of analysis. Immunofluorescent staining is a key technique for revealing 3-dimensional structure, but subsequent fluorescence microscopy is limited by the quantity of simultaneous targets that can be labeled and intrinsic lateral and isotropic axial point-spread function (PSF) blurring during the imaging process in a spectral and depth-dependent manner. Even after successful staining, imaging and optical deconvolution, the sheer density of filamentous processes in the neuropil significantly complicates analysis due to the difficulty of separating individual cells in a highly interconnected network of tightly woven cellular arbors. In order to solve these problems, a variety of methodologies were developed and validated for improved analysis of cortical anatomy. An enhanced immunofluorescent staining and imaging protocol was utilized to precisely locate specific functional regions within brain slices at high magnification and collect four-channel, complete cortical columns. A powerful deconvolution routine was established which collected depth variant PSFs using an optical phantom for image restoration. Fractional volume analysis (FVA) was used to provide preliminary data of the proportions of each stained component in order to statistically characterize the variability within and between the functional regions in a depth-dependent and depth-independent manner. Finally, using machine learning techniques, a supervised learning model was developed that could automatically classify neuronal and astrocytic nuclei within the large cortical column datasets based on perinuclear fluorescence. These annotated nuclei were then used as seed points within their corresponding fluorescent channel for cell individualization in a highly interconnected network. For astrocytes, this technique provides the first method for characterization of complex morphology in an automated fashion over large areas without laborious dye filling or manual tracing.




Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.



Rights Statement

Rights Statement

In Copyright. URI:
This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).