MRI sequences for detection of acute pulmonary embolism

Abstract

In recent years a range of imaging techniques have emerged to help diagnose patients with suspected acute Pulmonary Embolism (PE). This is particularly useful for those who are contraindicated (renal failure or allergies) to the contrast media that is needed to perform Computed Tomography Pulmonary Angiography (CTPA), which would be the usual diagnostic tool of choice. To aid the cohort of patients with this contraindication, we have investigated the option of using Magnetic Resonance Imaging (MRI) to diagnose PE. In this thesis, MRI sequences including gradient recall echo (more specifically balanced Steady State Free Precession [b-SSFP]) with different trajectories of data sampling, and diffusion weighted imaging (DWI) were assessed. None of the sequences investigated required the use of intravenous contrast media.

In Study I, we investigated a group of positive PE patients (verified by CTPA) alongside a volunteer group, who provided a negative PE control cohort. A b-SSFP sequence was assessed, using repetitive sampling of each slice position, in three different orthogonal planes. No triggering or breath hold techniques were used during imaging. This technique produced a large number of slices at each location for evaluation by radiologist. An excellent specificity and a good sensitivity were achieved.

In Study II, a group of positive PE patients (also verified by CTPA) and a control volunteer group were used to test the DWI technique, which is not used commonly for the investigation of thrombosis in the lungs. We compared DWI against the single slice per position approach of b-SSFP and CTPA, and demonstrated its capability to depict pulmonary embolism, finding a very high sensitivity but poor specificity for DWI.

In Study III, we tested two different sampling techniques for b-SSFP, Cartesian standard and golden angle radial sampling trajectories, to image the pulmonary arteries in ten volunteers and in two patients who had PE. We demonstrated the improvement of image quality when using radial trajectory sampling in comparison to the Cartesian technique. We also demonstrated that the post-reconstruction ‘sliding window’ method could be applied to the golden angle radial sampling schema when a different temporal resolution is needed.

In Study IV, we used the sequence tested in Study III (b-SSFP with golden angle radial and Cartesian sampling) in a clinical setting. The study included 64 patients who were suspected of having acute PE; all were examined while waiting for CTPA diagnostic testing. We compared radial sampling versus Cartesian, and also assessed post-reconstruction images of the radial sampling, with varying temporal resolution. The radial sampling with golden angle schema did not produce images of high enough quality to depict acute PE in patients.

In Study V, a retrospective overview of 57 patients (2012–2018) from our institution, with suspected acute PE was made. This group of patients was contraindicated to CTPA, and so were examined only using b-SSFP images. The clinical outcome of this cohort was obtained from the electronical medical record system up to twelve months after their MRI assessments. The MRI results allowed the clinicians to change or support their decision as to which treatment strategy they chose, in patients with or without PE.