Skip to main content

Table 3 Classification performance on Epistroma dataset

From: Texture features in the Shearlet domain for histopathological image classification

Method Perf RP + Magnitude RP Magnitude
SVM DTB SVM DTB SVM DTB
(1): Baseline performance of existing Shearlet-based methods
Vo et al. ACC% \(\underline{96} .\underline{44} \% \pm \underline{1} .\underline{47} \%\) \(95.42\% \pm 1.03\%\) \(95.50\% \pm 1.52\%\) \(94.26\% \pm 1.95\%\) \(94.48\% \pm 1.54\%\) \(94.26\% \pm 1.35\%\)
AUC \(\underline{0} .\underline{9895} \pm \underline{0} .\underline{0082}\) \(0.9897 \pm 0.0047\) \(0.9873 \pm 0.0078\) \(0.9839 \pm 0.0087\) \(0.9830 \pm 0.0083\) \(0.9763 \pm 0.0102\)
Sen \(\underline{0} .\underline{9685} \pm \underline{0} .\underline{0131}\) \(0.9443 \pm 0.0162\) \(0.9685 \pm 0.0117\) \(0.9467 \pm 0.0268\) \(0.9588 \pm 0.0182\) \(0.9564 \pm 0.0152\)
Prec \(\underline{0} .\underline{9722} \pm \underline{0} .\underline{0170}\) \(0.9787 \pm 0.0083\) \(0.9577 \pm 0.0260\) \(0.9576 \pm 0.0220\) \(0.9501\pm 0.0219\) \(0.9487 \pm 0.0193\)
Meshkini and Ghassemian ACC% \(\textit{96.08}\% \pm \textit{1.24}\%\) \(\textit{96.29}\% \pm \textit{1.86}\%\) \(\textit{95.35}\% \pm \textit{1.37}\%\) \(\textit{96.37} \% \pm \textit{2.30}\%\) \(95.35\% \pm 1.49\%\) \(96.44 \% \pm 2.20\%\)
AUC \(\textit{0.9926} \pm \textit{0.0046}\) \(\textit{0.9929} \pm \textit{0.0041}\) \(\textit{0.9887} \pm \textit{0.0058}\) \(\textit{0.9919} \pm \textit{0.0077}\) \(0.9899 \pm 0.0050\) \(0.9908 \pm 0.0068\)
Sen \(\textit{0.9710} \pm \textit{0.0190}\) \(\textit{0.9600} \pm \textit{0.0222}\) \(\textit{0.9685} \pm \textit{0.0207}\) \(\textit{0.9600} \pm \textit{0.0269}\) \(0.9722 \pm 0.0161\) \(0.9564 \pm 0.0274\)
Prec \(\textit{0.9642} \pm \textit{0.0169}\) \(\textit{0.9778} \pm \textit{0.0151}\) \(\textit{0.9550} \pm \textit{0.0183}\) \(\textit{0.9791} \pm \textit{0.0184}\) \(0.9517 \pm 0.0186\) \(0.9840 \pm 0.0181\)
Zhou et al. ACC% \(\textit{89.68}\% \pm \textit{1.33}\%\) \(\textit{88.30}\% \pm \textit{2.68}\%\) \(\textit{81.69}\% \pm \textit{2.53}\%\) \(\textit{79.65} \% \pm \textit{3.11}\%\) \(84.45\% \pm 2.98\%\) \(87.13 \% \pm 2.95\%\)
AUC \(\textit{0.9506} \pm \textit{0.0158}\) \(\textit{0.9400} \pm \textit{0.0223}\) \(\textit{0.8748} \pm \textit{0.0222}\) \(\textit{0.8721} \pm \textit{0.0367}\) \(0.9110 \pm 0.0194\) \(0.9416 \pm 0.0244\)
Sen \(\textit{0.9370} \pm \textit{0.0219}\) \(\textit{0.9103} \pm \textit{0.0287}\) \(\textit{0.8934} \pm \textit{0.0363}\) \(\textit{0.8666} \pm \textit{0.0343}\) \(0.9236 \pm 0.0141\) \(0.9102 \pm 0.0327\)
