There are many active research projects accessing and applying shared ADNI data. Use the search above to find specific research focuses on the active ADNI investigations. This information is requested annually as a requirement for data access.
| Principal Investigator | |
| Principal Investigator's Name: | Henry Tregidgo |
| Institution: | University College London |
| Department: | Medical Physics |
| Country: | |
| Proposed Analysis: | In collaboration with Juan Eugenio Iglesias and researchers at both UCL and MIT, we plan to use the ADNI dataset to evaluate novel thalamic segmentation techniques. Specifically, we aim to build on our previously published probabilistic atlas comprising 52 thalamic nuclei (Iglesias et. al. NeuroImage 2018). We plan to apply labels from this atlas to subjects from the ADNI using the original structural MRI based algorithm, an algorithm based on combined structural and diffusion MRI (Iglesias et. al. IPMI 2019, Tregidgo et. al. NeuroImage under-review) and a proposed convolutional neural network (CNN) implementation. The resulting segmentations and volumes would then be used to compare the three methods for both accuracy and utility in a downstream task. In the original atlas paper (Iglesias et. al. NeuroImage 2018), we describe a Bayesian inference algorithm to apply the atlas to structural MRI acquisitions. Subjects from the ADNI were then segmented and the resulting volumes used to discriminate between subjects with AD and healthy controls, allowing evaluation of the segmentation algorithms’s utility for downstream tasks. The structural MRI based algorithm allows contrast agnostic segmentation of the thalamus and identification of 52 histological labels in vivo. However, comparison of structural segmentations with diffusion MRI identified a lack of contrast between the thalamus and surrounding tissue as well as between individual thalamic nuclei on structural imaging (Iglesias et. al. IPMI 2019). This has led to our development of an algorithm to jointly segment structural and diffusion MRI (Tregidgo et. al. NeuroImage under-review). This joint segmentation algorithm shows improved identification of internal and extermal thalamic boundaries. We propose to develop a CNN implementation of our joint segmentation algorithm, with the aim of reducing both the impact of partial volume effects on the resulting segmentations and the required time to produce a segmentation. We would then apply all three segmentation algorithms to subjects from the ADNI in order to compare the three methods for both accuracy and utility in a downstream task. References: Iglesias, J.E., Insausti, R., Lerma-Usabiaga, G., Bocchetta, M., Leemput, K.V., et al., 2018. A probabilistic atlas of the human thalamic nuclei combining ex vivo MRI and histology. NeuroImage 183, 314 – 326.50 Iglesias, J.E., Van Leemput, K., Golland, P., Yendiki, A., 2019. Joint inference on structural and diffusion MRI for sequence-adaptive Bayesian segmentation of thalamic nuclei with probabilistic atlases, in: Chung, A.C.S., Gee, J.C., Yushkevich, P.A., Bao, S. (Eds.), Information Pro-54 cessing in Medical Imaging, Springer International Publishing, Cham. pp. 767–779 Tregidgo, H.F.J., Soskic, S., Althonayan, J., Maffei, C., Van Leemput, K., et al., under-review Accurate Bayesian segmentation of thalamic nuclei using diffusion MRI and an improved histological atlas |
| Additional Investigators |
