Title: Automated blood volume measurement on head CT

Main supervisor: Teresa Ullberg 

Reviewers: 

PhD, Associate professor of Radiology, Neuroradiologist, Anna Falk Delgado, Karolinska Institute, Stockholm

PhD, Neuroradiologist, Charalampos Georgiopoulos, Lund University, Lund 

Abstract

Background

Spontaneous Intracerebral hemorrhage (ICH) accounts for 10–20% of all strokes worldwide and is associated with a 30-day mortality rate of up to 40%. The strongest predictor of 30-day mortality and functional outcome for ICH patients is baseline hematoma volume which is rarely measured in clinical routine practice, primarily due to a lack of tools available for fast and reliable volumetric tools.Recently, automated image analysis software based on deep learning algorithms have been developed for detection and volume quantification of ICHon non-contrast computed tomography (NCCT). Implementing such automated imaging analysis tools in routine healthcare may improve early detection by prioritizing among radiological exams and reduce missed ICH diagnoses. However, the benefit of automated volumetric segmentation tools hinges on its accuracy.

Research questions

I. (Published)

Can hemorrhage volume be measured by automated image analysis software?
II. (Under final revision)
What is the degree of agreement between ICH diagnosis registered in Riksstroke and the radiological ICH diagnosis based on image evaluation? 

III. (Submitted to Neuroradiology, and currently under review)
What is the association between hematoma volume and the probability of Cerebral amyloid angiopathy (CAA) diagnosis in lobar ICH on NCCT?

IV. (In planning)
Where do currently commercially available automated solutions stand in comparison to manual segmentation for quantification of ICH volume on NCCT?

Preliminary results

I. 
Excellent agreement in volume quantification was shown between the fully automated software qER-NCCT and manual segmentation of ICH on NCCT

II. 
High rate of agreement was shown between ICH diagnoses registered in RS and the radiological evaluation. the most common coding error was traumatic bleeds that were classified as spontaneous ICH

III. 
Strong association was identified between baseline hematoma volume and an increased probability of CAA in lobar ICH patients on NCCT.

IV. 
No preliminary results yet. 

Significance

Hematoma volume is the single most important prognostic variable in ICH, even so it is rarely reported outside clinical trials. The main reason is that manual volume segmentation is time consuming and has a large inter-rater variability. With new fully automatic tools systematically evaluated within the proposed project, exact volume measurements will be only a couple of clicks away, and available for the radiologist at the initial image review. This will be a major advancement of ICH care and significantly aid in prognostication at an early stage, which may translate to better early treatment decisions.

Published studies

Accuracy of automated intracerebral hemorrhage volume measurement on non-contrast computed tomography: a Swedish Stroke Register cohort study.
Amir Hillal, Gabriella Sultani, Birgitta Ramgren, Johan Wassélius, Teresa Ullberg. Neuroradiology. 2023 Mar;65(3):479-488. Doi: 10.1007/s00234-022-03075-9.