Upload USGResearch.ipynb
Browse files- USGResearch.ipynb +1376 -0
USGResearch.ipynb
ADDED
|
@@ -0,0 +1,1376 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"nbformat": 4,
|
| 3 |
+
"nbformat_minor": 0,
|
| 4 |
+
"metadata": {
|
| 5 |
+
"colab": {
|
| 6 |
+
"provenance": [],
|
| 7 |
+
"gpuType": "T4"
|
| 8 |
+
},
|
| 9 |
+
"kernelspec": {
|
| 10 |
+
"name": "python3",
|
| 11 |
+
"display_name": "Python 3"
|
| 12 |
+
},
|
| 13 |
+
"language_info": {
|
| 14 |
+
"name": "python"
|
| 15 |
+
},
|
| 16 |
+
"accelerator": "GPU"
|
| 17 |
+
},
|
| 18 |
+
"cells": [
|
| 19 |
+
{
|
| 20 |
+
"cell_type": "markdown",
|
| 21 |
+
"source": [
|
| 22 |
+
"# ML tools for Thyroid Canciroma"
|
| 23 |
+
],
|
| 24 |
+
"metadata": {
|
| 25 |
+
"id": "mBaMWkb6qeFV"
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"cell_type": "code",
|
| 30 |
+
"execution_count": 1,
|
| 31 |
+
"metadata": {
|
| 32 |
+
"id": "M1a84OZrJkBy",
|
| 33 |
+
"colab": {
|
| 34 |
+
"base_uri": "https://localhost:8080/"
|
| 35 |
+
},
|
| 36 |
+
"outputId": "b19e9902-1b1d-4e57-eb23-6df8dcbb8bf2"
|
| 37 |
+
},
|
| 38 |
+
"outputs": [
|
| 39 |
+
{
|
| 40 |
+
"output_type": "stream",
|
| 41 |
+
"name": "stdout",
|
| 42 |
+
"text": [
|
| 43 |
+
"Archive: USG_db.zip\n",
|
| 44 |
+
" creating: USG/Benign/\n",
|
| 45 |
+
" inflating: USG/Benign/tiroides (60).jpg \n",
|
| 46 |
+
" inflating: USG/Benign/tiroides1.jpg \n",
|
| 47 |
+
" inflating: USG/Benign/tiroides10.jpg \n",
|
| 48 |
+
" inflating: USG/Benign/tiroides101.jpg \n",
|
| 49 |
+
" inflating: USG/Benign/tiroides103.jpg \n",
|
| 50 |
+
" inflating: USG/Benign/tiroides122.jpg \n",
|
| 51 |
+
" inflating: USG/Benign/tiroides125.jpg \n",
|
| 52 |
+
" inflating: USG/Benign/tiroides128.jpg \n",
|
| 53 |
+
" inflating: USG/Benign/tiroides13.jpg \n",
|
| 54 |
+
" inflating: USG/Benign/tiroides133.jpg \n",
|
| 55 |
+
" inflating: USG/Benign/tiroides137.jpg \n",
|
| 56 |
+
" inflating: USG/Benign/tiroides138.jpg \n",
|
| 57 |
+
" inflating: USG/Benign/tiroides144.jpg \n",
|
| 58 |
+
" inflating: USG/Benign/tiroides146.jpg \n",
|
| 59 |
+
" inflating: USG/Benign/tiroides151.jpg \n",
|
| 60 |
+
" inflating: USG/Benign/tiroides158.jpg \n",
|
| 61 |
+
" inflating: USG/Benign/tiroides16.jpg \n",
|
| 62 |
+
" inflating: USG/Benign/tiroides161.jpg \n",
|
| 63 |
+
" inflating: USG/Benign/tiroides162.jpg \n",
|
| 64 |
+
" inflating: USG/Benign/tiroides163.jpg \n",
|
| 65 |
+
" inflating: USG/Benign/tiroides164.jpg \n",
|
| 66 |
+
" inflating: USG/Benign/tiroides168.jpg \n",
|
| 67 |
+
" inflating: USG/Benign/tiroides2.jpg \n",
|
| 68 |
+
" inflating: USG/Benign/tiroides20.jpg \n",
|
| 69 |
+
" inflating: USG/Benign/tiroides276.jpg \n",
|
| 70 |
+
" inflating: USG/Benign/tiroides280.jpg \n",
|
| 71 |
+
" inflating: USG/Benign/tiroides281.jpg \n",
|
| 72 |
+
" inflating: USG/Benign/tiroides301.jpg \n",
|
| 73 |
+
" inflating: USG/Benign/tiroides31 (1).jpg \n",
|
| 74 |
+
" inflating: USG/Benign/tiroides31 (103).jpg \n",
|
| 75 |
+
" inflating: USG/Benign/tiroides31 (107).jpg \n",
|
| 76 |
+
" inflating: USG/Benign/tiroides31 (11).jpg \n",
|
| 77 |
+
" inflating: USG/Benign/tiroides31 (110).jpg \n",
|
| 78 |
+
" inflating: USG/Benign/tiroides31 (112).jpg \n",
|
| 79 |
+
" inflating: USG/Benign/tiroides31 (123).jpg \n",
|
| 80 |
+
" inflating: USG/Benign/tiroides31 (124).jpg \n",
|
| 81 |
+
" inflating: USG/Benign/tiroides31 (125).jpg \n",
|
| 82 |
+
" inflating: USG/Benign/tiroides31 (128).jpg \n",
|
| 83 |
+
" inflating: USG/Benign/tiroides31 (129).jpg \n",
|
| 84 |
+
" inflating: USG/Benign/tiroides31 (131).jpg \n",
|
| 85 |
+
" inflating: USG/Benign/tiroides31 (136).jpg \n",
|
| 86 |
+
" inflating: USG/Benign/tiroides31 (2).jpg \n",
|
| 87 |
+
" inflating: USG/Benign/tiroides31 (21).jpg \n",
|
| 88 |
+
" inflating: USG/Benign/tiroides31 (26).jpg \n",
|
| 89 |
+
" inflating: USG/Benign/tiroides31 (33).jpg \n",
|
| 90 |
+
" inflating: USG/Benign/tiroides31 (35).jpg \n",
|
| 91 |
+
" inflating: USG/Benign/tiroides31 (36).jpg \n",
|
| 92 |
+
" inflating: USG/Benign/tiroides31 (40).jpg \n",
|
| 93 |
+
" inflating: USG/Benign/tiroides31 (48).jpg \n",
|
| 94 |
+
" inflating: USG/Benign/tiroides31 (50).jpg \n",
|
| 95 |
+
" inflating: USG/Benign/tiroides31 (55).jpg \n",
|
| 96 |
+
" inflating: USG/Benign/tiroides31 (6).jpg \n",
|
| 97 |
+
" inflating: USG/Benign/tiroides31 (61).jpg \n",
|
| 98 |
+
" inflating: USG/Benign/tiroides31 (62).jpg \n",
|
| 99 |
+
" inflating: USG/Benign/tiroides31 (69).jpg \n",
|
| 100 |
+
" inflating: USG/Benign/tiroides31 (7).jpg \n",
|
| 101 |
+
" inflating: USG/Benign/tiroides31 (71).jpg \n",
|
| 102 |
+
" inflating: USG/Benign/tiroides31 (72).jpg \n",
|
| 103 |
+
" inflating: USG/Benign/tiroides31 (73).jpg \n",
|
| 104 |
+
" inflating: USG/Benign/tiroides31 (74).jpg \n",
|
| 105 |
+
" inflating: USG/Benign/tiroides31 (75).jpg \n",
|
| 106 |
+
" inflating: USG/Benign/tiroides31 (76).jpg \n",
|
| 107 |
+
" inflating: USG/Benign/tiroides31 (86).jpg \n",
|
| 108 |
+
" inflating: USG/Benign/tiroides31 (93).jpg \n",
|
| 109 |
+
" inflating: USG/Benign/tiroides31 (98).jpg \n",
|
| 110 |
+
" inflating: USG/Benign/tiroides32.jpg \n",
|
| 111 |
+
" inflating: USG/Benign/tiroides325.jpg \n",
|
| 112 |
+
" inflating: USG/Benign/tiroides36.jpg \n",
|
| 113 |
+
" inflating: USG/Benign/tiroides37.jpg \n",
|
| 114 |
+
" inflating: USG/Benign/tiroides39.jpg \n",
|
| 115 |
+
" inflating: USG/Benign/tiroides390.jpg \n",
|
| 116 |
+
" inflating: USG/Benign/tiroides4.jpg \n",
|
| 117 |
+
