search

uci-cbcl / DeepLung

WACV18 paper "DeepLung: Deep 3D Dual Path Nets for Automated Pulmonary Nodule Detection and Classification"
GNU General Public License v3.0
153 stars 54 forks source link

about det2cls.py #7

Closed anruoxi123 closed 5 years ago

anruoxi123 commented 5 years ago

Hello, in det2cls.py file, what is the file resmodelpath = './detcls-'+ STR (fold)+' old/ ckptgbit.t7 '?

I don't show this file here, but when I run the det2cls.py file, it always shows that the file is not found.

Thanks for your advice!

9db5b1cd42119240770f8f13a9fb41b cd1078b92d760c9df0b2c5a7412ca20

wentaozhu commented 5 years ago

How about run the main_nodcls.py file? It seems it is a test script.

anruoxi123 commented 5 years ago

orry, I didn't run the main_nodcls.py file.

That is to say, after I run the detection code, I want to continue running the classification after detection, I want to run the main_nodcls.py file first, not the det2cls.py file, right?

Is the det2cls.py file the test code for the classification section?

anruoxi123 commented 5 years ago

Is the preprocesspath path in the main_nodcls.py file the data path after preprocessing?

I added after preprocessing the data path DeepLung - master/the preprocess subset0 /,

I checked that all the files in this path end with clean. Npy, extendbox.npy, label.npy, mask. Npy, origin. Npy, metono.npy, etc.

But the code looks for files ending with -0.npy,

So I always report the following error when I run the main_nodcls.py file. What's the problem?

anruoxi123 commented 5 years ago

a985c4b812b18312348069d221aa4b6

wentaozhu commented 5 years ago

Please take a look at DeepLung/nodcls/data/ to generate data for classification model.

anruoxi123 commented 5 years ago

Is the classification model after detection using lidc-idri data?

Must the lidc-ldri data be processed with./data/extclsshpinfo.py, humanperformance.py, dimcls.py, nodclsgbit.py, pthumanperformance.py, etc.?

wentaozhu commented 5 years ago

Please check the readme carefully.

For nodule classification, first clean the data from LIDC-IDRI. Use the ./data/extclsshpinfo.py to extract nodule labels. humanperformance.py is used to get the performance of doctors. dimcls.py is used to get the classification based on diameter. nodclsgbt.py is used to get the performance based on GBM, nodule diameter and nodule pixel. pthumanperformance.py is used for patient-level diagnosis performance. kappatest.py is used for kappa value calculation in the paper. For classification using DPN, use the code in main_nodcls.py. Use the testdet2cls.py to test the trained model. You may revise the code a little bit for different test settings. For system's classification, that is classification based on detection. First, use the detection's test script in the run_training.sh to get the detected nodules for training CTs. Use the det2cls.py to train the model. And use the testdet2cls.py to test the trained model. You may revise the code a little bit for different test settings.

anruoxi123 commented 5 years ago

According to my understanding, if it is the classification after running detection, it is necessary to run det2cls.py instead of main_nodcls.py, right?

I would like to ask whether it is necessary to run programs such as./data/extclsshpinfo.py, humanperformance.py, dimcls.py, nodclsgbit.py, pthumanperformance.py before running det2cls.py

wentaozhu commented 5 years ago

Yes