Table 6 Summary of “Deep learning for phase processing”
Task | Reference | Input | Output | Network | Training dataset | Loss function |
|---|---|---|---|---|---|---|
Segmentation | Yi et al.282 | Phase of red blood cells | Segmentation map | FCN | Expt.: 35 pairs | --- |
Ahmadzadeh et al.284 | Phase of cardiomyocyte | Segmentation map | FCN | Expt.: 2000 pairs | --- | |
Kandel et al.285 | Phase of sperm cells | Segmentation map | U-Net | Expt.: --- | Cross entropy | |
Goswami et al.286 | Phase of virus particles | Segmentation map | U-Net | Expt.: 1000 pairs | Cross entropy | |
Hu et al.287 | Phase of ovary cells | Segmentation map | U-Net and EfficientNet | Expt.: 1536 pairs | Focal loss and dice loss | |
He et al.288 | Phase of HeLa cells | Segmentation map | U-Net and EfficientNet | Expt.: 2046 pairs | focal loss and dice loss | |
Zhang et al.289 | Phase of tissue slices | Segmentation map | mask R-CNN | Expt.: 196 pairs | Cross entropy | |
Jiang et al.290 | Phase and amplitude | Segmentation map | DeepLab-V3+ | Expt.: 1500 pairs | Cross entropy | |
Lee et al.291 | 2D RI tomogram | Segmentation map | U-Net | Expt.: 934 pairs | Cross entropy | |
Choi et al.292 | 3D RI tomogram | Segmentation map | 3D U-Net | Expt.: 105 pairs | Cross entropy and dice loss | |
Classification | Jo et al.293 | Phase of cells | Classification | CNN | Expt.: --- | Cross entropy |
Karandikar et al.314 | Phase of cells | Classification | CNN | Expt.: 300 | Cross entropy | |
Zhang et al.315 | Phase of tissue slices | Classification | VGG | Expt.: 1660 | Cross entropy | |
Butola et al.316 | Phase of sperm cells | Classification | CNN | Expt.: 10,163 | Cross entropy | |
Li et al.317 | Phase of cells | Classification | AlexNet | Expt.: 272 | Cross entropy | |
Shu et al.318 | Phase of cells | Classification | Cascaded ResNet | Expt.: 1521 | Cross entropy | |
Pitkäaho et al.319 | Phase and manual feature | Classification | CNN | Expt.: 2451 | --- | |
O’Connor et al.320 | Transfer-learning and manual feature from phase | Classification | LSTM | Expt.: 303 | --- | |
O’Connor et al.321 | Transfer-learning and manual feature from phase | Classification for COVID-19 | LSTM | Expt.: 1474 | --- | |
Ryu et al.322 | 3D RI tomogram | Classification (2 and 5 types) | 3D CNN | Expt.: 1782 | Cross entropy | |
Kim et al.323 | 3D RI tomogram | Classification (19 types) | 3D CNN | Expt.: 10,556 | Cross entropy | |
Wang et al.324 | Time-lapse amplitude and phase | Classification (3 types) | Pseudo-3D DensNet | Expt.: 16,309 | Cross entropy | |
Liu et al.325 | Time-lapse phase | Classification | Pseudo-3D DensNet | Expt.: 5622 | Cross entropy | |
Ben Baruch et al.326 | Phase and spatio-temporal fluctuation map | Classification | ResNet | Expt.: 216 videos | Cross entropy | |
Singla et al.327 | Phase of three wavelengths | Classification | CNN | Expt.: 16,200 | --- | |
Işıl et al.328 | Phase and amplitude of three wavelengths | Classification | DensNet | Expt.: 33,768 | Cross entropy | |
Pitkäaho et al.329 | Phase and amplitude | Classification | CNN | Expt.: --- | --- | |
Phase and amplitude | Classification | CNN | Sim.: >1000 Expt.: 4000 | --- | ||
Terbe et al.333 | Phase and amplitude in different defocus distances | Classification (7 types) | 3D ResNet | Expt.: >9000 | Cross entropy | |
Wu et al.334 | Real and imaginary | Classification (5 types) | ResNet | Expt.: 7000 | Cross entropy | |
Imaging modal transformation | Wu et al.342 | Real and imaginary | Bright-field image | U-Net | Expt.: 30,000 pairs | GAN loss |
Terbe et al.343 | Amplitude and phase | Bright-field image | U-Net | Expt.: 3000 unpaired | Cycle-GAN loss | |
Rivenson et al.344 | Phase of tissue slices | Stained bright-field image | U-Net | Expt.: >2000 pairs | GAN loss | |
Wang et al.345 | Phase of tissue slices | Stained bright-field or fluorescence image | U-Net | Expt.: 1000 unpaired | Cycle-GAN loss | |
Jiang et al.49 | Amplitude and phase of tissue slices | Stained bright-field or fluorescence image | Y-Net with phase attention | Expt.: ---unpaired | Cycle-GAN loss | |
Liu et al.346 | Amplitude and phase of three wavelength | Stained bright-field image | U-Net | Expt.: 8928 pairs | GAN loss | |
Nygate et al.347 | Phase and gradiences of sperm cells | Stained bright-field image | U-Net | Expt.: 1100 pairs | GAN loss | |
Guo et al.348 | Phase, retardance and Orientation | Fluorescence image | 2.5D U-Net | Expt.: 200 full brain sections | l1-norm | |
Phase | Fluorescence image | U-Net | Expt.: 30–3000 pairs | l2-norm | ||
Guo et al.351 | Phase at different depths | Fluorescence images at different depths | U-Net | Expt.: 200 pairs | l2-norm | |
Chen et al.352 | Three neighboring phase | Corresponding central fluorescence image | U-Net and EfficientNet | Expt.: 20 z-stacks | l2-norm | |
Jo et al.353 | 3D RI tomogram | 3D fluorescence image | 3D U-Net | Expt.: 1600 pairs | l2-norm and gradient difference |
