Table 3 Optimal feature selected for tumour response classification using SVM-RBF algorithm developed based on ultrasound data (Feature Set IV), ultrasound data + molecular subtype (Feature Set VII), and the best performed feature set (Feature Set VI) with leave-one-out cross validation approach.
No | Only ultrasound data (Feature Set IV) | Ultrasound data + Molecular subtype (Feature Set VII) | Best model feature set (Feature Set VI) |
|---|---|---|---|
1 | Core MBF-ENE-ENE | Luminal-A | Luminal-A |
2 | CMCR-SI | Core AAC-COR | Core SI-COR-HOM |
3 | CMCR-AAC | CMCR-AAC | Margin AAC-CON-COR |
4 | CMCR-ASD | Core SI-COR-HOM | Core MBF-CON-COR |
5 | Core SS-COR-ENE | CMCR-SI | Core SI-COR-ENE |
6 | Margin AAC-COR-HOM' | Core SI-COR | Margin AAC-COR-CON |
7 | CMR-SI | Core MBF-ENE | Margin MBF-CON-COR |
8 | Core MBF-ENE-COR | Core SI-COR-ENE | Core SS-COR-CON |
9 | Margin MBF | Core MBF-CON-COR | Core MBF-ENE-HOM |
10 | Margin MBF-CON-COR | CMCR-ASD | Margin AAC-COR-HOM |