" inflating: USG/Benign/tiroides406.jpg \n",
|
| 118 |
+
" inflating: USG/Benign/tiroides42.jpg \n",
|
| 119 |
+
" inflating: USG/Benign/tiroides46.jpg \n",
|
| 120 |
+
" inflating: USG/Benign/tiroides48.jpg \n",
|
| 121 |
+
" inflating: USG/Benign/tiroides52.jpg \n",
|
| 122 |
+
" inflating: USG/Benign/tiroides57.jpg \n",
|
| 123 |
+
" inflating: USG/Benign/tiroides59.jpg \n",
|
| 124 |
+
" inflating: USG/Benign/tiroides6.jpg \n",
|
| 125 |
+
" inflating: USG/Benign/tiroides61.jpg \n",
|
| 126 |
+
" inflating: USG/Benign/tiroides63.jpg \n",
|
| 127 |
+
" inflating: USG/Benign/tiroides66.jpg \n",
|
| 128 |
+
" inflating: USG/Benign/tiroides67.jpg \n",
|
| 129 |
+
" inflating: USG/Benign/tiroides72.jpg \n",
|
| 130 |
+
" inflating: USG/Benign/tiroides73.jpg \n",
|
| 131 |
+
" inflating: USG/Benign/tiroides78.jpg \n",
|
| 132 |
+
" inflating: USG/Benign/tiroides8.jpg \n",
|
| 133 |
+
" inflating: USG/Benign/tiroides80.jpg \n",
|
| 134 |
+
" inflating: USG/Benign/tiroides96.jpg \n",
|
| 135 |
+
" creating: USG/FTC/\n",
|
| 136 |
+
" inflating: USG/FTC/B_000103587_20130527_US_1_4.png \n",
|
| 137 |
+
" inflating: USG/FTC/B_000875063_20140218_US_1_5.png \n",
|
| 138 |
+
" inflating: USG/FTC/B_000875063_20140218_US_1_7.png \n",
|
| 139 |
+
" inflating: USG/FTC/B_000875063_20140703_US_1_1.png \n",
|
| 140 |
+
" inflating: USG/FTC/B_001903385_20130424_US_1_1.png \n",
|
| 141 |
+
" inflating: USG/FTC/B_001903385_20130424_US_1_10.png \n",
|
| 142 |
+
" inflating: USG/FTC/B_001903385_20130424_US_1_9.png \n",
|
| 143 |
+
" inflating: USG/FTC/B_001907930_20130826_US_1_10.png \n",
|
| 144 |
+
" inflating: USG/FTC/B_001907930_20130826_US_1_8.png \n",
|
| 145 |
+
" inflating: USG/FTC/B_001907930_20130826_US_1_9.png \n",
|
| 146 |
+
" inflating: USG/FTC/B_002086903_20190926_US_1_1.png \n",
|
| 147 |
+
" inflating: USG/FTC/B_002086903_20190926_US_1_2.png \n",
|
| 148 |
+
" inflating: USG/FTC/B_002086903_20190926_US_1_3.png \n",
|
| 149 |
+
" inflating: USG/FTC/B_002086903_20190926_US_1_4.png \n",
|
| 150 |
+
" inflating: USG/FTC/B_002165656_20150120_US_1_1.png \n",
|
| 151 |
+
" inflating: USG/FTC/B_002165656_20150120_US_1_2.png \n",
|
| 152 |
+
" inflating: USG/FTC/B_002165656_20150120_US_1_3.png \n",
|
| 153 |
+
" inflating: USG/FTC/B_002165656_20150120_US_1_6.png \n",
|
| 154 |
+
" inflating: USG/FTC/B_002468117_20170307_US_1_1.png \n",
|
| 155 |
+
" inflating: USG/FTC/B_002542312_20170823_US_1_7.png \n",
|
| 156 |
+
" inflating: USG/FTC/B_002542312_20170823_US_1_8.png \n",
|
| 157 |
+
" inflating: USG/FTC/B_002542312_20170823_US_1_9.png \n",
|
| 158 |
+
" inflating: USG/FTC/B_002648905_20180518_US_1_1.png \n",
|
| 159 |
+
" inflating: USG/FTC/B_002648905_20180518_US_1_3.png \n",
|
| 160 |
+
" inflating: USG/FTC/B_002648905_20180518_US_1_4.png \n",
|
| 161 |
+
" inflating: USG/FTC/B_002648905_20180518_US_1_7.png \n",
|
| 162 |
+
" inflating: USG/FTC/B_100028001_20170105_US_1_3.png \n",
|
| 163 |
+
" inflating: USG/FTC/B_100028001_20170105_US_1_5.png \n",
|
| 164 |
+
" inflating: USG/FTC/B_100028001_20170105_US_1_7.png \n",
|
| 165 |
+
" inflating: USG/FTC/B_100400895_20140423_US_1_6.png \n",
|
| 166 |
+
" inflating: USG/FTC/B_100402163_20130710_US_1_3.png \n",
|
| 167 |
+
" inflating: USG/FTC/B_100402163_20130710_US_1_4.png \n",
|
| 168 |
+
" inflating: USG/FTC/B_100402163_20130710_US_1_5.png \n",
|
| 169 |
+
" inflating: USG/FTC/B_100448043_20130509_US_1_10.png \n",
|
| 170 |
+
" inflating: USG/FTC/B_100448043_20130509_US_1_11.png \n",
|
| 171 |
+
" inflating: USG/FTC/B_100578414_20130311_US_1_2.png \n",
|
| 172 |
+
" inflating: USG/FTC/B_100578414_20130311_US_1_4.png \n",
|
| 173 |
+
" inflating: USG/FTC/B_100578414_20130311_US_1_6.png \n",
|
| 174 |
+
" inflating: USG/FTC/B_100578414_20130311_US_1_7.png \n",
|
| 175 |
+
" inflating: USG/FTC/B_100580629_20130318_US_1_2.png \n",
|
| 176 |
+
" inflating: USG/FTC/B_100580629_20130318_US_1_8.png \n",
|
| 177 |
+
" inflating: USG/FTC/B_100580629_20130318_US_1_9.png \n",
|
| 178 |
+
" inflating: USG/FTC/B_100585439_20130407_US_1_1.png \n",
|
| 179 |
+
" inflating: USG/FTC/B_100585439_20130407_US_1_2.png \n",
|
| 180 |
+
" inflating: USG/FTC/B_100585439_20130407_US_1_3.png \n",
|
| 181 |
+
" inflating: USG/FTC/B_100585439_20130407_US_1_4.png \n",
|
| 182 |
+
" inflating: USG/FTC/B_300032893_20181022_US_1_3.png \n",
|
| 183 |
+
" inflating: USG/FTC/B_300296050_20191225_US_1_2.png \n",
|
| 184 |
+
" inflating: USG/FTC/B_300296050_20191225_US_1_3.png \n",
|
| 185 |
+
" inflating: USG/FTC/B_300296050_20191225_US_1_5.png \n",
|
| 186 |
+
" inflating: USG/FTC/B_300296050_20191225_US_1_6.png \n",
|
| 187 |
+
" inflating: USG/FTC/B_300296050_20191225_US_1_8.png \n",
|
| 188 |
+
" inflating: USG/FTC/B_300296050_20191225_US_1_9.png \n",
|
| 189 |
+
" inflating: USG/FTC/B_300321057_20201215_US_1_3.png \n",
|
| 190 |
+
" inflating: USG/FTC/B_600016858_20180409_US_1_14.png \n",
|
| 191 |
+
" inflating: USG/FTC/B_600016858_20180409_US_1_15.png \n",
|
| 192 |
+
" inflating: USG/FTC/B_600016858_20180409_US_1_16.png \n",
|
| 193 |
+
" inflating: USG/FTC/B_600022518_20140220_US_1_4.png \n",
|
| 194 |
+
" inflating: USG/FTC/B_600027844_20130722_US_1_11.png \n",
|
| 195 |
+
" inflating: USG/FTC/B_600027844_20130722_US_1_4.png \n",
|
| 196 |
+
" inflating: USG/FTC/B_600081125_20140121_US_1_15.png \n",
|
| 197 |
+
" inflating: USG/FTC/B_600087068_20140127_US_1_12.png \n",
|
| 198 |
+
" inflating: USG/FTC/B_600087068_20140127_US_1_5.png \n",
|
| 199 |
+
" inflating: USG/FTC/B_600087068_20140127_US_1_6.png \n",
