Abstract |
Soft tissue tumors comprise a broad category of neoplasms with variable degree of
malignancy. Accurate and early tissue characterization yields a vital role in patient
management and disease prognosis. MRI is the imaging modality of choice for preoperative
assessment of soft tissue tumors as it offers supreme soft tissue contrast, multi plane coverage
and does not involve exposure to ionizing radiation. More importantly, multi modal MRI
imaging offers insight into tissue pathology from a number of different contrast mechanisms
each one highlighting a different aspect of tumor microenvironment. However, histopathologic
correlation of MR findings is a necessary step for the validation of MR findings and definite
tissue characterization.
Contrast on MR images can indirectly characterise these properties by adjusting image
contrast to be dependent on a sought-after property of tissue. Specifically, contrast on DW
images is related to cell density and vascularity as derived from water mobility in tissue. DCE
dynamic protocol highlights areas of increased vascular permeability through dynamic imaging
during contrast medium administration. Since biomarkers related to vascularity can be derived
by both DCE and DWI methods based on different theoretical assumptions, a study of
agreement attracts great interest. As DWI is also indicative of tissue cellularity, which along
with vascularity/permeability, is a very powerful metric of tumor aggressiveness. In paper III a
correlation study is presented between the two different methodologies (DWI-DCE) in terms of
statistical correlation and spatial agreement for increased tumor malignancy and conclude to a
visual guide of areas within the tumor with MR findings indicative of increased malignancy.
Paper IV is a study of the different DCE enhancement patterns that are indicative of viable
necrotic and hypoxic tumor sites.
Another robust quantitave MRI methodology dating from the early days of MR imaging is T2
and T2* relaxometry as it provides tissue specific metrics of magnetic properies, independent
of acquisition parameters and thus indicative of tissue properties. For this part of the study
benign lipoma patients were also enrolled and additionally phantoms were used for reference
measurements. This study focused selectively on tumors of adipocytic origin as liposarcomas
were the majority of soft tissue sarcomas in the patient cohort used in our study. Paper II and
V inferred tissue identity and composition as manifested indirectly in multi echo T2
relaxometry measurements. Paper II introduces Spin Coupling ratio (SCratio) metric indicative of signal loss related to the
spin coupling phenomenon which is a known phenomenon for healthy adipose tissue (bright
fat phenomenon) but has not been studied for other tissues of lipomatous origin, such as
lipomas or liposarcomas. This marker has the potential to be used for identification of areas of
increased / decreased tissue differentiation within a heterogeneous neoplasm and can be a
helpful tool for pre-operative tissue characterization for biopsy guiding. The study was
supported by preliminary phantom results published in paper I.
Paper V introduces a proposed methodology for multi exponential T2 relaxometry (Mexp) and
validates the results also in comparison with the well-established ILT method on a phantom as
a preliminary stage for the application of the proposed methodology to adipocytic tumors
(paper VI). The proposed technique has the added advantage over the gold standard ILT
method of producing voxel based parametric maps rather than ROI based T2 distributions,
which is essential taking into account tissue heterogeneity. Lipomatous tumors with different
degree of malignancy exhibit distinct behavior patterns as measured with Mexp, and thus the
proposed method can be used along with conventional imaging methods for preoperative
radiological assessment.
An advanced oncologic protocol hosting all abovementioned imaging techniques was deployed
in this study and resulting biomarkers were validated with histopathologic assessment in order
to constitute a set of robust and clinically relevant biomarkers for the characterization of soft
tissue neoplasms. Histopathologic analysis results were used for final tissue classification,
necessary for the analysis of MR findings. The results presented in this thesis are useful for
supporting radiological diagnosis and can also be a useful tool for optimizing imaging-driven
biopsies.
|