Skip to main content

Table 1 Classification performance on Kather 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% \(\underline{82} .\underline{34} \% \pm \underline{1} .\underline{87} \%\) \(79.42\% \pm 1.27\%\) \(78.46\% \pm 2.11\%\) \(75. 22\% \pm 2.17\%\) \(71.10\% \pm 1.65\%\) \(68.28\% \pm 1.09\%\)
AUC \(\underline{0} .\underline{9658} \pm \underline{0} .\underline{0038}\) \(0.9646 \pm 0.0038\) \(0.9545 \pm 0.0049\) \(0.9552 \pm 0.0066\) \(0.9283 \pm 0.0092\) \(0.9340 \pm 0.0036\)
Sen \(\underline{0} .\underline{8234} \pm \underline{0} .\underline{0186}\) \(0.7943 \pm 0.0127\) \(0.7846 \pm 0.0212\) \(0.7522 \pm 0.0218\) \(0.7111 \pm 0.0161\) \(0.6828 \pm 0.0107\)
Prec \(\underline{0} .\underline{8263} \pm \underline{0} .\underline{0191}\) \(0.7961 \pm 0.0131\) \(0.7858 \pm 0.0218\) \(0.7568 \pm 0.0213\) \(0.7102 \pm 0.0180\) \(0.6826 \pm 0.0131\)
Meshkini and Ghassemian ACC% \(\textit{80.06}\% \pm \textit{1.54}\%\) \(\textit{83.06}\% \pm \textit{1.33}\%\) \(\textit{37.72}\% \pm \textit{1.78}\%\) \(\textit{77.84}\% \pm \textit{1.17}\%\) \(83.38\% \pm 1.81\%\) \(83.74\% \pm 1.79\%\)
AUC \(\textit{0.9648} \pm \textit{0.0055}\) \(\textit{0.9800} \pm \textit{0.0081}\) \(\textit{0.7553} \pm \textit{0.0115}\) \(\textit{0.9649} \pm \textit{0.0028}\) \(0.9738 \pm 0.0036\) \(0.9816 \pm 0.0027\)
Sen \(\textit{0.8006} \pm \textit{0.0154}\) \(\textit{0.8307} \pm \textit{0.0134}\) \(\textit{0.3772} \pm \textit{0.0179}\) \(\textit{0.7784} \pm \textit{0.0117}\) \(0.8338 \pm 0.0181\) \(0.8374 \pm 0.0179\)
Prec \(\textit{0.8111} \pm \textit{0.0128}\) \(\textit{0.8325} \pm \textit{0.0124}\) \(\textit{0.4338} \pm \textit{0.0193}\) \(\textit{0.7792} \pm \textit{0.0105}\) \(0.8385 \pm 0.0170\) \(0.8400 \pm 0.0168\)
Zhou et al. ACC% \(\textit{64.26}\% \pm \textit{2.20}\%\) \(\textit{67.78}\% \pm \textit{1.38}\%\) \(\textit{51.34}\% \pm \textit{2.35}\%\) \(\textit{54.68}\% \pm \textit{1.79}\%\) \(60.06\% \pm 1.43\%\) \(64.42\% \pm 1.82\%\)
AUC \(\textit{0.9016} \pm \textit{0.0074}\) \(\textit{0.9337} \pm \textit{0.0051}\) \(\textit{0.8519} \pm \textit{0.0104}\) \(\textit{0.8846} \pm \textit{0.0045}\) \(0.8764 \pm 0.0050\) \(0.9132 \pm 0.0061\)
Sen \(\textit{0.6425} \pm \textit{0.0222}\) \(\textit{0.6778} \pm \textit{0.0138}\) \(\textit{0.5134} \pm \textit{0.0233}\) \(\textit{0.5468} \pm \textit{0.0177}\) \(0.6006 \pm 0.0143\) \(0.6442 \pm 0.0183\)
Prec \(\textit{0.6494} \pm \textit{0.0240}\) \(\textit{0.6842} \pm \textit{0.0208}\) \(\textit{0.5112} \pm \textit{0.0235}\) \(\textit{0.5540} \pm \textit{0.0210}\) \(0.6182 \pm 0.0198\) \(0.6511 \pm 0.0197\)