|
| 200 |
+
" inflating: USG/FTC/B_600128535_20140616_US_1_4.png \n",
|
| 201 |
+
" inflating: USG/FTC/B_600128535_20140616_US_1_5.png \n",
|
| 202 |
+
" inflating: USG/FTC/B_600137379_20140710_US_1_6.png \n",
|
| 203 |
+
" inflating: USG/FTC/B_600137379_20140710_US_1_7.png \n",
|
| 204 |
+
" inflating: USG/FTC/B_600395375_20171012_US_1_1.png \n",
|
| 205 |
+
" inflating: USG/FTC/B_600395375_20171012_US_1_2.png \n",
|
| 206 |
+
" inflating: USG/FTC/B_600395375_20171012_US_1_4.png \n",
|
| 207 |
+
" inflating: USG/FTC/B_600395375_20171012_US_1_5.png \n",
|
| 208 |
+
" inflating: USG/FTC/B_600403531_20151207_US_1_1.png \n",
|
| 209 |
+
" inflating: USG/FTC/B_600403531_20151207_US_1_4.png \n",
|
| 210 |
+
" inflating: USG/FTC/B_600403531_20151207_US_1_5.png \n",
|
| 211 |
+
" inflating: USG/FTC/B_600403531_20151207_US_1_9.png \n",
|
| 212 |
+
" inflating: USG/FTC/B_600405616_20151210_US_0001_0000.png \n",
|
| 213 |
+
" inflating: USG/FTC/B_600614740_20170215_US_0001_1792.png \n",
|
| 214 |
+
" inflating: USG/FTC/B_600651878_20170511_US_1_10.png \n",
|
| 215 |
+
" inflating: USG/FTC/B_600651878_20170511_US_1_11.png \n",
|
| 216 |
+
" inflating: USG/FTC/B_600667771_20170614_US_1_5.png \n",
|
| 217 |
+
" inflating: USG/FTC/B_600750495_20171027_US_1_6.png \n",
|
| 218 |
+
" inflating: USG/FTC/B_600750495_20171027_US_1_8.png \n",
|
| 219 |
+
" inflating: USG/FTC/B_600797960_20180131_US_1_2.png \n",
|
| 220 |
+
" inflating: USG/FTC/B_600797960_20180131_US_1_4.png \n",
|
| 221 |
+
" inflating: USG/FTC/B_600798437_20180201_US_1_10.png \n",
|
| 222 |
+
" inflating: USG/FTC/B_600798437_20180201_US_1_2.png \n",
|
| 223 |
+
" inflating: USG/FTC/B_600798437_20180201_US_1_8.png \n",
|
| 224 |
+
" inflating: USG/FTC/B_600861957_20180625_US_1_5.png \n",
|
| 225 |
+
" inflating: USG/FTC/B_600861957_20180625_US_1_6.png \n",
|
| 226 |
+
" inflating: USG/FTC/B_600861957_20180625_US_1_7.png \n",
|
| 227 |
+
" inflating: USG/FTC/B_600871983_20180712_US_1_10.png \n",
|
| 228 |
+
" inflating: USG/FTC/B_600871983_20180712_US_1_11.png \n",
|
| 229 |
+
" inflating: USG/FTC/B_600871983_20180712_US_1_12.png \n",
|
| 230 |
+
" inflating: USG/FTC/B_600871983_20180712_US_1_13.png \n",
|
| 231 |
+
" inflating: USG/FTC/B_600871983_20180712_US_1_6.png \n",
|
| 232 |
+
" inflating: USG/FTC/B_600871983_20180712_US_1_7.png \n",
|
| 233 |
+
" inflating: USG/FTC/B_600903396_20180827_US_1_11.png \n",
|
| 234 |
+
" inflating: USG/FTC/B_600921536_20180917_US_1_14.png \n",
|
| 235 |
+
" inflating: USG/FTC/B_600921536_20180917_US_1_18.png \n",
|
| 236 |
+
" creating: USG/MTC/\n",
|
| 237 |
+
" inflating: USG/MTC/MTC_0000189482_0.jpg \n",
|
| 238 |
+
" inflating: USG/MTC/MTC_0000189482_1.jpg \n",
|
| 239 |
+
" inflating: USG/MTC/MTC_0000189482_10.jpg \n",
|
| 240 |
+
" inflating: USG/MTC/MTC_0000189482_11.jpg \n",
|
| 241 |
+
" inflating: USG/MTC/MTC_0000189482_12.jpg \n",
|
| 242 |
+
" inflating: USG/MTC/MTC_0000189482_14.jpg \n",
|
| 243 |
+
" inflating: USG/MTC/MTC_0000189482_15.jpg \n",
|
| 244 |
+
" inflating: USG/MTC/MTC_0000189482_16.jpg \n",
|
| 245 |
+
" inflating: USG/MTC/MTC_0000189482_17.jpg \n",
|
| 246 |
+
" inflating: USG/MTC/MTC_0000189482_19.jpg \n",
|
| 247 |
+
" inflating: USG/MTC/MTC_0000189482_2.jpg \n",
|
| 248 |
+
" inflating: USG/MTC/MTC_0000189482_3.jpg \n",
|
| 249 |
+
" inflating: USG/MTC/MTC_0000189482_5.jpg \n",
|
| 250 |
+
" inflating: USG/MTC/MTC_0000189482_6.jpg \n",
|
| 251 |
+
" inflating: USG/MTC/MTC_0000189482_8.jpg \n",
|
| 252 |
+
" inflating: USG/MTC/MTC_0000189482_9.jpg \n",
|
| 253 |
+
" inflating: USG/MTC/MTC_000252531_20181221_US_1_1.jpg \n",
|
| 254 |
+
" inflating: USG/MTC/MTC_000252531_20181221_US_1_17.jpg \n",
|
| 255 |
+
" inflating: USG/MTC/MTC_000615923_20181213_US_1_1.jpg \n",
|
| 256 |
+
" inflating: USG/MTC/MTC_000615923_20181213_US_1_2.jpg \n",
|
| 257 |
+
" inflating: USG/MTC/MTC_000615923_20181213_US_1_5.jpg \n",
|
| 258 |
+
" inflating: USG/MTC/MTC_000615923_20181213_US_1_6.jpg \n",
|
| 259 |
+
" inflating: USG/MTC/MTC_000760566_20151020_US_1_4.jpg \n",
|
| 260 |
+
" inflating: USG/MTC/MTC_000800215_20201022_US_1_10.jpg \n",
|
| 261 |
+
" inflating: USG/MTC/MTC_000800215_20201022_US_1_5.jpg \n",
|
| 262 |
+
" inflating: USG/MTC/MTC_000800215_20201022_US_1_6.jpg \n",
|
| 263 |
+
" inflating: USG/MTC/MTC_000800215_20201022_US_1_7.jpg \n",
|
| 264 |
+
" inflating: USG/MTC/MTC_1.2.840.113663.1500.1.397381976.3.347.20220617.95946.109.jpg \n",
|
| 265 |
+
" inflating: USG/MTC/MTC_1.2.840.113663.1500.1.397381976.3.348.20220617.95952.953.jpg \n",
|
| 266 |
+
" inflating: USG/MTC/MTC_1.2.840.113663.1500.1.397381976.3.349.20220617.100004.406.jpg \n",
|
| 267 |
+
" inflating: USG/MTC/MTC_1.2.840.113663.1500.1.397381976.3.350.20220617.100017.281.jpg \n",
|
| 268 |
+
" inflating: USG/MTC/MTC_1.2.840.113663.1500.1.397381976.3.351.20220617.100029.890.jpg \n",
|
| 269 |
+
" inflating: USG/MTC/MTC_1.2.840.113663.1500.1.397381976.3.352.20220617.100055.484.jpg \n",
|
| 270 |
+
" inflating: USG/MTC/MTC_1.2.840.113663.1500.1.397381976.3.353.20220617.100105.296.jpg \n",
|
| 271 |
+
" inflating: USG/MTC/MTC_1.2.840.113663.1500.1.397381976.3.354.20220617.100116.796.jpg \n",
|
| 272 |
+
" inflating: USG/MTC/MTC_3074ef37dfc3fbc588ce7532eff8bde.jpg \n",
|
| 273 |
+
" inflating: USG/MTC/MTC_5de5e6ec7b7f10e5cc9541710840bf9.jpg \n",
|
| 274 |
+
" inflating: USG/MTC/MTC_636794304377500000.jpg \n",
|
| 275 |
+
" inflating: USG/MTC/MTC_636794304498125000.jpg \n",
|
| 276 |
+
