---

Enhancing Deep Learning Based Classifiers with Inpainting Anatomical Side Markers (L/R Markers) for Multi-center Trials

Requirements:

• Install python3.
• Install tensorflow (tested on Release 1.3.0, 1.4.0, 1.5.0, 1.6.0, 1.7.0).
• Install tensorflow toolkit neuralgym (run pip install git+https://github.com/JiahuiYu/neuralgym).
• Install MI2RLNet v1 L/R mark detection model (https://github.com/mi2rl/MI2RLNet)

Directory Architecture

|---------- inpaint_ops.py (inpainting operator)

|---------- checkpoint (input our pretrained model)

|---------- inpaint_model.py (inpainting model)

|---------- inpaint.yml (hyper parameter)

---

Preprocessing

• minmax scaling (npy format)
• image size : 1024 x 1024 x 1

---

Weight link

https://drive.google.com/drive/folders/17IiClqWW2YHUzPtKmgL4dR6RIKLoYNxK?usp=sharing

---

Inference

CUDA_VISIBLE_DEVICES='gpu_id' python [test.py](<http://test.py/>) \\
--image 'image path' \\
--mask 'detection mask path' \\
--output 'output path' \\
--checkpoint_dir './checkpoint/pretrained model path'

---

Result

---

References

[1] Generative Image Inpainting with Contextual Attention; https://arxiv.org/abs/2111.06377. https://github.com/JiahuiYu/generative_inpainting

[2] An Open Medical Platform to Share Source Code and Various Pre-Trained Weights for Models to Use in Deep Learning Research https://kjronline.org/DOIx.php?id=10.3348/kjr.2021.0170, https://github.com/mi2rl/MI2RLNet

---

Contributing**

If you'd like to have any suggestions for these guidelines, you can contact us at [email protected] or open an issue on this GitHub repository.

All contributions welcome!