• 43007 Tarragona, Spain

Title of Project : Multi-model imaging analysis: Biomarkers correlation between radiological and histopathological imaging
Student Name: Tewele W.Tareke​      

Summary:

Tewele Weletnsea Tareke, hailing from Ethiopia, holds a Bachelor's degree in Computer Science and Engineering from Mekelle University - MIT, where his B.Sc thesis focused on "Steganography-based secure data encryption and transmission." With a diverse work background, he has served as a Lecturer and Programmer at Debre Tabour University and as a Lecturer and Researcher at Mekelle University, both in Ethiopia. Tewele pursued his Master's Degree in the Erasmus Mundus Joint Master of Medical Imaging and Application, with a master's thesis titled "Automatic Classification of Nodules from 2D Ultrasound Images Using Deep Learning Networks." His skill set includes expertise in AI modeling, advanced image processing, machine/deep learning modeling, and programming in C++/Python. Currently, Tewele is undertaking a Ph.D. as a student in the BosomShield Horizon EU Project.

Objectives:
Studying the correlation between WSIs and radiological images, such as PET and/or MRI could improve the radiological imaging screening. Thus, this project will perform the following tasks: (1) to analyse the intra-tumour heterogeneity in WSIs through a novel digital image analysis (DIA) techniques of multi-labelling in bright or in fluorescence field, (2) to extend this analysis to other tumour locations and various immunohistochemical markers, as well to further understand the tumour biology, and link these information to multiparametric MRI, 2D/3D mammograms and PET image analysis, (3) to perform a comprehensive analysis of tumour heterogeneity at the level of autoradiography and immunohistochemistry and to compare the obtained quantitative information with multiparametric radiological MRI and PET imaging, (4) to conduct the developed approaches on various tumour situations at the preclinical level first (high grade glioma, brain and lung metastases from BC). Once validated, this correlation will then be performed at the clinical level (glioma for instance), (5) the correlation between changes extracted from the MRI images and the genetic information collected from medical centres will be done, and (6) stereology estimates of 3D structure and genetic information will be also compared to 3D imaging (MRI).

Outcomes:
The main result of this project is to improve the PET and MRI image analysis and avoid an invasive procedure for the easier cases of BC relapse. The outcomes of this project are: 1) a study of tumour heterogeneity to understand the epithelium development, and the micro-environment organisation. 2) an approach to address tumour heterogeneity based on bimodality analysis to predict aggressive tumour and patient relapse. 3) DIA for tissue pathology data retrieved in digital format and available for computational processing enabling high accuracy, reproducibility and capacity to enumerate various biological processes. 4) matching different image modalities in order to correlate the measurements performed in the same plan after a fine registration (either rigid or affine registration). 5) association between changes extracted from the MRI/PET images and the genetic information collected from medical centres.