J.SIM1, T.C.H.HUI.1, LOW.H.M.1, TAN.C.H.1
Tan Tock Seng Hospital1
Presence of microvascular invasion(MVI) indicates a poorer prognosis post-curative resection of hepatocellular carcinoma(HCC), with an increased chance of tumour recurrence. This study aims to evaluate the accuracy of texture analysis of pre-operative MRI in predicting MVI on histology of HCC.
IRB approval was obtained. Retrospective review of patients with new cases of HCC who underwent hepatectomy between 2007 and 2015 was performed. Exclusion criteria: no pre-operative MRI, significant movement artefacts, loss-to-follow-up, ruptured HCCs, previous hepatectomy and adjuvant therapy. Tumours were divided into MVI and non-MVI based on histology. Selected images of the tumour on post-contrast-enhanced T1-weighted MRI in the arterial, portovenous and equilibrium phases were analysed. Radiomics texture parameters were extracted based on the largest cross-sectional area of each tumor and analysed using MaZda software.
Five separate methods were performed. Methods 1, 2 and 3 exclusively made use of texture features derived from arterial, portovenous and equilibrium phases respectively. This was done to sieve out the features that had the most impact on classification accuracy. Methods 4 and 5 then made use of these comparatively significant features to attain optimal performance.
Fifty patients (43 male, mean age 67.0 years) with HCCs of varying sizes (mean:45.8 mm, range:12-120 mm) were divided into MVI(n=15) and non-MVI(n=35) groups. Methods 1 to 5 were able to classify patients into MVI and non-MVI groups with accuracies of 81.4%, 84.7%, 70.0%, 85.5% and 87.8% respectively.
Texture analysis of tumours on pre-operative MRI can predict the presence of MVI in HCC with accuracies of up to 87.8%.