An optimised patient-specific approach to

Eur Radiol DOI 10.1007/s00330-013-2919-6

COMPUTED TOMOGRAPHY

An optimised patient-specific approach to administration of contrast agent for CT pulmonary angiography Charbel Saade & Roger Bourne & Fadi El-Merhi & Arjuna Somanathan & Dev Chakraborty & Patrick Brennan

Received: 24 February 2013 / Revised: 27 April 2013 / Accepted: 30 April 2013 # European Society of Radiology 2013

Abstract Objectives To investigate pulmonary vasculature opacification during CTPA using an optimised patient-specific protocol for administering contrast agent. Methods CTPA was performed on 200 patients with suspected PE. Patients were assigned to two protocol groups: protocol A, fixed 80 ml contrast agent; protocol B used a patient-specific approach. The mean cross-sectional opacification profile of 8 central and 11 peripheral pulmonary arteries and veins was measured and the arteriovenous contrast ratio (AVCR) calculated. Protocols were compared using Mann–Whitney U nonparametric statistics. Jack-knife alternative free-response receiver-operating characteristic (JAFROC) analyses assessed diagnostic efficacy. Interobserver variations were investigated using kappa methods. Results A number of pulmonary arteries demonstrated increases in opacification (P180 images/case), it was not possible in the time frame of this study for every case to be examined for diagnostic efficacy; therefore for each contrast protocol 80 complete cases were randomly chosen with an equal number of normal and abnormal cases. All 80 abnormal cases showed thromboembolism of varying degrees as defined by the radiologists’ reports, and abnormalities were validated through senior respiratory physician consultation and clinical criteria (Scarvelis and Wells Clinical Prediction Rules and positive laboratory D-dimer>0.8 mg/l). The remaining images showed normal pulmonary vasculature. Readers viewed images in two sittings (see below), one for each protocol, and the ordering of the protocol was performed in a counterbalanced way, where the first radiologist looked at protocol A first whilst the second looked at protocol B first, and this alternating approach was used for all readers. All pathology was visible on the transaxial

Quantitative Analysis Vessel opacification was measured for all cases in the transaxial images within the largest circular ROI that would fit within the lumen and exclude the vessel wall. The mean cross-sectional opacification profile of 8 central segments and 11 peripheral pulmonary arteries was measured. In cases where PE was identified, care was taken not to include the emboli within the ROI. Arterial and venous measurements were performed at the heart, central pulmonary vasculature, and pulmonary segments with a minimum ROI diameter of

Fig. 1 Anatomical location of measurements of the pulmonary vasculature

Eur Radiol

images, but prevalence of abnormality within the image bank was not revealed to the readers. The multi-reader analysis consisted of eight cardiothoracic radiologists certified by the American Board of Radiology, The Royal College of Radiologists, or the Royal Australian and New Zealand College of Radiologists for a mean number of 8.9 years (minimum, 3 years; maximum, 18 years). Readers were permitted to manipulate the window and level of the images. Each reader using a mouse cursor indicated suspicious findings wherever they were located and provided a decision confidence Table 3 Mean vessel opacification (HU)

level from 1–6 where 6 indicated that vascular pathology was definitely present and 1 represented pathology definitely not present. Statistical Analysis Vessel opacification was compared using a non-parametric Mann–Whitney test. The JAFROC analysis employed the Dorfman-Berbaum-Metz (DBM-MRMC) approach using readers as random and cases as fixed [22]. The analysis treated cases as fixed on the basis that the limited image sample size

Anatomical location Cardiac chambers Right atrium Left atrium Right ventricle Left ventricle Pulmonary trunk Right main pulmonary artery Left main pulmonary artery Atrial Vein Right superior Right inferior Left superior Left inferior Superior vena cava Pulmonary Arteries Right superior anterior Right superior posterior Right medial Right lateral Right anterior-basal Right posterior-basal Left apico-posterior Left inferior lingular Left anteromedial basal Left posterior-basal Pulmonary Veins Right superior anterior Right superior posterior Right medial Right lateral Right anterior-basal Right posterior-basal Left apico-posterior Left inferior lingular Left anteromedial basal Left posterior-basal

Protocol A

Protocol B

P value

420±167 199±96 327±114 396±209 339±98 320±97 313±103

166±84 167±65 342±126 131±71 362±97 337±88 323±88

0.0001 0.0094 0.06 0.0001 0.14 0.11 0.56

Mediastinum

205±74

177±73

0.003

Mediastinum

172±70 205±83 187±78 1,077±730

154±67 182±72 157±61 606±257

0.007 0.034 0.003