Methods and algorithms of segmentation and deconvolution for quantitative analysis of Tissue Microarray images

Abstract : This thesis aims at developing dedicated methods for quantitative analysis of Tissue Microarray (TMA) images acquired by fluorescence scanners. We addressed these issues in biomedical image processing, including segmentation of objects of interest (i.e. tissue samples), correction of acquisition artifacts during the scanning process and improvement of acquired image resolution while taking into account imaging modality and scanner design. The developed algorithms allow envisaging a novel automated platform for TMA analysis, which is highly required in cancer research nowadays. On a TMA slide, multiple tissue samples which are collected from different donors are assembled according to a grid structure to facilitate their identification. In order to establish the link between each sample and its corresponding clinical data, we are not only interested in the localization of these samples but also in the computation of their array (row and column) coordinates according to the design grid because the latter is often very deformed during the manufacturing of TMA slides. However, instead of directly computing array coordinates as the existing approaches, we proposed to reformulate this problem as the approximation of the deformation of the theoretical TMA grid using ``thin plate splines'' given the result of tissue sample localization. We combined a wavelet-based detection and an ellipse-based segmentation to eliminate false alarms and thus improving the localization result of tissue samples. According to the scanner design, images are acquired pixel by pixel along each line, with a change of scan direction between two subsequent lines. Such scanning system often suffers from pixel mis-positioning (jitter) due to imperfect synchronization of mechanical and electronic components. To correct these scanning artifacts, we proposed a variational method based on the estimation of pixel displacements on subsequent lines. This method, inspired from optical flow methods, consists in estimating a dense displacement field by minimizing an energy function composed of a nonconvex data fidelity term and a convex regularization term. We used the half-quadratic splitting technique to decouple the original problem into two small sub-problems: one is convex and can be solved by a standard optimization algorithm, the other is non-convex but can be solved by a complete search. To improve the resolution of acquired fluorescence images, we introduced a method of image deconvolution by considering a family of convex regularizers. The considered regularizers are generalized from the concept of Sparse Variation which combines the $L_1$ norm and Total Variation (TV) to favors the co-localization of high-intensity pixels and high-magnitude gradient. The experiments showed that the proposed regularization approach produces competitive deconvolution results on fluorescence images, compared to those obtained with other approaches such as TV or the Schatten norm of Hessian matrix.
Document type :
Theses
Complete list of metadatas

Cited literature [137 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-01737764
Contributor : Hoai Nam Nguyen <>
Submitted on : Friday, March 23, 2018 - 11:32:32 AM
Last modification on : Wednesday, April 11, 2018 - 1:53:50 AM
Long-term archiving on : Thursday, September 13, 2018 - 10:26:46 AM

File

NHN_Manuscrit_these_final.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-01737764, version 1

Collections

Citation

Hoai Nam Nguyen. Methods and algorithms of segmentation and deconvolution for quantitative analysis of Tissue Microarray images. Image Processing [eess.IV]. Université Rennes 1, 2017. English. ⟨tel-01737764⟩

Share

Metrics

Record views

245

Files downloads

191