Julian de Wit

I think not.. First of all the model is trained on full luna dataset. I believe LUNA expects 10 fold CV. Also I think there are big differences in the...

There is a catch in the LUNA data. Nodules (masses) bigger than 3cm are not annotated. Since the NDSB date contained very big masses I thought to manually annotate the...

This is a speedup trick. If the convnet is looking far outside the rough lung mask, it knows that it does not need to predict anything. (ie. you will not...

diameter is a wrong variable name. At first I used diameter as a malignancy inidicator. Then I found the malignancy annotations and started to use them.. It was all very...

That is an old discarded feature based on a relative diameter.

I had my internal viewer as discussed in the blog post.

The data is removed from kaggle.. You can test on luna 16 challenge data

Yes you can. Other networks would be C3D (see my article) but this one is trained on video's and does not give good features for CT.

Hello, First of all it was a hasty job :) Indeed I use batchnorm to stabilize the NN computation. Esp with UNet it's sometimes hard to get them going. Basically...

Line 62 is part of cleanup of predictions. At row 282. `p = model.predict(batch_data, batch_size=batch_size)` What do you get in "p" ? That one is important to validate if the...