A Microstructure Model from Conventional Diffusion MRI of Meningiomas: Impact of Noise and Error Minimization

Microstructural parameters from DW‐MRI for tumour characterization and local recurrence prediction in particle therapy of skull‐base chordoma

“…From sensitivity analyses it was found that the proposed procedure requires a larger neighborhood (n) when considering noisy simulations to properly describe the local error made around the input data and so to better compute the corrective error (i.e., a wider neighborhood allows mitigating, by averaging, for the effects of noise). These results obtained from sensitivity and accuracy analyses are comparable with those found in preliminary analyses earlier proposed in Morelli et al 36 in which, however, only 30 noise realizations were performed, a SNR almost doubled was applied, and a different sequence was employed for the simulated DWI signals. Although further tests are needed, our study provides promising insights about the robustness of the estimates to different noise levels and different DWI protocols, which are critical aspects that need to be considered for repeatability and clinical applicability of the proposed approach.…”

“…31 Specifically, the error introduced by experimental noise and possible ambiguity in mapping ADC to microstructural parameters (different microstructural environment can give the same ADC) was investigated. Therefore, a procedure was implemented in order to propose a way to measure this error and then to refine the estimates accordingly, similarly to Morelli et al 36 Building on top of the original workflow (Figure 1a), such procedure (Figure 1b) involved:…”