Subject

Breast microcalcifications (MCs) are high-contrast small calcium deposits that are typically found in the breast tissue. They are considered to be an important biomarker for the early detection and diagnosis of breast cancer. Detecting and characterizing their properties is essential in assessing the likelihood of the malignancy at the very initial phase of the disease. ETRO has developed machine/deep learning based Computer Aided Detection and Diagnosis (CAD) systems by exploiting MCs’ individual properties. The results obtained have clearly demonstrated the link between breast MCs and malignancy. This work is based on high resolution 3D images, acquired by a microCT scanner. This technique is not available in vivo, hence we scanned biopsies containing MCs.

Thus far, all the experiments have considered only the morphological and texture features of MCs. However, it is well-reported that benign and malignant MCs have different chemical compositions. MCs are divided into the following main types: (a) type I composed of calcium oxalate (CaC2O4), related mostly with benign lesions (b) type II composed of calcium phosphate, mainly hydroxyapatite (Ca5(PO4)3(OH)), most often related with malignant lesions (c) the existence of a new sub-type is reported composed of magnesium-substituted hydroxyapatite ((CaMg)5(PO4)3(OH)), only found in malignant lesions.

A few studies, performed in phantoms with embedded MCs, have reported that dual-energy X-Ray imaging can differentiate between benign and malignant MCs based only on their material/chemical composition. In collaboration with Ghent University, this proposed master thesis aims to conduct a proof of concept study using dual energy MicroCT scanning (a) to explore and (b) to analyze the possibilities of discriminating the chemical compositions of MCs, particularly in simulated environments. The experimental work will start with phantom studies, involving the scanning of MCs structures composed of the pure chemical compositions mentioned above. The student will characterize the attenuation properties of the different materials and explore the influence of the different scanning parameters on the separability of the different attenuation curves.


[1] Niki Martini et al. ‘Dual energy x-ray methods for the characterization, quantification and imag- ing of calcification minerals and masses in breast’. In: Crystals 10.3 (2020), p. 198.

[2] Bahaa Ghammraoui et al. ‘Classification of breast microcalcifications using dual-energy mammography’. In: Journal of Medical Imaging 6.1 (2019), pp. 013502–013502.

[3] Kim Hyuk, Lee Minjong, and Kim Hyuk Jin. ‘Dual energy-based quantification method for determination of breast microcalcification types’. In: Journal of Instrumentation 17.11 (2022), p. C11009.

[4]?Tracy E. Kirkbride et al. ‘Discrimination between calcium hydroxyapatite and calcium oxalate us- ing multienergy spectral photon-counting CT’. In: AJR Am J Roentgenol 209.5 (2017), pp. 1088– 92.

[5] Redona Brahimetaj et al. ‘Improved automated early detection of breast cancer based on high resolution 3D micro-CT microcalcification images’. In: BMC cancer 22.1 (2022), pp. 1–13.

[6] Muller, Florence M., et al. "Performance evaluation of a micro-CT system for laboratory animal imaging with iterative reconstruction capabilities."?Medical Physics?49.5 (2022): 3121-3133.

[7] https://www.molecubes.com/systems/

Kind of work

Literature review about: (a) dual-energy scanning principle (b) the chemical composition of breast MCs (c) dual-energy studies employed for breast MCs material characterization.

Find/prepare (solid) phantom MC structures composed of the above-mentioned chemical compositions. Obtain attenuation curves for each of the materials.

Explore and analyze the influence of scanning parameters on the discrimination capabilities of each material characterization. Propose an optimal acquisition setup.

Evaluate and validate the proposed proof of concept protocol on real paraffin blocks containing MCs structures. Analyze in depth the obtained results.

Thesis writing and presentation.

Number of Students

1

Expected Student Profile

Interest in biomedical image processing and acquisition

Very good programming and analytic skills

Experience with image acquisition principles and processing

Ability to write scientific report and communicate research results.

As the thesis is guided by staff from both universities, the student should be prepared to work at both universities when necessary.