CT is very accurate for the diagnosis of appendicitis,¹ however, it exposes patients to ionizing radiation and iodinated contrast, which can cause significant adverse events.² Alternatively, MRI has neither of these limitations. The purpose of this work is to determine the accuracy of a novel MRI protocol including unenhanced, contrast-enhanced, and diffusion-weighted imaging to diagnose appendicitis, using a combination of surgical and pathological findings and clinical follow up as the reference standard.

This is a HIPAA-compliant, IRB-approved prospective study of a convenience sample of patients presenting with abdominal pain to the emergency department of an academic medical center. Patients were eligible for enrollment if they were over 11 years old and had CT imaging ordered to evaluate for possible appendicitis. Patients underwent CT and MR imaging serially, within approximately 1 hour of each other.

CT scans of the abdomen and pelvis were performed using a 64 x 0.625 detector configuration 64 slice multi-detector CT (VCT, GE Healthcare, Waukesha, WI) following oral contrast and IV iohexol (Omnipage-300, GE Healthcare, London, UK) administration in the portal venous phase. Size-specific protocols for small, medium, and large body habitus ranged from 100-140 kVp, NI = 15-21, and Smart mA with mA range = 30-600. Images were reconstructed with 5 mm slice thickness at 3 mm intervals using a 40% ASIR blend in the axial, sagittal, and coronal planes.

MRI was performed on clinical 1.5T scanners (Signa HDxt CRM or TwinSpeed, Discovery MR450w) using an 8-channel body phased array coil. For contrast-enhanced T1w imaging, 0.1mmol/kg of gadobenate dimeglumine was administered at 2ml/s, followed by a 20ml saline flush injected at the same rate. Example images are shown in Figure 1. The MR protocol consisted of the following sequences:

1. T2w-SSFSE with (axial) and without fat-suppression in axial, coronal, and sagittal planes with the following imaging parameters: 320 x 256 matrix, 36cm FOV, 4mm slice/0mm gap, TR/TE=min/80ms, BW=±83kHz, and ARC parallel imaging (R=2) to minimize blurring.

2. 3D T1w fat-suppressed spoiled gradient echo images (LAVA) acquired during a 22s breath-hold, prior to contrast (axial), 40s after contrast (axial), 90s after contrast (coronal) and 3min after contrast (axial), with the following imaging parameters: 256 x 192 x 100 matrix, 256 x 192, 38cm x 30cm FOV, 3mm slice thickness, TR/TE=3.6/1.7, ARC parallel imaging (R=2.75).

3. Diffusion-weighted imaging (DWI) in the axial plane with the following imaging parameters: 128 x 128 matrix, 35cm FOV, 5mm slice/1mm gap, and b=50, b=500 (8 signal averages), acquired using respiratory triggering.

Three fellowship-trained abdominal radiologists, blinded to the original CT read, interpreted all MR and CT images independently and in random order, using a standardized data collection form. Multiple parameters were documented for each image set including characteristics of the appendix (size, location, etc), the likelihood of appendicitis, and the time required to interpret the images. Follow-up consisted of a chart review for pathological/surgical findings for patients who underwent appendectomy and follow-up phone interview with chart review for all others.

Continuous variables were summarized with descriptive statistics including 95% confidence intervals. Receiver operating characteristic (ROC) curve analysis for the likelihood of appendicitis are reported with area under the curve (AUC). Regression analysis was also performed to demonstrate the value of individual imaging characteristics when evaluating for appendicitis.

We enrolled 226 patients from 2/2012 to 8/2014 though images from the first 20 patients were used as a training set for our study readers and were therefore excluded from analysis, leaving CT and MR images from 206 patients for our final analysis. Of this cohort, there were 118 women (57%), the mean age was 31.6 years (range 13-75), and the prevalence of appendicitis was 31.1%. Summary test characteristics are listed in Table 1. ROC and AUC are shown in Figure 2. Mean total interpretation times were 4.3 minutes for MR and 2.1 minutes for CT. Increased DWI signal, appendiceal wall thickening, and periappendiceal inflammation had the greatest predictive power for the presence of appendicitis (Table 2).We found that the accuracy of this contrast-enhanced MRI protocol for suspected appendicitis in the general population is non-inferior to that of CE-CT. This study is also novel for several reasons: 1) the general population was studied rather than targeted populations (e.g. pregnant women), 2) all patients underwent both CT and MRI, allowing for direct comparison of the two technologies, and 3) this MRI protocol incorporates both intravenous contrast enhancement and DWI.We suggest that this MRI protocol is a suitable primary test for diagnosing appendicitis, though a prospective, multi-center study would more definitely prove this.