Pre-Doctoral Research & Experiences

Role of Susceptibility Weighted Imaging (SWI) in Acute ischemic Stroke Penumbra Detection

Motivation: Definition and extent of penumbra in acute stroke patients, is an important information for thrombolytic/recanalizationtherapies which target to recover the salvageable brain tissue. Hypo-perfusion is established as standard to calculate penumbra and mismatch ratio with core of the infarct highlighted on diffusion-weighted- imaging(DWI). Various methods of estimating hypo-perfusion are reported for MRI-based penumbra calculation: such as dynamic-susceptibility-contrast(DSC) based metrics (MTT or rCBF). Pseudo-continuous-arterial-spin-labeling(pCASL) has also been used as a method of qualitatively assessing strokepenumbra area in recent studies and it has been proven as comparable with DSC perfusion based penumbra indications. All these reported methods are either invasive or time- consuming. In acute ischemic stroke, there is a sudden reduction in oxygen, whichfurther increases the OEF and results in higher concentration of deoxyhaemoglobin in the draining vein. These can be visualized asa prominent hypo-intense- vein-sign(PHVS) in the ischemic region on susceptibility weighted imaging(SWI) sequence. Studies suggest that PHVS can be an equivalent sign of penumbra region. However, visualization of PHVS and penumbra identificationbased on it, is rather poor compared to perfusion based methods.

Work: Developed a framework that enhances venous-structures seen on acute-ischemic-stroke SWI images, i.e. prominent-hypo-intense-vein-signs (PHVS) and identify ischemic-penumbra high-density PHVS presence. Quantitation of SWI/ADC mismatch-ratioand its validation with pCASL/ADC mismatch-ratio can lead to scan-time saving. Combination of minimum-intensity-projection, multi-scale Frangi-2D & 3D filtering and morphological-operation based approach was designed to enhance PHVS-features on SWI. ) stroke penumbra quantification from enhanced PHVS images was automatized using gabor texture and iterative graph-cut based segmentation. The infarct-core-volume (ICV) from ADC, SWI-penumbra-volume (SPV) from PHVS-feature-maps and pCASL-penumbra-volume (PPV) from ASL-CBF-maps were used to quantify mismatch ratios.