" inflating: USG/MTC/MTC_636794304549218750.jpg \n",
|
| 277 |
+
" inflating: USG/MTC/MTC_636794304588281250.jpg \n",
|
| 278 |
+
" inflating: USG/MTC/MTC_636820165127812500.jpg \n",
|
| 279 |
+
" inflating: USG/MTC/MTC_636820166051718750.jpg \n",
|
| 280 |
+
" inflating: USG/MTC/MTC_636820166231406250.jpg \n",
|
| 281 |
+
" inflating: USG/MTC/MTC_636820166551093750.jpg \n",
|
| 282 |
+
" inflating: USG/MTC/MTC_636820167186562500.jpg \n",
|
| 283 |
+
" inflating: USG/MTC/MTC_636825604143523275.jpg \n",
|
| 284 |
+
" inflating: USG/MTC/MTC_636825604305062515.jpg \n",
|
| 285 |
+
" inflating: USG/MTC/MTC_636825604367246071.jpg \n",
|
| 286 |
+
" inflating: USG/MTC/MTC_636825604825892305.jpg \n",
|
| 287 |
+
" inflating: USG/MTC/MTC_636825605032124100.jpg \n",
|
| 288 |
+
" inflating: USG/MTC/MTC_636825605132589847.jpg \n",
|
| 289 |
+
" inflating: USG/MTC/MTC_636825605244206231.jpg \n",
|
| 290 |
+
" inflating: USG/MTC/MTC_636825605365733182.jpg \n",
|
| 291 |
+
" inflating: USG/MTC/MTC_636825605424086519.jpg \n",
|
| 292 |
+
" inflating: USG/MTC/MTC_636825605537413001.jpg \n",
|
| 293 |
+
" inflating: USG/MTC/MTC_636825605672390721.jpg \n",
|
| 294 |
+
" inflating: USG/MTC/MTC_636825605955446911.jpg \n",
|
| 295 |
+
" inflating: USG/MTC/MTC_636825606241973300.jpg \n",
|
| 296 |
+
" inflating: USG/MTC/MTC_636825606346659287.jpg \n",
|
| 297 |
+
" inflating: USG/MTC/MTC_636825606435674379.jpg \n",
|
| 298 |
+
" inflating: USG/MTC/MTC_636825606512488772.jpg \n",
|
| 299 |
+
" inflating: USG/MTC/MTC_636825606626685304.jpg \n",
|
| 300 |
+
" inflating: USG/MTC/MTC_636825606794864923.jpg \n",
|
| 301 |
+
" inflating: USG/MTC/MTC_636825607097802250.jpg \n",
|
| 302 |
+
" inflating: USG/MTC/MTC_636928246285226522.jpg \n",
|
| 303 |
+
" inflating: USG/MTC/MTC_636928246507689246.jpg \n",
|
| 304 |
+
" inflating: USG/MTC/MTC_636928246564262482.jpg \n",
|
| 305 |
+
" inflating: USG/MTC/MTC_636928246613795315.jpg \n",
|
| 306 |
+
" inflating: USG/MTC/MTC_636928246673328720.jpg \n",
|
| 307 |
+
" inflating: USG/MTC/MTC_636928246700400269.jpg \n",
|
| 308 |
+
" inflating: USG/MTC/MTC_636928246759903672.jpg \n",
|
| 309 |
+
" inflating: USG/MTC/MTC_636928246804896246.jpg \n",
|
| 310 |
+
" inflating: USG/MTC/MTC_636928246841578344.jpg \n",
|
| 311 |
+
" inflating: USG/MTC/MTC_636928246911782359.jpg \n",
|
| 312 |
+
" inflating: USG/MTC/MTC_636928246957784990.jpg \n",
|
| 313 |
+
" inflating: USG/MTC/MTC_636928247087702421.jpg \n",
|
| 314 |
+
" inflating: USG/MTC/MTC_636928247209689399.jpg \n",
|
| 315 |
+
" inflating: USG/MTC/MTC_636928247264552536.jpg \n",
|
| 316 |
+
" inflating: USG/MTC/MTC_636928247318705634.jpg \n",
|
| 317 |
+
" inflating: USG/MTC/MTC_636930060694375000.jpg \n",
|
| 318 |
+
" inflating: USG/MTC/MTC_636930060874531250.jpg \n",
|
| 319 |
+
" inflating: USG/MTC/MTC_636930060924218750.jpg \n",
|
| 320 |
+
" inflating: USG/MTC/MTC_636930061144062500.jpg \n",
|
| 321 |
+
" inflating: USG/MTC/MTC_636930061226406250.jpg \n",
|
| 322 |
+
" inflating: USG/MTC/MTC_636930061287187500.jpg \n",
|
| 323 |
+
" inflating: USG/MTC/MTC_636935260437187500.jpg \n",
|
| 324 |
+
" inflating: USG/MTC/MTC_636935261555312500.jpg \n",
|
| 325 |
+
" inflating: USG/MTC/MTC_636935263982500000.jpg \n",
|
| 326 |
+
" inflating: USG/MTC/MTC_636935264064218750.jpg \n",
|
| 327 |
+
" inflating: USG/MTC/MTC_636947155801178819.jpg \n",
|
| 328 |
+
" inflating: USG/MTC/MTC_636947155997480047.jpg \n",
|
| 329 |
+
" inflating: USG/MTC/MTC_636947156186860879.jpg \n",
|
| 330 |
+
" inflating: USG/MTC/MTC_636947156520189944.jpg \n",
|
| 331 |
+
" inflating: USG/MTC/MTC_636971645042850640.jpg \n",
|
| 332 |
+
" inflating: USG/MTC/MTC_636971645195689382.jpg \n",
|
| 333 |
+
" inflating: USG/MTC/MTC_636971645270463659.jpg \n",
|
| 334 |
+
" inflating: USG/MTC/MTC_636971645464224742.jpg \n",
|
| 335 |
+
" inflating: USG/MTC/MTC_636971645642764954.jpg \n",
|
| 336 |
+
" creating: USG/PTC/\n",
|
| 337 |
+
" inflating: USG/PTC/A_013001-TIRADS4A-03-L-V-2.png \n",
|
| 338 |
+
" inflating: USG/PTC/A_091912-TIRADS4C-01-L-V-2.png \n",
|
| 339 |
+
" inflating: USG/PTC/A_093002-TIRADS4A-03-L-V-2.png \n",
|
| 340 |
+
" inflating: USG/PTC/A_101004-TIRADS4B-01-R-V-2.png \n",
|
| 341 |
+
" inflating: USG/PTC/A_110506-TIRADS4A-01-L-V-2.png \n",
|
| 342 |
+
" inflating: USG/PTC/A_110715-TIRADS4A-01-L-V-2.png \n",
|
| 343 |
+
" inflating: USG/PTC/A_111213-TIRADS4A-01-R-V-2.png \n",
|
| 344 |
+
" inflating: USG/PTC/A_111301-TIRADS4B-01-R-V-2.png \n",
|
| 345 |
+
" inflating: USG/PTC/A_111903-TIRADS4B-01-R-V-2.png \n",
|
| 346 |
+
" inflating: USG/PTC/A_120303-TIRADS5-01-X-V-2.png \n",
|
| 347 |
+
" inflating: USG/PTC/A_120908-TIRADS5-01-L-V-2.png \n",
|
| 348 |
+
" inflating: USG/PTC/A_122204-TIRADS5-01-R-V-2.png \n",
|
| 349 |
+
" inflating: USG/PTC/A_b17170109135813.png \n",
|
| 350 |
+
" inflating: USG/PTC/A_b17170110072746.png \n",
|
| 351 |
+
" inflating: USG/PTC/A_b17170110075828.png \n",
|
| 352 |
+
" inflating: USG/PTC/A_b17170110080622.png \n",
|
| 353 |
+
" inflating: USG/PTC/A_b17170110083358.png \n",
|
| 354 |
+
" inflating: USG/PTC/A_b17170110083444.png \n",
|
| 355 |
+
" inflating: USG/PTC/A_b17170110083527.png \n",
|
| 356 |
+
" inflating: USG/PTC/A_b17170110084327.png \n",
|
| 357 |
+
" inflating: USG/PTC/A_b17170110084425.png \n",
|
| 358 |
+
" inflating: USG/PTC/A_b17170112072149.png \n",
|
| 359 |
+
" inflating: USG/PTC/A_b17170112133310.png \n",
|
| 360 |
+
" inflating: USG/PTC/A_b17170112133625.png \n",
|
| 361 |
+
" inflating: USG/PTC/A_b17170112134637.png \n",
