Application Of Quantitative Neuroimaging (Qni) Segmentations In Dementia


1 Southend University Hospital, Southend on sea, United Kingdom

2 Queen Square, University College London, London, United Kingdom

3 Southend university hospital


In the current radiological practice, there is no gold standard for the diagnosis of dementia and we are dependent on qualitative visual assessment of images such as visual scores of atrophy. To increase the accuracy of the diagnosis, we aim to generate an automated analysis of grey matter atrophy by segmentation using the software Geodesical information flows (GIF) to give us a quantitative assessment result. The most determining factor when analysing images via this method is the regional volume of the different areas of the brain derived.
Segmentation is routinely used for research purposes at the moment.
 There are four stages in this study: credibility, accuracy, management and socio-economic study. The credibility and accuracy studies are presented in this project. In the first stage, the credibility study, T1 weighted images of 20 patients with Alzheimers disease (AD), frontotemporal dementia (FTD) and controls were assessed and run through GIF. In the accuracy study, 47 cases were analysed by two raters, each with and without the segmentation report from GIF.
 Overall, a significant relationship was seen between the visual scores and regional volumes. However, no specific pattern was observed in parameters analyses between with and without
GIF segmentation report. The addition of QNI report showed an improvement in the Kappa value in terms of agreement with the pathological diagnosis in both raters.

Volume 34, Issue 4
December 2018
Pages 104-104
  • Receive Date: 26 October 2018
  • Accept Date: 26 October 2018
  • First Publish Date: 01 December 2018