Journal: bioRxiv

Article Title: Axon Diameter Measurements using Diffusion MRI are Infeasible

doi: 10.1101/2020.10.01.320507

Figure Lengend Snippet: Scatter plot of fitted diameters with mean fitted diameter (green line) and 80% confidence interval (blue lines). For each diameter between 0.1 μ m and 5 μ m, we generated 10000 noisy restricted signals using and Gaussian noise of SNR 30 (left) and 300 (right). The signals are generated for realistic in-vivo setting ( D 0 = 2 μ m 2 /ms) with a Connectom-like acquisition (single “direction”, G = 300 mT/m, δ = Δ = 40 ms). The orange line corresponds to d min using the framework by .

Article Snippet: For human in-vivo measurements, the Siemens Connectom MRI scanner (Siemens Healthineers, Erlangen, Germany) is the system that produces by far the strongest diffusion gradients ( G max = 300 mT/m).

Techniques: Generated, In Vivo

Journal: bioRxiv

Article Title: Axon Diameter Measurements using Diffusion MRI are Infeasible

doi: 10.1101/2020.10.01.320507

Figure Lengend Snippet: Example of the unresolvability of distribution fitting. The ground truth signal was generated from a combination of 2 parallel cylinders; 30% signal fraction with diameter d 1 = 4.5 μ m and 70% d 2 = 3.5 μ m (shown as white dot in the center plot) with in-vivo diffusivity ( D 0 = 2 μ m 2 /ms) and a Connectom-like acquisition with three gradient pulse durations (G = 300 mT/m, Δ = 50 ms, δ = [ , , ] ms). The parameters were selected so that the smallest diameter was comfortably above the “typical” diameter limit for δ = 30 (compared to the limit for SNR = 30, this experiment is noiseless). The 9 subplots represent all combinations of diameters between 0.1 and 6 μ m, sliced uniformly at signal fractions between 10% and 50%. The blue “path” corresponds to parameter combinations yielding a signal less than 1% signal decay different than the noiseless ground truth. It forms a surface spanning most of the 3D parameter space, rendering any distribution fitting impossible for non-absurd SNR. Section A.4 showcases the same experiment for diameters closer to human axons.

Article Snippet: For human in-vivo measurements, the Siemens Connectom MRI scanner (Siemens Healthineers, Erlangen, Germany) is the system that produces by far the strongest diffusion gradients ( G max = 300 mT/m).

Techniques: Generated, In Vivo

Journal: bioRxiv

Article Title: Axon Diameter Measurements using Diffusion MRI are Infeasible

doi: 10.1101/2020.10.01.320507

Figure Lengend Snippet: Example of the unresolvability of distribution fitting. The ground truth signal was generated using a gamma distribution of diameter count (shown as a red dot in the center plot) with in-vivo diffusivity ( D 0 = 2 μ m 2 /ms) and a Connectom-like acquisition with three different gradient pulse durations (G = 300 mT/m, Δ = 50 ms, δ = [ , , ] ms). The center plot represents all combinations of shape and peak location characterizing different gamma distributions. The dark blue “path” corresponds to parameter combinations yielding a signal less than 0.1% signal decay different than the noiseless ground truth. It forms a path spanning across most of the 2D parameter space, rendering distribution fitting unreliable for non-absurd SNR. The 4 side plots show examples of various gamma distributions from the center plot of wildly different shapes generating roughly indistinguishable signals.

Article Snippet: For human in-vivo measurements, the Siemens Connectom MRI scanner (Siemens Healthineers, Erlangen, Germany) is the system that produces by far the strongest diffusion gradients ( G max = 300 mT/m).

Techniques: Generated, In Vivo