Table 1 Frequently used notations.
Symbol | Definition |
|---|---|
\(T_{div}\) | Threshold for classifying all pixel points during MRI image preprocessing |
\(N_{{g_{val} }}\) | Number of gray scale values in MRI image preprocessing |
\(I_{norm}\) | Gray scale values of MRI images after normalization of color values |
\(a_{ij} i\) | Attention weight |
\(\tilde{\omega }_{ki}\) | The degree of normalization of pixel \(p_{i}\) in the kth object region |
\(f_{k}\) | Representation of the kth object region |
\({\mathcal{K}}\left( {x.f} \right)\) | Unnormalized Relation Function |
\(R_{ik}\) | Relationship of the \(i\)-th pixel to the kth object region |
\(y_{i}\) | Contextual representation of pixel \(i\) |
\(z_{i}^{new\_map}\) | Final representation of pixel \(i\) after feature enhancement |
\(I_{h,w}^{final\_out}\) | The final output of the model |
\(TL\left( {\alpha ,\beta } \right)_{os}\) | Tversky Loss Function with α and β as variables |
\(p_{i}^{os}\) | Predict the result of the \(i\) th pixel point value, take the value of 0 or 1 |
\(g_{i}^{os}\) | The value of the \(i\)-th pixel point of the true result, taking the value 0 or 1 |
\(\mu\) | Value added to avoid division by 0, taking the value 1e-8 |
