A novel higher order appearance texture analysis to diagnose lung cancer based on a modified local ternary pattern

Ahmed Alksas, Ahmed Shaffie, Mohammed Ghazal, Fatma Taher, Adel Khelifi, Maha Yaghi, Ahmed Soliman, Eric Van Bogaert, Ayman El-Baz

Abstract: Highlights•Current lung cancer diagnostic methods fail to diagnose small malignant lung nodules and large nodules located away from large-diameter airways.•Lung nodules’ texture and morphology can be used to differentiate malignant and benign nodules.•The traditional local binary/ternary pattern are not able to accurately classify nodule inhomogeneity.•Using three different inhomogeneity levels (instead of two) capture the nodule’s inhomogeneity as well as its morphology in a more accurate and flexible way.•Hyper-tuned stacking-based classification architecture get the best of its embedded meta-models.