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

In neuro–oncology microstructural imaging techniques, like diffusion–weighted MRI (DW–MRI), have been investigated to non–invasively derive patient–specific parameters that can be used for tumour characterization, treatment personalisation and monitoring, response assessment and prediction of radiotherapy outcomes. In particular, DW–MRI is opening up promising perspectives in radiotherapy applications as it is suitable for characterizing tissues at a microscopic scale (microstructure). However, as advanced MRI is rarely acquired in clinical settings, most studies propose metrics extracted from the conventional apparent diffusion coefficient (ADC), despite it being a sensitive but non–specific metric that encapsulates many features of the underlying tissue. Starting from conventional ADC, a recently published computational framework showed its potential for tumour characterization at the microscopic scale by means of synthetic cell substrates (which mimic the cellular packing of a tumour tissue) and a simulation tool. The aim of this study was (i) to evaluate the effectiveness of an error correction procedure; (ii) to provide a method that accounts for noise in the computational framework; (iii) to obtain a quantitative description of tumour microstructure from DW–MRI images of meningiomas that helps differentiating patients according to their histological sub–type.