Dong et al. ACC% \(\textit{81.10}\% \pm \textit{1.90}\%\) \(\textit{77.50}\% \pm \textit{1.17}\%\) \(\textit{70.66}\% \pm \textit{1.81}\%\) \(\textit{72.10}\% \pm \textit{0.97}\%\) \(76.62\% \pm 1.86\%\) \(72.08\% \pm 1.65\%\)
AUC \(\textit{0.9621} \pm \textit{0.0046}\) \(\textit{0.9620} \pm \textit{0.0033}\) \(\textit{0.9297} \pm \textit{0.0082}\) \(\textit{0.9457} \pm \textit{0.0037}\) \(0.9504 \pm 0.0063\) \(0.9448 \pm 0.0046\)
Sen \(\textit{0.8110} \pm \textit{0.0188}\) \(\textit{0.7751} \pm \textit{0.0116}\) \(\textit{0.7066} \pm \textit{0.0179}\) \(\textit{0.7210} \pm \textit{0.0100}\) \(0.7662 \pm 0.0185\) \(0.7207 \pm 0.0166\)
Prec \(\textit{0.8146} \pm \textit{0.0187}\) \(\textit{0.7753} \pm \textit{0.0140}\) \(\textit{0.7114} \pm \textit{0.0187}\) \(\textit{0.7263} \pm \textit{0.0106}\) \(0.7681 \pm 0.0190\) \(0.7210 \pm 0.0172\)
(2): Proposed Shearlet-based methods for textured bio-medical image classification
CM ACC% \(86.24\% \pm 1.27\%\) \(80.98\% \pm 1.54\%\) \(82.82\% \pm 1.47\%\) \(78.48\% \pm 1.30\%\) \(80.28\% \pm 1.64\%\) \(76.30\% \pm 1.16\%\)
AUC \(0.9761 \pm 0.0027\) \(0.9712 \pm 0.0031\) \(0.9697 \pm 0.0035\) \(0.9636 \pm 0.0046\) \(0.9603 \pm 0.0039\) \(0.9568 \pm 0.0051\)
Sen \(0.8624 \pm 0.0128\) \(0.8098 \pm 0.0155\) \(0.8282 \pm 0.0148\) \(0.7849 \pm 0.01274\) \(0.8028 \pm 0.0166\) \(0.7630 \pm 0.0114\)
Prec \(0.8649 \pm 0.0130\) \(0.8127 \pm 0.0141\) \(0.8322 \pm 0.0147\) \(0.7884 \pm 0.0115\) \(0.8037 \pm 0.0157\) \(0.7642 \pm 0.0127\)
LBP ACC% \(84.24\% \pm 0.91\%\) \(77.90\% \pm 1.75\%\) \(81.36\% \pm 1.16\%\) \(75.10\% \pm 2.20\%\) \(79.36\% \pm 1.39\%\) \(75.38\% \pm 1.71\%\)
AUC \(0.9724 \pm 0.0026\) \(0.9630 \pm 0.0043\) \(0.9649 \pm 0.0039\) \(0.9559 \pm 0.0053\) \(0.9590 \pm 0.0037\) \(0.9541 \pm 0.0064\)
Sen \(0.8424 \pm 0.0092\) \(0.7791 \pm 0.0174\) \(0.8136 \pm 0.0115\) \(0.7511 \pm 0.0216\) \(0.7936 \pm 0.0140\) \(0.7537 \pm 0.0172\)
Prec \(0.8454 \pm 0.0106\) \(0.7829 \pm 0.0185\) \(0.8171 \pm 0.0129\) \(0.7551 \pm 0.0231\) \(0.7964 \pm 0.0131\) \(0.7573 \pm 0.0164\)
LOSIB ACC% \(\underline{86} .\underline{32} \% \pm \underline{1} .\underline{36} \%\) \(79.78\% \pm 2.11\%\) \(82.36\% \pm 1.82\%\) \(79.76\% \pm 1.35\%\) \(76.52\% \pm 1.69\%\) \(71.24\% \pm 1.01\%\)
AUC \(\underline{0} .\underline{9773} \pm \underline{0} .\underline{0021}\) \(0.9703 \pm 0.0046\) \(0.9671 \pm 0.0045\) \(0.9689 \pm 0.0024\) \(0.9495 \pm 0.0048\) \(0.9402 \pm 0.0037\)
