Local Means Threshold shows higher accuracy than current standard ROI detection methods.
<p>(A) Schematic of <i><i>C. elegans</i></i> body plan, with imaging regions displayed in B and C boxed in red and dark blue, respectively. (B) Top: Straightened <i><i>crop</i></i> extracted from the full-size confocal image below. Bottom: 63X co...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | , |
| Published: |
2025
|
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | <p>(A) Schematic of <i><i>C. elegans</i></i> body plan, with imaging regions displayed in B and C boxed in red and dark blue, respectively. (B) Top: Straightened <i><i>crop</i></i> extracted from the full-size confocal image below. Bottom: 63X confocal image of an L4-stage worm that is overexpressing the synaptic active zone marker CLA-1::GFP in the cholinergic motor neuron DA9, whose axon extends in the dorsal nerve cord. The crop outline is shown in green. Note the nearby dark signal from the gut (dark gray), including gut granules (pink asterisks). Scale Bars, 5µm. (C) Top: Straightened <i><i>crop</i></i> extracted from the same region as displayed in the full-size confocal image below. Bottom: 63X confocal image of endogenously tagged active zone marker Neurexin::Skylan-S in the dorsal nerve cord of an L4-stage worm. Gut granules indicated with pink asterisk. Outline of crop from 100X image of the same worm shown in green. Note the lower signal-to-noise ratio compared to B. Scale Bars, 5µm. (D) Example Crop from 63X confocal image with annotated ROIs (Green) from different methods showing: <i><i>i</i></i>. The raw crop; <i><i>ii.</i></i> 2D ROIs from Ground Truth annotation; <i><i>iii.</i></i> Projected 2D ROIs from 1D peak finding algorithm, with valid peaks labeled by black Xs and erroneous peaks (not corresponding to real puncta) labeled with orange Xs; <i><i>iv.</i></i> 2D ROIs from Global Thresholding algorithm, with missing puncta highlighted in red; and <i><i>v.</i></i> 2D ROIs from Local Means Thresholding algorithm. The local means thresholding method demonstrates the highest fidelity to the ground truth annotation. Scale Bar, 5µm. € Sensitivity (True Positive Rate) for each of the three thresholding methods, measured in the Overexpressed Synaptic Fluorophore Dataset (N = 52, red) and the Endogenous Synaptic Fluorophore dataset (N = 56, blue). (F) Specificity (True Negative Rate) for each of the same three thresholding methods, measured in the Overexpressed Synaptic Fluorophore Dataset (N = 52, red) and Endogenous Synaptic Fluorophore dataset (N = 56, blue). (G) Balanced Accuracy – the average of Sensitivity and Specificity – for each of the three thresholding methods, measured in the Overexpressed Synaptic Fluorophore Dataset (N = 52, red) and Endogenous Synaptic Fluorophore dataset (N = 56, blue).</p> |
|---|