Prec \(\textit{0.8961} \pm \textit{0.0183}\) \(\textit{0.8968} \pm \textit{0.0275}\) \(\textit{0.8186} \pm \textit{0.0251}\) \(\textit{0.8088} \pm \textit{0.0301}\) \(0.8357 \pm 0.0330\) \(0.8799 \pm 0.0282\)
Dong et al. ACC% \(\textit{95.72}\% \pm \textit{1.54}\%\) \(\textit{95.28}\% \pm \textit{1.66}\%\) \(\textit{94.84}\% \pm \textit{1.59}\%\) \(\textit{93.39}\% \pm \textit{2.11}\%\) \(95.86\% \pm 1.32\%\) \(93.75\% \pm 1.15\%\)
AUC \(\textit{0.9882} \pm \textit{0.0071}\) \(\textit{0.9837} \pm \textit{0.0089}\) \(\textit{0.9836} \pm \textit{0.0071}\) \(\textit{0.9776} \pm \textit{0.0132}\) \(0.9884 \pm 0.0082\) \(0.9836 \pm 0.0101\)
Sen \(\textit{0.9551} \pm \textit{0.0162}\) \(\textit{0.9503} \pm \textit{0.0271}\) \(\textit{0.9503} \pm \textit{0.0218}\) \(\textit{0.9187} \pm \textit{0.0300}\) \(0.9685 \pm 0.0103\) \(0.9600 \pm 0.0182\)
Prec \(\textit{0.9735} \pm \textit{0.0227}\) \(\textit{0.9707} \pm \textit{0.0178}\) \(\textit{0.9637} \pm \textit{0.0211}\) \(\textit{0.9697} \pm \textit{0.0201}\) \(0.9630 \pm 0.0194\) \(0.9381 \pm 0.0231\)
(2): Proposed Shearlet-Based Methods For Textured Bio-medical Image Classification
CM ACC% \(\underline{97} .\underline{24} \% \pm \underline{1} .\underline{27} \%\) \(94.41\% \pm 1.48\%\) \(96.66\% \pm 1.14\%\) \(94.99\% \pm 1.43\%\) \(96.00\% \pm 1.30\%\) \(94.19\% \pm 2.19\%\)
AUC \(\underline{0} .\underline{9917} \pm \underline{0} .\underline{0067}\) \(0.9819 \pm 0.0126\) \(0.9925 \pm 0.0064\) \(0.9828 \pm 0.0116\) \(0.9863 \pm 0.0087\) \(0.9828 \pm 0.0091\)
Sen \(\underline{0} .\underline{9733} \pm \underline{0} .\underline{0126}\) \(0.9419 \pm 0.0258\) \(0.9672 \pm 0.0129\) \(0.9503 \pm 0.0134\) \(0.9636 \pm 0.0163\) \(0.9418 \pm 0.0294\)
Prec \(\underline{0} .\underline{9807} \pm \underline{0} .\underline{0150}\) \(0.9646 \pm 0.0202\) \(0.9769 \pm 0.0130\) \(0.9658 \pm 0.0175\) \(0.9696 \pm 0.0116\) \(0.9607 \pm 0.0186\)
LBP ACC% \(95.64\% \pm 1.32\%\) \(95.57\% \pm 2.04\%\) \(95.50\% \pm 1.06\%\) \(94.04\% \pm 1.56\%\) \(95.71\% \pm 1.39\%\) \(93.90\% \pm 1.75\%\)
AUC \(0.9890 \pm 0.0073\) \(0.9905 \pm 0.0069\) \(0.9871 \pm 0.0076\) \(0.9850 \pm 0.0054\) \(0.9876 \pm 0.0088\) \(0.9800 \pm 0.0113\)
Sen \(0.9624 \pm 0.0146\) \(0.9479 \pm 0.0261\) \(0.9661 \pm 0.0178\) \(0.9394 \pm 0.0221\) \(0.9600 \pm 0.0153\) \(0.9406 \pm 0.0239\)
Prec \(0.9653 \pm 0.0203\) \(0.9779 \pm 0.0210\) \(0.9594 \pm 0.0133\) \(0.9607 \pm 0.0176\) \(0.9687 \pm 0.0206\) \(0.9573 \pm 0.0194\)
LOSIB ACC% \(96.29\% \pm 1.47\%\) \(95.13\% \pm 1.61\%\) \(95.50\% \pm 1.32\%\) \(96.65\% \pm 1.51\%\) \(95.35\% \pm 1.85\%\) \(93.10\% \pm 1.29\%\)
AUC \(0.9900 \pm 0.0069\) \(0.9870 \pm 0.0074\) \(0.9868 \pm 0.0073\) \(0.9911 \pm 0.0065\) \(0.9892 \pm 0.0087\) \(0.9825 \pm 0.0103\)
