We aimed to evaluate radiomic features’ stability across different region of interest (ROI) sizes in CT and MR images. We chose a phantom with a homogenous internal structure so no differences for a feature extracted from ROIs of different sizes would be expected. For this, we scanned a plastic cup filled with sodium chloride solution ten times in CT and per MR sequence (T1-weighted-gradient-echo and T2-weighted-turbo-inversion-recovery-magnitude). We placed sphere-shaped ROIs of different diameters (4, 8, and 16 mm, and 4, 8, and 16 pixels) into the phantom’s center. Features were extracted using PyRadiomics. We assessed feature stability across ROI sizes with overall concordance correlation coefficients (OCCCs). Differences were tested for significance with the Mann–Whitney U-test. Of 93 features, 87 T1w-derived, 87 TIRM-derived, and 70 CT-derived features were significantly different between ROI sizes. Among MR-derived features, OCCCs showed excellent (>0.90) agreement for mean, median, and root mean squared for ROI sizes between 4 and 16 mm and pixels. We further observed excellent agreement for 10th and 90th percentile in T1w and 10th percentile in T2w TIRM images. There was no excellent agreement among the OCCCs of CT-derived features. In summary, many features indicated significant differences and only few showed excellent agreement across varying ROI sizes, although we examined a homogenous phantom. Since we considered a small phantom in an experimental setting, further studies to investigate this size effect would be necessary for a generalization. Nevertheless, we believe knowledge about this effect is crucial in interpreting radiomics studies, as features that supposedly discriminate disease entities may only indicate a systematic difference in ROI size.

No abstract

Purpose To assess the value of magnetic resonance imaging (MRI) in detecting craniofacial fibrous dysplasia (CFD) and diagnosing and differentiating it from intraosseous meningioma. Additionally, the MRI appearance of the typical computed tomography (CT) imaging feature, the ground glass phenomenon, was evaluated. Material and methods MRI datasets of 32 patients with CFD were analysed retrospectively. Detectability in MRI was assessed by analysis of 10 randomly selected patients with CFD and 10 normal controls by two blinded readers. Changes of affected bone, internal lesion structure, T1 and T2 signal intensity, and contrast enhancement of the lesion in general and ground glass areas in particular were assessed. Ten patients with intraosseous meningioma (one in each) served as differential diagnosis for CFD. Results All 10 CFD lesions were reliably detected in MRI. In 32 patients 36 CFD lesions were evaluated. In 66.7% CFD were iso- to hypointense in T1 and hyperintense in T2; this proportion was similar for ground glass areas (65.7%). Ground glass areas were more homogeneously structured than the whole CFD lesion in both T1 (100% vs. 56%, respectively) and T2 (91% vs. 61%, respectively). Contrast enhancement was found in 97% of complete CFD lesions and 93% of ground glass areas. The accuracy for CFD vs. intraosseous meningioma was 100% for ‘no soft-tissue component’ and 98% for ‘bone broadening’ in MRI. Conclusions Distinct morphological changes of CFD are reliably detected in MRI and allow differentiation from intraosseous meningioma. Areas with ground glass phenomenon in CT show a predominantly homogenous internal structure in MRI with contrast enhancement.

Aim was to develop a user-friendly method for creating parametric maps that would provide a comprehensible visualization and allow immediate quantification of radiomics features. For this, a self-explanatory graphical user interface was designed, and for the proof of concept, maps were created for CT and MR images and features were compared to those from conventional extractions. Especially first-order features were concordant between maps and conventional extractions, some even across all examples. Potential clinical applications were tested on CT and MR images for the differentiation of pulmonary lesions. In these sample applications, maps of Skewness enhanced the differentiation of non-malignant lesions and non-small lung carcinoma manifestations on CT images and maps of Variance enhanced the differentiation of pulmonary lymphoma manifestations and fungal infiltrates on MR images. This new and simple method for creating parametric maps makes radiomics features visually perceivable, allows direct feature quantification by placing a region of interest, can improve the assessment of radiological images and, furthermore, can increase the use of radiomics in clinical routine.