1746. A Technique to Identify Isoattenuating Gallstones With Dual-Energy CT: An In Vivo Study
Authors * Denotes Presenting Author
  1. Todd Soesbe; UT Southwestern Medical Center
  2. Yin Xi; UT Southwestern Medical Center
  3. Matthew Lewis *; UT Southwestern Medical Center
  4. Hirokazu Saito; Kumamoto Chuo Hospital
  5. Lakshmi Ananthakrishnan; UT Southwestern Medical Center
  6. Kazuhiro Katahira; Kumamoto Chuo Hospital
  7. John Leyendecker; UT Southwestern Medical Center
To develop a dual-energy CT method for differentiating isoattenuating gallstones from bile and compare it with previously reported dual-energy CT methods by using a retrospective in vivo reader study.

Materials and Methods:
Dual-energy CT scans for 20 isoattenuating gallstone positive patients and 20 gallstone negative control patients were retrospectively collected from two separate institutions. All 40 patients (21 men; mean age, 62; age range, 30-88 years) were scanned on a Philips IQon dual-layer dual-energy CT system between July 2015 and Oct 2019. Positive and negative gallstone diagnoses were confirmed with either ultrasound, MRCP, or ERCP performed within 90 days of the CT scan. Conventional CT images, 40 keV monoenergetic images, and effective Z images were created for each patient. Two classes of segmented images were also created for each patient using two-dimensional (2D) histograms of either 1) Compton scattering and photoelectric attenuation data, or 2) denoised 200 keV and 40 keV monoenergetic data. Six readers evaluated the presence of isoattenuating gallstones in each of the five image types for each patient (300 images total). Intra- and inter-reader agreement was measured by weighted kappa, diagnostic performance was evaluated by using the mean area under the receiver operating characteristic curve (AUC) estimates, and pairwise comparisons between different image types were performed using Delong’s method.

For the 20 isoattenuating gallstone positive patients, 20 had gallbladder stones, 1 had cystic duct stones, 5 had common bile duct stones, and 4 had contrast-enhanced scans. For the 20 gallstone negative control patients, 15 had contrast-enhanced scans. Comparing the 2D histograms from all isoattenuating gallstone positive patients to all gallstone negative patients revealed that isoattenuating gallstones appear in a separate and unique 2D location relative to other tissues. A global ROI drawn around this unique 2D location permitted material differentiation and visual segmentation of the isoattenuating gallstones within the conventional CT images (i.e., the segmented images). Both classes of segmentation were successful at accurately differentiating isoattenuating gallstones from bile in a single comprehensive image. The reader study is currently in progress and will take 5 to 7 weeks to complete.

A recent ex vivo isoattenuating gallstone report used a similar 2D histogram method and reader study (1). This study showed that segmented images offered the highest mean area under the receiver operating characteristic (ROC) curve estimates (0.98-1.00) compared with previously reported dual-energy CT methods. Most notably for gallstones with diameters 9 mm or smaller. Furthermore, the vendor-neutral aspects of our improved in vivo method have the potential to make dual-energy CT the reference standard to evaluate gallstone disease (2). This would reduce the need for further gastrointestinal imaging with ultrasound or MRI. Leading to a reduction in imaging cost, time-to-diagnosis, and risk for patients with cholelithiasis or choledocholithiasis.