Fig. 1

Workflow. After donor inclusion, post-mortem in-situ 3DT1 and DTI were collected [34], from which cortical thickness was derived with Freesurfer [41] and free-water corrected cortical mean diffusivity (FWC-MD) was derived with FSL [44], DIPY [47, 48] and Freesurfer [41] (purple box). After the MRI scan, autopsy was performed, and 9 cortical regions were selected from the right hemisphere (RH) for further analysis. To match these regions to the MR images, regions of interest were selected from the Lausanne atlas [42, 43] (yellow box). Brain tissue was processed for immunohistochemistry against phosphorylated Ser129 α-synuclein (pSer129-αSyn, clone EP1536Y), phosphorylated-tau (p-tau, clone AT8), amyloid β (Aβ, clone 4G8), neurofilament light chain (NfL, amino acid sequence 1-376) and phosphorylated neurofilaments medium and heavy chains (p-NfM/H, clone SMI312), which were quantified using QuPath [50]. The correlations between neurofilament immunoreactivity and pathology load, and between neurofilament immunoreactivity and MRI outcome measures were investigated via linear mixed models (yellow and purple dashed arrows, respectively). Aβ: amyloid beta; FWC-MD: free water corrected mean diffusivity; IHC: immunohistochemistry; LB: Lewy Body; NfL: neurofilament light chain; pSer129-αSyn: phosphorylated Ser129 alpha synuclein; p-NfM/H: phosphorylated neurofilament medium and heavy chain; p-tau: phosphorylated tau; RH: right hemisphere