|
| 362 |
+
" inflating: USG/PTC/A_b17170112134701.png \n",
|
| 363 |
+
" inflating: USG/PTC/A_b17170112134749.png \n",
|
| 364 |
+
" inflating: USG/PTC/A_b17170117080854.png \n",
|
| 365 |
+
" inflating: USG/PTC/A_b17170117080912.png \n",
|
| 366 |
+
" inflating: USG/PTC/A_b17170117081007.png \n",
|
| 367 |
+
" inflating: USG/PTC/A_b17170209084245.png \n",
|
| 368 |
+
" inflating: USG/PTC/A_b17170308073111.png \n",
|
| 369 |
+
" inflating: USG/PTC/A_b17170406090806.png \n",
|
| 370 |
+
" inflating: USG/PTC/A_b17170522075032.png \n",
|
| 371 |
+
" inflating: USG/PTC/A_b17170608090324.png \n",
|
| 372 |
+
" inflating: USG/PTC/A_b17170908153837.png \n",
|
| 373 |
+
" inflating: USG/PTC/A_b17171010085802.png \n",
|
| 374 |
+
" inflating: USG/PTC/A_b17171010085928.png \n",
|
| 375 |
+
" inflating: USG/PTC/A_b17171011074315.png \n",
|
| 376 |
+
" inflating: USG/PTC/A_b17171011081947.png \n",
|
| 377 |
+
" inflating: USG/PTC/A_b17171011091207.png \n",
|
| 378 |
+
" inflating: USG/PTC/A_b17171011091244.png \n",
|
| 379 |
+
" inflating: USG/PTC/A_b17171011091533.png \n",
|
| 380 |
+
" inflating: USG/PTC/A_b17171011092647.png \n",
|
| 381 |
+
" inflating: USG/PTC/A_b17171011092717.png \n",
|
| 382 |
+
" inflating: USG/PTC/A_b17171011093018.png \n",
|
| 383 |
+
" inflating: USG/PTC/A_b17171011093834.png \n",
|
| 384 |
+
" inflating: USG/PTC/A_b17171011093956.png \n",
|
| 385 |
+
" inflating: USG/PTC/A_b17171011095957.png \n",
|
| 386 |
+
" inflating: USG/PTC/A_b17171011100259.png \n",
|
| 387 |
+
" inflating: USG/PTC/A_b17171011100319.png \n",
|
| 388 |
+
" inflating: USG/PTC/A_b17171012075630.png \n",
|
| 389 |
+
" inflating: USG/PTC/A_b17171012075850.png \n",
|
| 390 |
+
" inflating: USG/PTC/A_b17171012075919.png \n",
|
| 391 |
+
" inflating: USG/PTC/A_b17171012081602.png \n",
|
| 392 |
+
" inflating: USG/PTC/A_b17171012084011.png \n",
|
| 393 |
+
" inflating: USG/PTC/A_b17171012084427.png \n",
|
| 394 |
+
" inflating: USG/PTC/A_b17171012142230.png \n",
|
| 395 |
+
" inflating: USG/PTC/A_b17171012142256.png \n",
|
| 396 |
+
" inflating: USG/PTC/A_b17171012144234.png \n",
|
| 397 |
+
" inflating: USG/PTC/A_b17171012144421.png \n",
|
| 398 |
+
" inflating: USG/PTC/A_b17171016073541.png \n",
|
| 399 |
+
" inflating: USG/PTC/A_b17171016073735.png \n",
|
| 400 |
+
" inflating: USG/PTC/A_b17171016074619.png \n",
|
| 401 |
+
" inflating: USG/PTC/A_b17171016100654.png \n",
|
| 402 |
+
" inflating: USG/PTC/A_b17171017073834.png \n",
|
| 403 |
+
" inflating: USG/PTC/A_b17171017073909.png \n",
|
| 404 |
+
" inflating: USG/PTC/A_b17171017073959.png \n",
|
| 405 |
+
" inflating: USG/PTC/A_b17171017075151.png \n",
|
| 406 |
+
" inflating: USG/PTC/A_b17171017081042.png \n",
|
| 407 |
+
" inflating: USG/PTC/A_b17171017081122.png \n",
|
| 408 |
+
" inflating: USG/PTC/A_b17171017085025.png \n",
|
| 409 |
+
" inflating: USG/PTC/A_b17171017085114.png \n",
|
| 410 |
+
" inflating: USG/PTC/A_b17171017085139.png \n",
|
| 411 |
+
" inflating: USG/PTC/A_b17171017091203.png \n",
|
| 412 |
+
" inflating: USG/PTC/A_b17171017091250.png \n",
|
| 413 |
+
" inflating: USG/PTC/A_b17171017091328.png \n",
|
| 414 |
+
" inflating: USG/PTC/A_b17171018073349.png \n",
|
| 415 |
+
" inflating: USG/PTC/A_b17171018074030.png \n",
|
| 416 |
+
" inflating: USG/PTC/A_b17171018074055.png \n",
|
| 417 |
+
" inflating: USG/PTC/A_b17171018074117.png \n",
|
| 418 |
+
" inflating: USG/PTC/A_b17171018103941.png \n",
|
| 419 |
+
" inflating: USG/PTC/A_b17171018155334.png \n",
|
| 420 |
+
" inflating: USG/PTC/A_b17171018155546.png \n",
|
| 421 |
+
" inflating: USG/PTC/A_b17171019074311.png \n",
|
| 422 |
+
" inflating: USG/PTC/A_b17171019075456.png \n",
|
| 423 |
+
" inflating: USG/PTC/A_b17171019075543.png \n",
|
| 424 |
+
" inflating: USG/PTC/A_b17171019075801.png \n",
|
| 425 |
+
" inflating: USG/PTC/A_b17171019092726.png \n",
|
| 426 |
+
" inflating: USG/PTC/A_b17171019092848.png \n",
|
| 427 |
+
" inflating: USG/PTC/A_b17171021112137.png \n",
|
| 428 |
+
" inflating: USG/PTC/A_b17171021112331.png \n",
|
| 429 |
+
" inflating: USG/PTC/A_b17171023083335.png \n",
|
| 430 |
+
" inflating: USG/PTC/A_b17171023083403.png \n",
|
| 431 |
+
" inflating: USG/PTC/A_b17171023085148.png \n",
|
| 432 |
+
" inflating: USG/PTC/A_b17171023085247.png \n",
|
| 433 |
+
" inflating: USG/PTC/A_b17171024083510.png \n",
|
| 434 |
+
" inflating: USG/PTC/A_b17171024083633.png \n",
|
| 435 |
+
" inflating: USG/PTC/A_b17171026073152.png \n"
|
| 436 |
+
]
|
| 437 |
+
}
|
| 438 |
+
],
|
| 439 |
+
"source": [
|
| 440 |
+
"!unzip USG_db.zip"
|
| 441 |
+
]
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"cell_type": "markdown",
|
| 445 |
+
"source": [
|
| 446 |
+
"# Data Preprocessing\n",
|
| 447 |
+
"\n",
|
| 448 |
+
"This dataset and research is all about thyroid cancer detection. We have gathered 3(three) types of ultrasonography image .\n",
|
| 449 |
+
"\n",
|
| 450 |
+
"1. **FTC (Follicular Thyroid Carcinoma) – 100 images**\n",
|
| 451 |
+
"\n",
|
| 452 |
+
"2. **PTC (Papillary Thyroid Carcinoma) – 99 images**\n",
|
| 453 |
+
"\n",
|
| 454 |
+
"3. **MTC (Medullary Thyroid Carcinoma) – 99 images**\n",
|
| 455 |
+
"\n",
|
| 456 |
+
"4. **Benign (Normal Thyroid) - 90 images**"
|
| 457 |
+
],
|
| 458 |
+
"metadata": {
|
| 459 |
+
"id": "AoeGM3-NKDwS"
|
| 460 |
+
}
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"cell_type": "code",
|
| 464 |
+
"source": [
|
| 465 |
+
"import os\n",
|
| 466 |
+
"import random\n",
|
| 467 |
+
