Background Patients with neurofibromatosis type 1 (NF1) develop benign (BPNST), pre-malignant atypical (ANF), and malignant (MPNST) peripheral nerve sheath tumors. Radiological differentiation of these entities is challenging. Therefore, we aimed to evaluate the value of an MRI-based radiomics machine learning classifier for differentiation of these three entities of internal peripheral nerve sheath tumors in NF1 patients. Methods MRI was performed at 3T in 36 NF1 patients (20 male; age: 31±11 years). Segmentation of 117 BPNSTs, 17 MPNSTs, and 8 ANFs was manually performed using T2w SPAIR sequences. One hundred seven features per lesion were extracted using PyRadiomics and applied for BPNST vs. MPNST differentiation. A 5-feature radiomics signature was defined based on the most important features and tested for signature-based BPNST vs. MPNST classification (random forest (RF) classification, leave-one-patient-out evaluation). In a second step, signature feature expressions for BPNSTs, ANFs and MPNSTs were evaluated for radiomics-based classification for these three entities. Results The mean AUC for the radiomics-based BPNST vs. MPNST differentiation was 0.94, corresponding to correct classification of on average 16/17 MPNSTs and 114/117 BPNSTs (sensitivity: 94%, specificity: 97%). Exploratory analysis with the 8 ANFs revealed intermediate radiomic feature characteristics in-between BPNST and MPNST tumor feature expression. Conclusion In this proof-of-principle study, machine learning using MRI-based radiomics characteristics allows sensitive and specific differentiation of benign and malignant peripheral nerve sheath tumors in NF1 patients. Feature expression of pre-malignant atypical tumors were distributed in-between benign and malignant tumor feature expressions, which illustrates biological plausibility of the considered radiomics characteristics.

Objectives To compare the accuracy and precision of 3D-Dixon and 2D-SSFP MR-imaging for assessment of aortic diameter in Marfan patients. Methods This prospective single-center study investigated respiratory-gated 3D-Dixon and breath-hold 2D-SSFP non-contrast MR-imaging at 3 T in 47 Marfan patients (36.0 ± 13.2 years, 28♀,19♂). Two radiologists performed individual diameter measurements at five levels of the thoracic aorta and evaluated image quality on a four-grade scale (1 = poor, 4 = excellent) and artifacts (1 = severe, 4 = none). Aortic root diameters acquired by echocardiography served as a reference standard. Intraclass correlation coefficient, Bland-Altman analyses, F-test, t-test, and regression analyses were used to assess agreement between observers and methods. Results Greatest aortic diameters were observed at the level of the sinuses of Valsalva (SOV) for 3D-Dixon (38.2 ± 6.8 mm) and 2D-SSFP (38.3 ± 7.1 mm) (p = 0.53). Intra- and interobserver correlation of diameter measurements was excellent at all aortic levels for both 3D-Dixon (r = 0.94–0.99 and r = 0.94–0.98) and 2D-SSFP (r = 0.96–1.00 and r = 0.95–0.99). 3D-Dixon-derived and 2D-SSFP-derived diameter measurements at the level of the SOV revealed a strong correlation with echocardiographic measurements (r = 0.92, p < 0.001 and r = 0.93, p < 0.001, respectively). The estimated mean image quality at the level of SOV was higher for 2D-SSFP compared to that for 3D-Dixon (3.3 (95%-CI: 3.1–3.5) vs. 2.9 (95%-CI: 2.7–3.1)) (p < 0.001). Imaging artifacts were less at all aortic levels for 3D-Dixon compared to 2D-SSFP (3.4–3.8 vs. 2.8–3.1) (all p < 0.002). Conclusion Respiratory-gated 3D-Dixon and breath-hold 2D-SSFP MR-imaging provide accurate and precise aortic diameter measurements. We recommend 3D-Dixon imaging for monitoring of aortic diameter in Marfan patients due to fewer imaging artifacts and the possibility of orthogonal multiplanar reformations of the aortic root. Key Points • Respiratory-gated 3D-Dixon and breath-hold 2D-SSFP imaging provide accurate and precise aortic diameter measurements in patients suffering from Marfan syndrome. • Imaging artifacts are stronger in 2D-SFFP imaging than in 3D-Dixon imaging. • We recommend 3D-Dixon imaging for monitoring of aortic diameter in Marfan patients due to fewer imaging artifacts and the possibility of orthogonal multiplanar reformations.

Objectives To define and compare growth rates of the distal aorta in Marfan patients with and without aortic root replacement using serial MR angiography (MRA). Methods We retrospectively included 136 Marfan patients with a total of 645 MRAs who underwent a median of five MRAs (range: 2–13) at 1.5 T and 3 T in annual intervals. Of these, 41 patients (34.8 ± 12 years) had undergone aortic root replacement. The remaining 95 patients (29.0 ± 17 years) still had a native aorta and served as the control group. Thoracic aortic diameters were independently measured at eleven predefined levels. Estimated growth rates were calculated using a mixed effects model adjusted for sex, age, BMI, and medication. Results Marfan patients with aortic root replacement revealed the highest mean estimated growth rate in the proximal descending aorta (0.77 mm/year, CI: 0.31–1.21). Mean growth rates at all levels of the distal thoracic aorta were significantly higher in patients with aortic root replacement (0.28–0.77 mm/year) when compared to patients without aortic root replacement (0.03–0.07 mm/year) (all p < 0.001). Antihypertensive medication, gender, and BMI had no significant impact on the distal aortic growth rates. Conclusion Distal thoracic aortic diameters increase at a significantly higher rate in Marfan patients with aortic root replacement compared to Marfan patients without aortic root replacement. Further studies are warranted to investigate if the increased growth rate of the distal thoracic aorta after aortic root replacement is caused by altered hemodynamics due to the rigid aortic root graft or due to the general genetic disposition of post-operative Marfan patients. Clinical relevance statement High growth rates of the distal aorta after aortic root replacement underline the need for careful life-long aortic imaging of Marfan patients after aortic root replacement. Key Points • Aortic growth rates in Marfan patients with aortic root replacement are highest in the mid-aortic arch, the proximal- and mid-descending aorta. • Growth rates of the distal thoracic aorta are significantly higher in Marfan patients with aortic root replacement compared to Marfan patients without aortic root replacement. • Antihypertensive medication, gender, and BMI have no significant impact on distal aortic growth rates in Marfan patients.