Sen \(\underline{0} .\underline{8632} \pm \underline{0} .\underline{0134}\) \(0.7978 \pm 0.0209\) \(0.8236 \pm 0.0180\) \(0.7976 \pm 0.0136\) \(0.7653 \pm 0.0169\) \(0.7124 \pm 0.0098\)
Prec \(\underline{0} .\underline{8664} \pm \underline{0} .\underline{0128}\) \(0.8017 \pm 0.0216\) \(0.8274 \pm 0.0169\) \(0.8008 \pm 0.0127\) \(0.7672 \pm 0.0191\) \(0.7128 \pm 0.0123\)
SFTA ACC% \(82.92\% \pm 1.41\%\) \(78.24\% \pm 1.21\%\) \(78.10\% \pm 1.43\%\) \(75.04\% \pm 1.44\%\) \(79.72\% \pm 1.07\%\) \(75.42\% \pm 2.23\%\)
AUC \(0.9682 \pm 0.0030\) \(0.9645 \pm 0.0035\) \(0.9554 \pm 0.0031\) \(0.9570 \pm 0.0047\) \(0.9597 \pm 0.0032\) \(0.9574 \pm 0.0054\)
Sen \(0.8292 \pm 0.0143\) \(0.7825 \pm 0.0124\) \(0.7810 \pm 0.0141\) \(0.7504 \pm 0.0144\) \(0.7972 \pm 0.0110\) \(0.7543 \pm 0.0224\)
Prec \(0.8331 \pm 0.0139\) \(0.7845 \pm 0.0117\) \(0.7877 \pm 0.0144\) \(0.7539 \pm 0.0133\) \(0.7986 \pm 0.0109\) \(0.7587 \pm 0.0248\)
(3): Integrating Shearlet-based existing techniques with our proposed methods
Fusion #1 ACC% \(87.54\% \pm 1.85\%\) \(81.40\% \pm 1.21\%\) \(85.32\% \pm 1.72\%\) \(80.30\% \pm 1.52\%\) \(83.92\% \pm 1.84\%\) \(77.76\% \pm 1.21\%\)
AUC \(0.9823 \pm 0.0042\) \(0.9734 \pm 0.0019\) \(0.9774 \pm 0.0043\) \(0.9707 \pm 0.0024\) \(0.9740 \pm 0.0049\) \(0.9619 \pm 0.0031\)
Sen \(0.8755 \pm 0.0186\) \(0.8141 \pm 0.0122\) \(0.8533 \pm 0.0173\) \(0.8029 \pm 0.0155\) \(0.8392 \pm 0.0185\) \(0.7776 \pm 0.0121\)
Prec \(0.8764 \pm 0.0189\) \(0.8154 \pm 0.0135\) \(0.8539 \pm 0.0179\) \(0.8078 \pm 0.0153\) \(0.8397 \pm 0.0178\) \(0.7798 \pm 0.0127\)
Fusion #2 ACC% \(88.46\% \pm 1.14\%\) \(82.28\% \pm 1.46\%\) \(85.14\% \pm 1.05\%\) \(80.26\% \pm 2.11\%\) \(83.54\% \pm 1.30\%\) \(77.20\% \pm 1.92\%\)
AUC \(0.9804 \pm 0.0023\) \(0.9746 \pm 0.0034\) \(0.9749 \pm 0.0029\) \(0.9699 \pm 0.0049\) \(0.9693 \pm 0.0033\) \(0.9606 \pm 0.0040\)
Sen \(0.8846 \pm 0.0113\) \(0.8228 \pm 0.0148\) \(0.8514 \pm 0.0105\) \(0.8026 \pm 0.0211\) \(0.8354 \pm 0.0133\) \(0.7720 \pm 0.0195\)
Prec \(0.8874 \pm 0.0108\) \(0.8253 \pm 0.0149\) \(0.8551 \pm 0.0107\) \(0.8070 \pm 0.0198\) \(0.8375 \pm 0.0130\) \(0.7744 \pm 0.0182\)
Fusion #3 ACC% \(\underline{92} .\underline{54} \% \pm \underline{1} .\underline{32} \%\) \(88.60\% \pm 2.03\%\) - - - -
AUC \(\underline{0} .\underline{9906} \pm \underline{0} .\underline{0019}\) \(0.9900 \pm 0.0020\) - - - -
Sen \(\underline{0} .\underline{9254} \pm \underline{0} .\underline{0132}\) \(0.8860 \pm 0.0203\) - - - -
Prec \(\underline{0} .\underline{9267} \pm \underline{0} .\underline{0129}\) \(0.8881 \pm 0.0194\) - - - -
  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