Abstracts

Rationale: The comparison between MRI and histopathology is essential to validate new MRI techniques and to improve the detection of epileptogenic lesions and surgical treatment. Recently, several attempts have been made to coregister in vivo MRI with the corresponding histopathology in surgically resected tissues (Eriksson et al, 2005; Goubran et al, 2015) and an intermediate step through the ex vivo high resolution MRI of the sample can be useful providing an enormous gain in image quality and allowing to compensate the deformations due to tissue processing (Goubran et al, 2013; Zucca et al, 2016). In the present work, as an implementation of this procedure, we propose a method to reconstruct the anatomical localization of intracerebral electrodes, with the identification of single leads, on the histological sections. We developed a registration protocol to match the in vivo onto the ex vivo MR imaging with final aim to correlate the electrophysiological signal derived from a single lead with the corresponding neuropathology. Methods: Four en bloc surgical resections, after fixation, were scanned using a 7T MRI and T2-weighted images were acquired. Then, the specimens were sectioned into serial 5-mm thick slabs according to the ex vivo image-acquisition geometries, paraffin embedded and processed for neuropathology. An in vivo/ex vivo coregistration protocol was developed. Since ex vivo images were natively in direct geometrical correspondence with histology slices, the goal of the coregistration process was the inverse transform function T in,ex  that maps the clinical MRI space to the ex vivo MRI. Finally, the ex vivo MR images were compared with digitized histological sections. Conductive leads, visible in CT-MRI superposition images, were segmented and translate on histological slices. A qualitative and quantitative validation of the final result was performed. Results: Step-by step, qualitative visual inspection showed an increasing matching between the native ex vivo MRI with the transformed in vivo MRI in terms of anatomical structures or boundaries and electrodes position. The quantitative evaluation of the registration protocol based on intracerebral leads displacement in the two different modalities report a mean error ranging between 0.57 and 2.42 mm with better accuracy when the electrode trajectory was in the core of the sample. The electrode tracks were clearly identifiable in histological and immunohistological staining. Further to the in vivo/ex vivo coregistration protocol, we were able to identify the electrode and the specific lead generating that track and its related SEEG signal. Conclusions: We propose a coregistration pipeline to provide morphological correspondence between in vivo MRI, histology and intracerebral electrical signal. This method could be used to study the anatomo-pathological features of the tissue in relation to the locally generated electrical activity. Funding: European Union grant FP7 (DESIRE) No. 602531 Italian Ministry of Health grant RF-2011-02350578Associazione ‘‘Paolo Zorzi’’ per le Neuroscienze