"from PIL import Image, ImageEnhance, ImageOps\n",
|
| 468 |
+
"import shutil\n",
|
| 469 |
+
"\n",
|
| 470 |
+
"\n",
|
| 471 |
+
"base_dir = \"USG\"\n",
|
| 472 |
+
"output_dir = \"thyroid_preprocessed\"\n",
|
| 473 |
+
"os.makedirs(output_dir, exist_ok=True)\n",
|
| 474 |
+
"\n",
|
| 475 |
+
"classes = [\"FTC\", \"PTC\", \"MTC\",\"Benign\"]\n",
|
| 476 |
+
"all_images = []\n",
|
| 477 |
+
"\n",
|
| 478 |
+
"for cls in classes:\n",
|
| 479 |
+
" img_dir = os.path.join(base_dir, cls)\n",
|
| 480 |
+
" imgs = [os.path.join(img_dir, f) for f in os.listdir(img_dir) if f.endswith((\".png\", \".jpg\", \".jpeg\"))]\n",
|
| 481 |
+
" all_images.extend([(img, cls) for img in imgs])\n",
|
| 482 |
+
"\n",
|
| 483 |
+
"random.shuffle(all_images)\n",
|
| 484 |
+
"\n",
|
| 485 |
+
"n = len(all_images)\n",
|
| 486 |
+
"train_split, val_split = int(0.7 * n), int(0.9 * n)\n",
|
| 487 |
+
"\n",
|
| 488 |
+
"train_data = all_images[:train_split]\n",
|
| 489 |
+
"val_data = all_images[train_split:val_split]\n",
|
| 490 |
+
"test_data = all_images[val_split:]\n",
|
| 491 |
+
"\n",
|
| 492 |
+
"def augment_image(img):\n",
|
| 493 |
+
" # Random brightness, contrast, and flips\n",
|
| 494 |
+
" if random.random() < 0.5:\n",
|
| 495 |
+
" img = ImageEnhance.Brightness(img).enhance(random.uniform(0.7, 1.3))\n",
|
| 496 |
+
" if random.random() < 0.5:\n",
|
| 497 |
+
" img = ImageEnhance.Contrast(img).enhance(random.uniform(0.7, 1.3))\n",
|
| 498 |
+
" if random.random() < 0.5:\n",
|
| 499 |
+
" img = ImageOps.mirror(img)\n",
|
| 500 |
+
" if random.random() < 0.5:\n",
|
| 501 |
+
" img = img.rotate(random.choice([90, 180, 270]))\n",
|
| 502 |
+
" return img\n",
|
| 503 |
+
"\n",
|
| 504 |
+
"# Save helper\n",
|
| 505 |
+
"def save_image(img, dest_dir, cls, fname):\n",
|
| 506 |
+
" os.makedirs(os.path.join(dest_dir, cls), exist_ok=True)\n",
|
| 507 |
+
" img.save(os.path.join(dest_dir, cls, fname))\n",
|
| 508 |
+
"\n",
|
| 509 |
+
"# Copy and augment\n",
|
| 510 |
+
"def process_dataset(data, dest_dir, apply_aug=False):\n",
|
| 511 |
+
" for img_path, cls in data:\n",
|
| 512 |
+
" img = Image.open(img_path).convert(\"RGB\")\n",
|
| 513 |
+
" fname = os.path.basename(img_path)\n",
|
| 514 |
+
" save_image(img, dest_dir, cls, fname)\n",
|
| 515 |
+
"\n",
|
| 516 |
+
" if apply_aug:\n",
|
| 517 |
+
" num_aug = int(0.3 * len(data)) # augment 30%\n",
|
| 518 |
+
" aug_samples = random.sample(data, num_aug)\n",
|
| 519 |
+
" for img_path, cls in aug_samples:\n",
|
| 520 |
+
" img = Image.open(img_path).convert(\"RGB\")\n",
|
| 521 |
+
" img = augment_image(img)\n",
|
| 522 |
+
" fname = \"aug_\" + os.path.basename(img_path)\n",
|
| 523 |
+
" save_image(img, dest_dir, cls, fname)\n",
|
| 524 |
+
"\n",
|
| 525 |
+
"process_dataset(train_data, os.path.join(output_dir, \"train\"), apply_aug=True)\n",
|
| 526 |
+
"process_dataset(val_data, os.path.join(output_dir, \"val\"), apply_aug=False)\n",
|
| 527 |
+
"process_dataset(test_data, os.path.join(output_dir, \"test\"), apply_aug=True)\n",
|
| 528 |
+
"\n",
|
| 529 |
+
"print(f\"Total images: {n}\")\n",
|
| 530 |
+
"print(f\"Train: {len(train_data)}, Val: {len(val_data)}, Test: {len(test_data)}\")\n",
|
| 531 |
+
"print(f\"Augmented (train): {int(0.3 * len(train_data))}, (test): {int(0.3 * len(test_data))}\")\n"
|
| 532 |
+
],
|
| 533 |
+
"metadata": {
|
| 534 |
+
"colab": {
|
| 535 |
+
"base_uri": "https://localhost:8080/"
|
| 536 |
+
},
|
| 537 |
+
"id": "mFoil1duKt90",
|
| 538 |
+
"outputId": "85cb36bc-42b7-47ae-e74f-d124001bca55"
|
| 539 |
+
},
|
| 540 |
+
"execution_count": 2,
|
| 541 |
+
"outputs": [
|
| 542 |
+
{
|
| 543 |
+
"output_type": "stream",
|
| 544 |
+
"name": "stdout",
|
| 545 |
+
"text": [
|
| 546 |
+
"Total images: 388\n",
|
| 547 |
+
"Train: 271, Val: 78, Test: 39\n",
|
| 548 |
+
"Augmented (train): 81, (test): 11\n"
|
| 549 |
+
]
|
| 550 |
+
}
|
| 551 |
+
]
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"cell_type": "code",
|
| 555 |
+
"source": [
|
| 556 |
+
"from torchvision import datasets, transforms\n",
|
| 557 |
+
"from torch.utils.data import DataLoader\n",
|
| 558 |
+
"\n",
|
| 559 |
+
"normalize = transforms.Compose([\n",
|
| 560 |
+
" transforms.Resize((224,224)),\n",
|
| 561 |
+
" transforms.ToTensor(),\n",
|
| 562 |
+
" transforms.Normalize([0.485,0.456,0.406],[0.229,0.224,0.225])\n",
|
| 563 |
+
"])\n",
|
| 564 |
+
"\n",
|
| 565 |
+
"train_dataset = datasets.ImageFolder(\"thyroid_preprocessed/train\", transform=normalize)\n",
|
| 566 |
+
"val_dataset = datasets.ImageFolder(\"thyroid_preprocessed/val\", transform=normalize)\n",
|
| 567 |
+
"test_dataset = datasets.ImageFolder(\"thyroid_preprocessed/test\", transform=normalize)\n",
|
| 568 |
+
"\n",
|
| 569 |
+
"train_loader = DataLoader(train_dataset, batch_size=16, shuffle=True, num_workers=2)\n",
|
| 570 |
+
"val_loader = DataLoader(val_dataset, batch_size=16, shuffle=False, num_workers=2)\n",
|
| 571 |
+
"test_loader = DataLoader(test_dataset, batch_size=16, shuffle=False, num_workers=2)\n"
|
| 572 |
+
],
|
| 573 |
+
"metadata": {
|
| 574 |
+
"id": "yoeePErHLOdN"
|
| 575 |
+
},
|
| 576 |
+
"execution_count": 1,
|
| 577 |
+
"outputs": []
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"cell_type": "code",
|
| 581 |
+
"source": [
|
| 582 |
+
"import torch.nn as nn\n",
|
| 583 |
+
"from torch.utils.checkpoint import checkpoint_sequential\n",
|
| 584 |
+
"\n",
|
| 585 |
+
"chunks = 2 # how many chunks for checkpointing\n",
|
| 586 |
+
"\n",
|
| 587 |
+
"class CheckpointedAlexNet(nn.Module):\n",
|
| 588 |
