Ferran Prados Carrasco - Nooshin Ghavami - CMIC/WEISS joint Seminar Series

Ferran Prados Carrasco - Title: Procedures to harmonise MRI data: image storage and transfer

Abstract: Nowadays, research groups are starting to manage a large amount of data. In a few years, we moved from small studies with hundreds of subjects acquired cross-sectionally, to large multi-modal studies with several timepoints and thousands of subjects. It also has increased collaboration across centres and consequently the need to merge different studies. All these new scenarios highlighted the key importance of efficient data storage, managing and sharing. An optimised use of a large amount of data that research groups are handling have an incredibly positive impact in the research results, ease the follow-up of the studies, facilitate the transfer of studies across researchers, allows multi-user work over the same dataset, optimise the computational resources avoiding computing again and again the same steps and, nevertheless, it saves money. In this seminar, I’m going to explain how CMIC has contributed to optimise data curation and handling within the Queen Square Multiple Sclerosis Centre.

Nooshin Ghavami - Title: Investigating the Impact of Network Architecture on the Accuracy of Volume Measurement and MRI-Ultrasound Prostate Registration

Abstract: Convolutional neural networks (CNNs) have recently led to significant advances in automatic segmentation of anatomical structures in medical images, and a wide variety of network architectures are now available to the research community. For applications such as segmentation of the prostate in MR images, the results of the PROMISE12 online algorithm evaluation platform have demonstrated differences between the best-performing segmentation algorithms in terms of numerical accuracy using standard metrics such as the Dice score and boundary distance. These small differences in the segmented regions/boundaries outputted by different algorithms may potentially have an unsubstantial impact on the results of downstream image analysis tasks, such as estimating organ volume and multimodal image registration, which inform clinical decisions. This impact has not been previously investigated. In this talk, I will describe our experience in quantifying the accuracy of six different CNNs in segmenting the prostate in 3D patient T2-weighted MRI scans and compared the accuracy of organ volume estimation and MRI-ultrasound registration errors using the prostate segmentations produced by different networks. The result provides a real-world example that these networks with different segmentation performances may potentially provide indistinguishably adequate registration accuracies to assist prostate cancer imaging applications. I will conclude by recommending that the differences in the accuracy of downstream image analysis tasks that make use of data output by automatic segmentation methods, such as CNNs, within a clinical pipeline should be taken into account when selecting between different network architectures, in addition to reporting the segmentation accuracy.