Fig. 1
From: Deep learning for Dixon MRI-based attenuation correction in PET/MRI of head and neck cancer patients
The eleventh patient of the cohort, a 52-year-old male with right base of the tongue cancer and lymph node involvement (T2N1M0). Each row from top to bottom shows an axial, coronal and sagittal slice of: the reference CT; the vendor-provided atlas-based attenuation map (Atlas); the deep learning derived attenuation map (Deep); PETCT; PETAtlas; PETDeep; the relative difference map between PETCT and PETAtlas (ΔPETAtlas); the relative difference map between PETCT and PETDeep (ΔPETDeep);. The involved lymph node is delineated in green for the axial images. Notice, that the atlas-based attenuation map does not classify the trachea as air and the overall reduced PET error for the deep learning method, which is apparent from the difference maps