Sen \(0.9733 \pm 0.0096\) \(0.9454 \pm 0.0245\) \(0.9721 \pm 0.0152\) \(0.9624 \pm 0.0218\) \(0.9758 \pm 0.0141\) \(0.9503 \pm 0.0276\)
Prec \(0.9656 \pm 0.0215\) \(0.9729 \pm 0.0179\) \(0.9541 \pm 0.0203\) \(0.9816 \pm 0.0132\) \(0.9487 \pm 0.0239\) \(0.9362 \pm 0.0194\)
SFTA ACC% \(95.71\% \pm 1.54\%\) \(93.90\% \pm 1.81\%\) \(95.28\% \pm 2.16\%\) \(91.57\% \pm 1.91\%\) \(94.33\% \pm 1.52\%\) \(92.01 \% \pm 2.07\%\)
AUC \(0.9881 \pm 0.0064\) \(0.9841 \pm 0.0106\) \(0.9870 \pm 0.0068\) \(0.9775 \pm 0.0093\) \(0.9824 \pm 0.0080\) \(0.9732 \pm 0.0125\)
Sen \(0.9661 \pm 0.0137\) \(0.9297 \pm 0.0282\) \(0.9624 \pm 0.0145\) \(0.9006 \pm 0.0265\) \(0.9636 \pm 0.0114\) \(0.9455 \pm 0.0221\)
Prec \(0.9628 \pm 0.0169\) \(0.9676 \pm 0.0184\) \(0.9593 \pm 0.0233\) \(0.9572 \pm 0.0266\) \(0.9438 \pm 0.024\) \(0.9232 \pm 0.0178\)
(3): Integrating Shearlet-based existing techniques with our proposed methods
Fusion #1 ACC% \(96.59\% \pm 1.24\%\) \(95.93\% \pm 1.95\%\) \(96.37\% \pm 1.14\%\) \(94.98\% \pm 0.73\%\) \(95.71\% \pm 1.25\%\) \(95.13 \% \pm 1.37\%\)
AUC \(0.9889 \pm 0.0111\) \(0.9910 \pm 0.0070\) \(0.9889 \pm 0.0103\) \(0.9838 \pm 0.0131\) \(0.9877 \pm 0.0104\) \(0.9841 \pm 0.0104\)
Sen \(0.9612 \pm 0.0126\) \(0.9624 \pm 0.0224\) \(0.9612 \pm 0.0179\) \(0.9406 \pm 0.0194\) \(0.9563 \pm 0.0155\) \(0.9503 \pm 0.0201\)
Prec \(0.9816 \pm 0.0140\) \(0.9696 \pm 0.0175\) \(0.9780 \pm 0.0135\) \(0.9752 \pm 0.0139\) \(0.9721 \pm 0.0185\) \(0.9682 \pm 0.0172\)
Fusion #2 ACC% \(97.09\% \pm 1.49\%\) \(94.98\% \pm 1.56\%\) \(96.37\% \pm 1.45\%\) \(95.86\% \pm 2.11\%\) \(96.73\% \pm 1.33\%\) \(95.28 \% \pm 1.54\%\)
AUC \(0.9910 \pm 0.0099\) \(0.9829 \pm 0.0111\) \(0.9908 \pm 0.0092\) \(0.9884 \pm 0.0113\) \(0.9892 \pm 0.0097\) \(0.9872 \pm 0.0106\)
Sen \(0.9660 \pm 0.0205\) \(0.9382 \pm 0.0211\) \(0.9600 \pm 0.0182\) \(0.9527 \pm 0.0271\) \(0.9685 \pm 0.0192\) \(0.9612 \pm 0.0124\)
Prec \(0.9853 \pm 0.0111\) \(0.9778 \pm 0.0213\) \(0.9791 \pm 0.0128\) \(0.9779 \pm 0.0201\) \(0.9771 \pm 0.0158\) \(0.9607 \pm 0.0250\)
Fusion #3 ACC% \(\underline{97} .\underline{46} \% \pm \underline{1} .\underline{24} \%\) \(96.22\% \pm 1.60\%\) - - - -
AUC \(\underline{0} .\underline{9925} \pm \underline{0} .\underline{0093}\) \(0.9920 \pm 0.0047\) - - - -
Sen \(\underline{0} .\underline{9769} \pm \underline{0} .\underline{0156}\) \(0.9551 \pm 0.0229\) - - - -
Prec \(\underline{0} .\underline{9809} \pm \underline{0} .\underline{0165}\) \(0.9814 \pm 0.0103\) - - - -
  1. The results shown in italic are of experiments that are not explored in the original research papers
  2. The underlined classification results represent the highest results for each corresponding section