Home > Journals > Journal of Radiological Review > Past Issues > Il Giornale Italiano di Radiologia Medica 2019 Gennaio-Febbraio;6(1) > Il Giornale Italiano di Radiologia Medica 2019 Gennaio-Febbraio;6(1):25-30



To subscribe
Recommend to your librarian


Publication history
Cite this article as



Il Giornale Italiano di Radiologia Medica 2019 Gennaio-Febbraio;6(1):25-30

DOI: 10.23736/S2283-8376.19.00164-5


language: English, Italian

Feasibility of imaging phenotyping (radiomics) on nuclear MRI of major lower limb nerves

Alberto S. TAGLIAFICO 1, 2, Federica ROSSI 1, 2 , Francesca VALDORA 2, Marina GRANDIS 1, 3, Luana BENEDETTI 1, Bianca BIGNOTTI 1, 2, Angelo SCHENONE 1, 3, Carlo MARTINOLI 1, 2

1 San Martino Polyclinic Hospital and Institute for Cancer Research and Care, Genoa, Italy; 2 Department of Health Sciences (DISSAL), University of Genoa, Genoa, Italy; 3 Department of Neuroscience, Rehabilitation, Ophthalmology, Genetics, Maternal and Child Health (DINOGMI), University of Genoa, Genoa, Italy


BACKGROUND: To demonstrate the feasibility of imaging phenotyping (radiomics) on magnetic resonance imaging (MRI) of major lower limb nerves.
METHODS: The work was done according to the Declaration of Helsinki and its most recent updates. We retrospectively reviewed clinical 1.5-T MRI data between January 2017 and January 2018 of 10 adult patients (5 men and 5 women; mean age 52.4±6.6 years) with lower limb peripheral neuropathy and 10 age- and sex-matched controls (5 men and 5 women; mean age 58.9±6.7). The femoral, sciatic, interdigital and tibial nerves were evaluated with a radiomic approach on morphological fast spin echo axial T1-weighted images. Nerve contour was manually traced by two radiologists in consensus after training. A total of 104 radiomic features were evaluated with software. Radiomic features of normal and pathological lower limb nerves were compared. Receiver operating characteristic (ROC) analysis was performed. Reading time to identify and analyze the nerves was calculated.
RESULTS: The pathological group included: dysimmune neuropathies (N.=4), entrapments (N.=4), Morton neuroma (N.=2). Statistically significant differences (P<0.05) were found for seven selected features: skewness, energy, entropy, kurtosis, 90th percentile, dissimilarity, homogeneity. Three of seven features showed the highest accuracy with an area under the curve (AUC) of 0.88 (95% CI: 0.72-1.00), 0.92 (95% CI: 0.82-1.00) and 0.91 (95% CI: 0.80-1.00). Mean time to study major lower limb nerves with Radiomics was 2 hours ±15 minutes.
CONCLUSIONS: Seven first level features were found to differentiate lower limb pathological nerves from normal nerves. Three of seven features showed a satisfied accuracy.

KEY WORDS: Lower extremity - Peripheral nerves - Magnetic resonance imaging - Phenotype

top of page