+
" def __init__(self, num_classes=4):\n",
|
| 589 |
+
" super().__init__()\n",
|
| 590 |
+
" self.features = nn.Sequential(\n",
|
| 591 |
+
" nn.Conv2d(3, 64, 11, stride=4, padding=2),\n",
|
| 592 |
+
" nn.ReLU(inplace=True),\n",
|
| 593 |
+
" nn.MaxPool2d(3, stride=2),\n",
|
| 594 |
+
" nn.Conv2d(64,192,5,padding=2),\n",
|
| 595 |
+
" nn.ReLU(inplace=True),\n",
|
| 596 |
+
" nn.MaxPool2d(3,stride=2),\n",
|
| 597 |
+
" nn.Conv2d(192,384,3,padding=1),\n",
|
| 598 |
+
" nn.ReLU(inplace=True),\n",
|
| 599 |
+
" nn.Conv2d(384,256,3,padding=1),\n",
|
| 600 |
+
" nn.ReLU(inplace=True),\n",
|
| 601 |
+
" nn.Conv2d(256,256,3,padding=1),\n",
|
| 602 |
+
" nn.ReLU(inplace=True),\n",
|
| 603 |
+
" nn.MaxPool2d(3,stride=2)\n",
|
| 604 |
+
" )\n",
|
| 605 |
+
" self.avgpool = nn.AdaptiveAvgPool2d((6,6))\n",
|
| 606 |
+
" self.classifier = nn.Sequential(\n",
|
| 607 |
+
" nn.Dropout(),\n",
|
| 608 |
+
" nn.Linear(256*6*6,4096),\n",
|
| 609 |
+
" nn.ReLU(inplace=True),\n",
|
| 610 |
+
" nn.Dropout(),\n",
|
| 611 |
+
" nn.Linear(4096,4096),\n",
|
| 612 |
+
" nn.ReLU(inplace=True),\n",
|
| 613 |
+
" nn.Linear(4096,num_classes)\n",
|
| 614 |
+
" )\n",
|
| 615 |
+
"\n",
|
| 616 |
+
" def forward(self,x):\n",
|
| 617 |
+
" x = checkpoint_sequential(self.features, chunks, x, use_reentrant=False)\n",
|
| 618 |
+
" x = self.avgpool(x)\n",
|
| 619 |
+
" x = x.view(x.size(0), 256*6*6)\n",
|
| 620 |
+
" x = self.classifier(x)\n",
|
| 621 |
+
" return x\n"
|
| 622 |
+
],
|
| 623 |
+
"metadata": {
|
| 624 |
+
"id": "rFz09FESNQYs"
|
| 625 |
+
},
|
| 626 |
+
"execution_count": 2,
|
| 627 |
+
"outputs": []
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"cell_type": "code",
|
| 631 |
+
"source": [
|
| 632 |
+
"import torch\n",
|
| 633 |
+
"\n",
|
| 634 |
+
"device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
|
| 635 |
+
"model = CheckpointedAlexNet(num_classes=4).to(device)\n",
|
| 636 |
+
"\n",
|
| 637 |
+
"criterion = nn.CrossEntropyLoss()\n",
|
| 638 |
+
"optimizer = torch.optim.Adam(model.parameters(), lr=1e-4)\n"
|
| 639 |
+
],
|
| 640 |
+
"metadata": {
|
| 641 |
+
"id": "9QODqeToNTgo"
|
| 642 |
+
},
|
| 643 |
+
"execution_count": 3,
|
| 644 |
+
"outputs": []
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"cell_type": "code",
|
| 648 |
+
"source": [
|
| 649 |
+
"num_epochs = 20\n",
|
| 650 |
+
"\n",
|
| 651 |
+
"for epoch in range(num_epochs):\n",
|
| 652 |
+
" model.train()\n",
|
| 653 |
+
" running_loss = 0.0\n",
|
| 654 |
+
" for imgs, labels in train_loader:\n",
|
| 655 |
+
" imgs, labels = imgs.to(device), labels.to(device)\n",
|
| 656 |
+
" optimizer.zero_grad()\n",
|
| 657 |
+
" outputs = model(imgs)\n",
|
| 658 |
+
" loss = criterion(outputs, labels)\n",
|
| 659 |
+
" loss.backward()\n",
|
| 660 |
+
" optimizer.step()\n",
|
| 661 |
+
" running_loss += loss.item() * imgs.size(0)\n",
|
| 662 |
+
"\n",
|
| 663 |
+
" epoch_loss = running_loss / len(train_loader.dataset)\n",
|
| 664 |
+
" print(f\"Epoch {epoch+1}/{num_epochs}, Loss: {epoch_loss:.4f}\")\n",
|
| 665 |
+
"\n",
|
| 666 |
+
" # optionally, validation accuracy\n",
|
| 667 |
+
" model.eval()\n",
|
| 668 |
+
" correct = 0\n",
|
| 669 |
+
" total = 0\n",
|
| 670 |
+
" with torch.no_grad():\n",
|
| 671 |
+
" for imgs, labels in val_loader:\n",
|
| 672 |
+
" imgs, labels = imgs.to(device), labels.to(device)\n",
|
| 673 |
+
" outputs = model(imgs)\n",
|
| 674 |
+
" _, preds = torch.max(outputs,1)\n",
|
| 675 |
+
" correct += (preds==labels).sum().item()\n",
|
| 676 |
+
" total += labels.size(0)\n",
|
| 677 |
+
" val_acc = correct/total\n",
|
| 678 |
+
" print(f\"Validation Accuracy: {val_acc:.4f}\")\n",
|
| 679 |
+
" torch.save(model.state_dict(), f\"model_{epoch}.pth\")\n"
|
| 680 |
+
],
|
| 681 |
+
"metadata": {
|
| 682 |
+
"colab": {
|
| 683 |
+
"base_uri": "https://localhost:8080/"
|
| 684 |
+
},
|
| 685 |
+
"id": "snr2aZmqNWFc",
|
| 686 |
+
"outputId": "2e82853d-c35d-42e7-b16e-09498ad144a8"
|
| 687 |
+
},
|
| 688 |
+
"execution_count": 4,
|
| 689 |
+
"outputs": [
|
| 690 |
+
{
|
| 691 |
+
"output_type": "stream",
|
| 692 |
+
"name": "stdout",
|
| 693 |
+
"text": [
|
| 694 |
+
"Epoch 1/20, Loss: 1.3756\n",
|
| 695 |
+
"Validation Accuracy: 0.3590\n",
|
| 696 |
+
"Epoch 2/20, Loss: 1.1746\n",
|
| 697 |
+
"Validation Accuracy: 0.6026\n",
|
| 698 |
+
"Epoch 3/20, Loss: 0.9701\n",
|
| 699 |
+
"Validation Accuracy: 0.6538\n",
|
| 700 |
+
"Epoch 4/20, Loss: 0.8224\n",
|
| 701 |
+
"Validation Accuracy: 0.6923\n",
|
| 702 |
+
"Epoch 5/20, Loss: 0.6945\n",
|
| 703 |
+
"Validation Accuracy: 0.6026\n",
|
| 704 |
+
"Epoch 6/20, Loss: 0.5991\n",
|
| 705 |
+
"Validation Accuracy: 0.7436\n",
|
| 706 |
+
"Epoch 7/20, Loss: 0.5588\n",
|
| 707 |
+
"Validation Accuracy: 0.7949\n",
|
| 708 |
+
"Epoch 8/20, Loss: 0.4317\n",
|
| 709 |
+
"Validation Accuracy: 0.8462\n",
|
| 710 |
+
"Epoch 9/20, Loss: 0.3715\n",
|
| 711 |
+
"Validation Accuracy: 0.8590\n",
|
| 712 |
+
"Epoch 10/20, Loss: 0.4791\n",
|
| 713 |
+
"Validation Accuracy: 0.8333\n",
|
| 714 |
+
"Epoch 11/20, Loss: 0.3604\n",
|
| 715 |
+
"Validation Accuracy: 0.8333\n",
|
| 716 |
+
"Epoch 12/20, Loss: 0.2914\n",
|
| 717 |
+
"Validation Accuracy: 0.8718\n",
|
| 718 |
+
"Epoch 13/20, Loss: 0.2471\n",
|
| 719 |
+
"Validation Accuracy: 0.8077\n",
|
| 720 |
+
"Epoch 14/20, Loss: 0.2251\n",
|
| 721 |
+
"Validation Accuracy: 0.8718\n",
|
| 722 |
+
"Epoch 15/20, Loss: 0.1646\n",
|
| 723 |
+
"Validation Accuracy: 0.7821\n",
|
| 724 |
+
"Epoch 16/20, Loss: 0.1744\n",
|
| 725 |
+
"Validation Accuracy: 0.8590\n",
