Zongwei Zhou

Hi @hajra90 What do you mean by "test code"? If you want to fine-tune the pre-trained model on your own data, you must build the training script by yourself, such...

Hi @MinaJf Did you refer "augmented data" to the proxy or target tasks? Best, Zongwei

Hi @AgustinaLaGreca Thanks for reaching out. If your interesting range is -50 to 200 HU, I think it is still okay to use Models Genesis because we observe performance gain...

Hi @AgustinaLaGreca Sorry for the delay. No, I don't have a fancy hyperparameter search or optimizer for fine-tuning. Everything is the default in keras/pytorch setup ('Adam'). Every layer is trainable....

Hi @akiliyiu and @Tianyangg I need more details to resolve this issue. Would you please send me an email so that we can arrange a time to look into your...

Hi @yaosh09 I have released the reference code for the target segmentation task, with data preprocessing and visualization. Processing data is not very complicated—the final input is 64x64x32 and the...

Hi @babbu3682 Skip connection itself does not play an important role in the reconstruction task and sometimes gives a trivial solution if the input and output are too similar. Referring...

Additionally, skip connections might be helpful for the purpose of anomaly detection. Referring to Sec. 5.6 in the Models Genesis paper: "Could Models Genesis detect infected regions from images autonomously?"...

Hi @Godthumb Adding an adversarial loss could be a good idea for higher-quality image restoration, but whether better restoration (more realistic) means better representation is unknown. This paper (https://link.springer.com/chapter/10.1007/978-3-030-59719-1_24) used...

Hi @Med-Process Question 1: I recommend scaling the input into [0, 1] because the pre-trained model was trained in such configuration. There are two popular pre-processing strategies: min-max and z-score....