282 chapter 3 were all significantly lower in follicular adenoma than in follicular thyroid carcinoma. ROC curve analysis revealed that the 25th percentile resulted in the highest AUC of 87%, with an optimal cutoff value of 0.353 × 10-3 mm²/s. A lower ADC value in follicular adenoma compared to follicular carcinoma seems contradictory with results of Nakahira et al., where lower ADCs were found in malignant nodules. The probable reason for this is that Nakahira et al. predominantly included PTC with histological characteristics of calcifications and desmoplastic reactions, which cause restriction of free water movement in the cellular environment and reduce ADC values, whereas follicular neoplasms including Hürthle cell nodules are known for their varying colloid tissue involvement and thus histologically contain more fluid. Thus, DW-MRI would inaccurately provide a more benign image [328, 422]. DWI has been more extensively investigated in the differentiation between benign and malignant unselected thyroid nodules. A 2014 meta-analysis of Wu et al. summarised seven studies with 358 subjects and presented a pooled sensitivity of 91%; a specificity of 93%, a LR+ of 12.24; a LR- of 0.99; a diagnostic OR of 123.78; and an AUC of the summary ROC of 94% [423]. In 2016, a metaanalysis by Chen et al. summarised 15 studies with 765 lesions and presented a pooled sensitivity of 90%; a specificity of 95%; a LR+ of 16.49; a LR- of 0.11; a diagnostic OR of 150.73; and an AUC of the summary ROC of 95% [329]. Most studies showed a significantly lower mean ADC value in malignant lesions compared to benign lesions, because of larger nuclei, denser stroma and higher cell counts, all of which led to increased cellularity and reduced extracellular space. However, no absolute cutoff was found. This could be attributed to heterogeneous methodology such as varying b-values and differences in ADC measurements. Other explanations could be a diversity in patient population or components with high diffusivity in malignant lesions, like cystic components, central necrosis, or intratumoural haemorrhage. DWI seems a promising non-invasive, non-radiative and accurate technique for the pre-operative differentiation of (cytologically indeterminate) thyroid nodules. Nevertheless, while the worldwide availability of MRI scanners is growing, MRI is still considered an expensive technique in terms of hardware, overhead costs, and the relatively long scan duration. Large-scale trials are necessary to assess and validate its clinical value, to establish harmonisation in methodology, to determine cut-off values, and to study cost-effectiveness, specifically in FNAC indeterminate thyroid nodules. Magnetic resonance spectroscopy Magnetic resonance spectroscopy (MRS) is an analytical method used for the in vivo chemical characterisation of tissue, measuring the presence and concentration of various metabolites. Magnetic resonance principles are used to detect various nuclei, such as hydrogen-1 (1H), which all can provide valuable metabolic and physiological information [424]. 1H-MRS is able to capture the metabolic profile of a lesion, by determination of the relative concentrations and physical properties of a variety of biochemicals. These include several low molecular weight metabolites such as choline, creatine, glutamate, lactate, and different amino acids. Spectroscopy uses the chemical
RkJQdWJsaXNoZXIy MTk4NDMw