Fig. 1

Pipeline for quantification of forebrain projections to the LHA. a Neurons projecting to the LHA were identified by injecting rAAV-retro encoding Cre into the LHA of WT (or mSOD1) carrying a floxed ZsGreen reporter allele (panel 1). Brains were serially sectioned (2) and each section was manually annotated in WholeBrain (3). For precise quantification of neurons in each anatomical structure, the registered sections were parcellated and refined in Fiji (panel 4), and then segmented using a purpose-trained ilastik project (Density Counting workflow) (5) to determine the final neuronal counts per image. The WholeBrain dataset was also used to derive an independent anatomical annotation in the Allen Brain Atlas (backward warp transform field [atlas plate to section], 6.1; forward warp transformation field [section to atlas plate] 6.2) and 3D reconstructions (6.3). b Training of the ilastik cell density counting suite. Accuracy improvement of the classifier according to the number of images used for training. The increasing need of computing power is indicated by the time needed to process 60 images. Accuracy was calculated by comparing randomly selected and manually analyzed images. A final accuracy of 94% was achieved with the use of 250 images after adaptation of parameters for the Random Regression Forest