VSD signal processing

We followed the previously developed procedure of drawing cell regions, image alignments, and normalization to extract the VSD signals from the sequence of images (Fathiazar and Kretzberg, 2015). The processing steps are explained in more details in the following sections.

The first step of VSD signal analyses was to define the cells' regions of interests. Although the maximum spatial resolution of the camera was 512 × 512 pixels, recordings with the desired temporal frequency of almost 100 frames per second could only be achieved with a lower spatial resolution of 64 × 128 pixels. Snapshot images with the full resolution of the camera (512 × 512 pixels), taken prior or right after the VSD recordings, were only used to define the regions of interests characterizing individual cells. Due to the location of the cell bodies on the spherical surface of leech ganglia, we identified neurons in up to three snapshot images from different depths in the stack. The regions of interests from the different depths were mapped to one selected snapshot image. Figure ​Figure1B1B depicts regions of interests representing cell bodies with cyan, purple, and red circles, detected in two different layers of a stack in a sample VSD experiment. The resulting picture was sampled down to fit the spatial resolution of VSD recordings (64 × 128 pixels).

To extract the cells' response traces from the sequence of images, we tracked the average luminance of the pixels inside the cells' regions of interest. Since the positions of the cells might change slightly over time due to movements of cells, ganglia, or the skin preparation, the defined regions of interest might not fit the actual cells' positions on all successive frames, leading to the so-called movement artifact. To remove this artifact, all the successive frames of each recording were mapped to the sixth frame using an optical flow method (Fathiazar and Kretzberg, 2015).

The optical flow method, however, could not correct extreme cell movements especially in tactile preparations with attached body-wall. During desheating, some cells were loosed, resulting in displacement from their original location. This displacement occurred most frequently in body-wall preparations, where the ganglion was pulled slightly toward posterior, to fit the small hole cut into the skin. In this condition, the glia sheath of the ganglion was removed while some cells were under strain. Therefore, cells for which the calculated flow field indicated extreme movement were excluded from further analysis (e.g., cells marked in black in Figures 6G, 7B).

The absolute signal level of each trace depended on the amount of the dye bound to the cell membrane and the relative positions of the cell to the microscope lens. Cells closer to the microscope or with a higher concentration of dye appeared brighter in the images, having higher signal values. Hence, to enable comparison of the cells within or across different experiments, the signal level of each cell was normalized to its mean luminance level, which was calculated by averaging the luminance level over 30 frames before the start of the stimulation (ΔF/F¯).

Photo-bleaching leads to a reduced signal level over time, due to light imposed dye destruction. To estimate the effect of bleaching, we recorded a control trial without stimulation following each sequence of recordings with stimuli. For each unstimulated recording, a linear line, representing the bleaching decay, was fitted to the VSD trace of the pixels in the ROI. We then removed the bleaching artifact from the cell responses to stimulation by subtracting this fitted line to the control trial, similar to (Fathiazar and Kretzberg, 2015). The processed VSD signals of two sample cells in response to intracellular P cell stimulations are shown in Figures 1G,H.