Attention-Based Deep Learning Approach for Breast Cancer Histopathological Image Multi-Classification

. 2024 Jul 1;14(13):1402. doi: 10.3390/diagnostics14131402

Algorithm 2 Reinhard Color Normalization

Input:
Source image $I_{s}$ in RGB, Target image $I_{t}$ in RGB
Output:
Normalized source image $I_{s}^{'}$ in RGB
1:
Convert $I_{s}$ from RGB to Lab color space to get $L_{s}, a_{s}, b_{s}$
2:
Convert $I_{t}$ from RGB to Lab color space to get $L_{t}, a_{t}, b_{t}$
3:
Compute mean $μ_{L_{s}}$ , $μ_{a_{s}}$ , $μ_{b_{s}}$ and standard deviation $σ_{L_{s}}$ , $σ_{a_{s}}$ , $σ_{b_{s}}$ of $I_{s}$
4:
Compute mean $μ_{L_{t}}$ , $μ_{a_{t}}$ , $μ_{b_{t}}$ and standard deviation $σ_{L_{t}}$ , $σ_{a_{t}}$ , $σ_{b_{t}}$ of $I_{t}$
5:
for each channel $C \in {L, a, b}$ do
6:
Subtract mean from source: $C_{s}^{'} \leftarrow C_{s} - μ_{C_{s}}$
7:
Scale by standard deviation ratio: $C_{s}^{″} \leftarrow C_{s}^{'} \times (σ_{C_{t}} / σ_{C_{s}})$
8:
Add target mean: $C_{s}^{‴} \leftarrow C_{s}^{″} + μ_{C_{t}}$
9:
end for
10:
Merge normalized channels to get Lab image $I_{s}^{'} \leftarrow {L_{s}^{‴}, a_{s}^{‴}, b_{s}^{‴}}$
11:
Convert $I_{s}^{'}$ from Lab to RGB color space