| Introduction: We developed novel computer-aided diagnostic software (MADARA, Inspeedia, Japan) to quantify EUS/CE-EUS findings. Aim and methods: This study evaluated the utility of (1) EUS with MADARA for diagnosis of chronic pancreatitis(CP) compared with the Rosemont classification(RC), and (2) CE-EUS with MADARA for differential diagnosis between pancreatic cancer(PC) and CP with pseudotumor. Using our software, a region of interest (ROI) divided into 100 cells was placed to cover an area within the pancreatic lesion on EUS/CE-EUS images. Then the number of adjoining cells with a gray scale difference was automatically calculated in each frame (MADARA index: MI). Results: (1) EUS images of 46 patients with suspected CP were analyzed. The mean MI for a diagnosis of normal, indeterminate, suspicious, and consistent according to the RC was 40.4, 112.8, 107.5, and 134, respectively, and MI was strongly correlated with the RC (p<0.001). (2) When CE-EUS video images of 39 patients with PC and 20 with CP were analyzed, the mean MI was significantly higher for PC than CP (15.6 vs. 6.1, p<0.001). Combined assessment of the mean MI and time-intensity curve, using an MI cut-off value of 10.3, yielded a sensitivity of 94.3% and specificity of 95% for differentiating PC from CP with pseudotumor. Conclusion: Performing EUS/CE-EUS with “MADARA” diagnostic software to quantify the findings is useful for diagnosis of pancreatic diseases. |
