Purpose:　An　accurate　and　reliable　brain　atlas,　as　the　role　of　navigation,　should　be　effective　and　vital　to　differentiate　the　patients　with　temporal　lobe　epilepsy　(TLE)　from　normal　controls　(NCs).　The　purpose　of　this　study　is　to　compare　the　classification　performance　of　identifying　TLE　patients　based　on　different　atlases,　which　were　Desikan-killiany　(DK)　atlas,　Destrieux　(DS)　atlas　and　Brainnetome　(BN)　atlas.　Methods:　Twenty-three　patients　with　TLE　and　thirty　NCs　were　recruited　for　our　study.　Seven　morphological　features　of　ROIs　were　calculated　firstly.　Then　individual　morphological　brain　network　were　constructed.　After　that,　least　absolute　shrinkage　and　selection　operator　(LASSO)　algorithm　was　used　in　feature　selection.　Finally,　classification　with　support　vector　machine　(SVM)　and　leave-one-out　cross-validation　(LOOCV)　were　employed　for　the　training　and　evaluation　of　the　classifiers.　Results:　The　performance　of　the　experiments　using　BN　atlas　was　better　than　DK　atlas　and　DS　atlas.　LASSO　algorithm　used　for　feature　selection　can　improve　the　classification　performance.　The　SVM　analysis　using　BN　atlas　revealed　best　classification　with　accuracy　of　92.45%　and　90.57%　respectively　based　on　network　properties　and　morphological　features.　Conclusion:　This　study　suggested　that　the　choice　of　atlases　is　important　in　the　computer-aided　classification　of　TLE.　©　2021,　Taiwanese　Society　of　Biomedical　Engineering.