Imaging Cerebral Haemorrhage using MRI: Improved Sensitivity of Susceptibility Weighted Imaging (SWI) Compared to Gradient Echo Sequences (GRE)

Author(s): Punitha P, Karunanithi Rajamanickam, Krishnamoorthy A, Einstein A, Alex Daniel Prabhu K, Murugesan R

Background: The appearance of cerebral haemorrhage (CH) as common MR imaging findings raise the question of how MR image acquisition and haemorrhage area quantification can help in understanding brain pathological changes as the result of various disease conditions such as acute brain injury, hypertension amyloid angiopathy and other diverse pathology. Neuroimaging is essential for the clinician to identify the cause of haemorrhage. In this study, we examined 25 cerebral haemorrhage MR images with an objective of comparing two different sequences viz., susceptibility weighted imaging (SWI) and gradient recalled echo (GRE) on same patients and also to quantify the area of haemorrhage using manual selection and gray area threshold (automated selection) in the digital image.

Materials and Methods: In this prospective study, MR images were acquired from 25 subjects with acute brain trauma (16) , hypertension (5), and amyloid angiopathy (4), using SWI and GRE MR Sequences, with section thicknesses (3- 5 mm), and magnetic field strengths (1.5T). Individual CH was manually identified and analyzed for haemorrhage area using signal intensity changes by manual and automatic selection of binary threshold hypointense signal by ImageJ software package.

Results: By other parameters set as constant, we were able to quantify hypointense signal changes at regions of cerebral haemorrhage on SWI when compared to no such notable changes in GRE in the case of amyloid angiopathy patients. Apart from this, lesions prospectively identified on SWI had a significantly greater area measured on the SWI image than those not prospectively identified on GRE.

 Conclusions: There is an increased sensitivity in the CH detection in patients with amyloid angiopathy while using SWI sequence when compared to GRE in which we could not identify any signal changes in specific areas. Furthermore, the haemorrhage areas quantified by an automated cropping technique using ImageJ software yielded a precise lesion area method when compared to manual cropping. However, observation on increased sample size in this disease condition is required to confirm these preliminary findings to prognosticate CH.

© 2016-2022, Copyrights Fortune Journals. All Rights Reserved