Impact of A structured report template based liver imaging reporting and data system (LI-RADS) on diagnostic accuracy

PURPOSE: To assess impact of implementing a structured report template on the diagnostic accuracy of magnetic resonance imaging (MRI) computed tomography (CT) reports by evaluating them against the elements of LI-RADSv2017 based structured report template. MATERIAL AND METHODS: This retrospective study was approved by our Institutional Review Board (IRB) and included 238 patients with liver lesions in cirrhotic liver imaged with abdominal CT and MRI between January 2009 and February 2018. One radiologist interpreted all examinations using structured reporting template. The template was created based on LI-RADS version 2017. After the interpretation of examinations, both structured reports (SR) and free-text reports were assessed separately for documentation of the presence or absence of major imaging features and ancillary features. RESULTS: 238 patients (172 female, 66 male; mean age 62 years [range: 27-89 years]), CT [n=66 (27.7%)] and MRI [n=172 (%72.2)] free-text repots and structured reports were evulated. For seven LI-RADS categories each contained 34 patients [Categories: LR-1, LR-2, LR-3, LR-4, LR-5, not categorizable(LR-NC) and tumor in vein(LR-TIV)]. SR resulted in increased documentation of 5 of 5 major features including arterial phase hyperenhancement (p< .001), wash-out (p<.001), threshold growth (p=0.004 ), enhancing capsule (p<.001),lesion diameter (p<.001). SR resulted in increased documentation of 2 of 7 ancillary features favoring benignity including marked T2 hyperintensity (p< .001) and size stability ≥ 2 years (p< .001). SR resulted in increased documentation of 4 of 5 ancillary features favoring HCC in particular including; mosaic architecture (p< .001), nodule-in-nodule architecture, (p=0.008), fat in mass, more than adjacent liver (p=0.031), nonenhancing capsule (p=0.016). SR resulted in increased documentation of 2of 9 ancillary features favoring malignancy including mild-moderate T2 hyperintensity (p< .001) and restricted diffusion (p< .001). CONCLUSION: The use of LI-RADS v2017 based structured report template results in significantly increased reporting of the major and ancillary features compared with free-text reporting.