Search

IntroductionThe MDS-PSP criteria have shown high sensitivity for the PSP diagnosis, but do not discriminate the phenotype diversity. Our purpose was to search for anatomopathological differences among PSP phenotypes resulting from the application of the MDS-PSP criteria comparing with the previous ones.MethodsThirty-four PSP cases from a single brain bank were retrospectively classified according to the criteria used by Respondek et al. in 2014 and the PSP-MDS criteria at 3 years (MDS-3y), 6 years (MDS-6y) and at the last clinical evaluation before death (MDS-last). Semiquantitative measurement of total, cortical and subcortical tau load was compared. For comparative analysis, PSP-Richardson syndrome and PSP postural instability were grouped (PSP-RS/PI) as well as the PSP atypical cortical phenotypes (PSP-Cx).ResultsApplying the Respondek's criteria, PSP phenotypes were distributed as follow: 55.9% PSP-RS/PI, 26.5% PSP-Cx, 11.8% PSP-Parkinsonism (PSP-P), and 5.9% PSP-Cerebellum. PSP-RS/PI and PSP-Cx had a higher total tau load than PSP-P; PSP-Cx showed a higher cortical tau load than PSP-RS/PI and PSP-P; and PSP-RS/PI had a higher subcortical tau load than PSP-P. Applying the MDS-3y, MDS-6y and MDS-last criteria; the PSP-RS/PI group increased (67.6, 70.6 and 70.6% respectively) whereas the PSP-Cx group decreased (8.8, and 8.8 and 11.8%). Then, only differences in total and subcortical tau burden between PSP-RS/PI and PSP-P were observed.InterpretationAfter the retrospective application of the new MDS-PSP criteria, total and subcortical tau load is higher in PSP-RS/PI than in PSP-P whereas no other differences in tau load between phenotypes were found, as a consequence of the loss of phenotypic diversity.

The discrimination power of a hyperspectral imaging system for image segmentation or object detection is determined by the illumination, the camera spatial–spectral resolution, and both the pre-processing and analysis methods used for image processing. In this study, we methodically reviewed the alternatives for each of those factors for a case study from the food industry to provide guidance in the construction and configuration of hyperspectral imaging systems in the visible near infrared range for food quality inspection. We investigated both halogen- and LED-based illuminations and considered cameras with different spatial–spectral resolution trade-offs. At the level of the data analysis, we evaluated the impact of binning, median filtering and bilateral filtering as pre- or post-processing and compared pixel-based classifiers with convolutional neural networks for a challenging application in the food industry, namely ingredient identification in a flour–seed mix. Starting from a basic configuration and by modifying the combination of system aspects we were able to increase the mean accuracy by at least 25 %. In addition, different trade-offs in performance-complexity were identified for different combinations of system parameters, allowing adaptation to diverse application requirements.