|
| 726 |
+
"Epoch 17/20, Loss: 0.1534\n",
|
| 727 |
+
"Validation Accuracy: 0.9103\n",
|
| 728 |
+
"Epoch 18/20, Loss: 0.1267\n",
|
| 729 |
+
"Validation Accuracy: 0.8590\n",
|
| 730 |
+
"Epoch 19/20, Loss: 0.1756\n",
|
| 731 |
+
"Validation Accuracy: 0.8077\n",
|
| 732 |
+
"Epoch 20/20, Loss: 0.0903\n",
|
| 733 |
+
"Validation Accuracy: 0.8590\n"
|
| 734 |
+
]
|
| 735 |
+
}
|
| 736 |
+
]
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"cell_type": "code",
|
| 740 |
+
"source": [
|
| 741 |
+
"import torch\n",
|
| 742 |
+
"import os\n",
|
| 743 |
+
"from sklearn.metrics import classification_report, confusion_matrix\n",
|
| 744 |
+
"\n",
|
| 745 |
+
"def evaluate_all_models(model_class, test_loader, classes, model_dir=\".\", device=\"cuda\" if torch.cuda.is_available() else \"cpu\"):\n",
|
| 746 |
+
" model_files = sorted([f for f in os.listdir(model_dir) if f.startswith(\"model_\") and f.endswith(\".pth\")])\n",
|
| 747 |
+
" if not model_files:\n",
|
| 748 |
+
" print(\"No model_*.pth files found.\")\n",
|
| 749 |
+
" return\n",
|
| 750 |
+
"\n",
|
| 751 |
+
" for model_file in model_files:\n",
|
| 752 |
+
" path = os.path.join(model_dir, model_file)\n",
|
| 753 |
+
" print(f\"\\nEvaluating {model_file} ...\")\n",
|
| 754 |
+
"\n",
|
| 755 |
+
" model = model_class(num_classes=len(classes)).to(device)\n",
|
| 756 |
+
" model.load_state_dict(torch.load(path, map_location=device))\n",
|
| 757 |
+
" model.eval()\n",
|
| 758 |
+
"\n",
|
| 759 |
+
" all_preds, all_labels = [], []\n",
|
| 760 |
+
" with torch.no_grad():\n",
|
| 761 |
+
" for imgs, labels in test_loader:\n",
|
| 762 |
+
" imgs, labels = imgs.to(device), labels.to(device)\n",
|
| 763 |
+
" outputs = model(imgs)\n",
|
| 764 |
+
" _, preds = torch.max(outputs, 1)\n",
|
| 765 |
+
" all_preds.extend(preds.cpu().numpy())\n",
|
| 766 |
+
" all_labels.extend(labels.cpu().numpy())\n",
|
| 767 |
+
"\n",
|
| 768 |
+
" print(\"\\n=== Classification Report ===\")\n",
|
| 769 |
+
" print(classification_report(all_labels, all_preds, target_names=classes))\n",
|
| 770 |
+
" print(\"\\n=== Confusion Matrix ===\")\n",
|
| 771 |
+
" print(confusion_matrix(all_labels, all_preds))\n",
|
| 772 |
+
" print(\"=============================================\")\n"
|
| 773 |
+
],
|
| 774 |
+
"metadata": {
|
| 775 |
+
"id": "hjamtIpkN2bS"
|
| 776 |
+
},
|
| 777 |
+
"execution_count": 5,
|
| 778 |
+
"outputs": []
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"cell_type": "code",
|
| 782 |
+
"source": [
|
| 783 |
+
"classes = [\"FTC\", \"PTC\", \"MTC\",\"Benign\"]\n",
|
| 784 |
+
"evaluate_all_models(CheckpointedAlexNet, test_loader, classes, model_dir=\".\", device=device)\n"
|
| 785 |
+
],
|
| 786 |
+
"metadata": {
|
| 787 |
+
"colab": {
|
| 788 |
+
"base_uri": "https://localhost:8080/"
|
| 789 |
+
},
|
| 790 |
+
"id": "53tkura8voG_",
|
| 791 |
+
"outputId": "9ef59975-8dd0-4d9a-be33-46437110b631"
|
| 792 |
+
},
|
| 793 |
+
"execution_count": 6,
|
| 794 |
+
"outputs": [
|
| 795 |
+
{
|
| 796 |
+
"output_type": "stream",
|
| 797 |
+
"name": "stdout",
|
| 798 |
+
"text": [
|
| 799 |
+
"\n",
|
| 800 |
+
"Evaluating model_0.pth ...\n",
|
| 801 |
+
"\n",
|
| 802 |
+
"=== Classification Report ===\n",
|
| 803 |
+
" precision recall f1-score support\n",
|
| 804 |
+
"\n",
|
| 805 |
+
" FTC 0.86 0.40 0.55 15\n",
|
| 806 |
+
" PTC 0.00 0.00 0.00 10\n",
|
| 807 |
+
" MTC 0.17 0.10 0.12 10\n",
|
| 808 |
+
" Benign 0.41 1.00 0.58 15\n",
|
| 809 |
+
"\n",
|
| 810 |
+
" accuracy 0.44 50\n",
|
| 811 |
+
" macro avg 0.36 0.38 0.31 50\n",
|
| 812 |
+
"weighted avg 0.41 0.44 0.36 50\n",
|
| 813 |
+
"\n",
|
| 814 |
+
"\n",
|
| 815 |
+
"=== Confusion Matrix ===\n",
|
| 816 |
+
"[[ 6 0 5 4]\n",
|
| 817 |
+
" [ 1 0 0 9]\n",
|
| 818 |
+
" [ 0 0 1 9]\n",
|
| 819 |
+
" [ 0 0 0 15]]\n",
|
| 820 |
+
"=============================================\n",
|
| 821 |
+
"\n",
|
| 822 |
+
"Evaluating model_1.pth ...\n"
|
| 823 |
+
]
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"output_type": "stream",
|
| 827 |
+
"name": "stderr",
|
| 828 |
+
"text": [
|
| 829 |
+
"/usr/local/lib/python3.12/dist-packages/sklearn/metrics/_classification.py:1565: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
|
| 830 |
+
" _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
|
| 831 |
+
"/usr/local/lib/python3.12/dist-packages/sklearn/metrics/_classification.py:1565: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
|
| 832 |
+
" _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n",
|
| 833 |
+
"/usr/local/lib/python3.12/dist-packages/sklearn/metrics/_classification.py:1565: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
|
| 834 |
+
" _warn_prf(average, modifier, f\"{metric.capitalize()} is\", len(result))\n"
|
| 835 |
+
]
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"output_type": "stream",
|
| 839 |
+
"name": "stdout",
|
| 840 |
+
"text": [
|
| 841 |
+
"\n",
|
| 842 |
+
"=== Classification Report ===\n",
|
| 843 |
+
" precision recall f1-score support\n",
|
| 844 |
+
"\n",
|
| 845 |
+
" FTC 0.82 0.60 0.69 15\n",
|
| 846 |
+
" PTC 0.46 0.60 0.52 10\n",
|
| 847 |
+
" MTC 0.29 0.20 0.24 10\n",
|
| 848 |
+
" Benign 0.68 0.87 0.76 15\n",
|
| 849 |
+
"\n",
|
| 850 |
+
" accuracy 0.60 50\n",
|
| 851 |
+
" macro avg 0.56 0.57 0.55 50\n",
|
| 852 |
+
"weighted avg 0.60 0.60 0.59 50\n",
|
| 853 |
+
"\n",
|
| 854 |
+
"\n",
|
| 855 |
+
"=== Confusion Matrix ===\n",
|
| 856 |
+
"[[ 9 0 5 1]\n",
|
| 857 |
+
" [ 1 6 0 3]\n",
|
| 858 |
+
" [ 0 6 2 2]\n",
|
| 859 |
+
" [ 1 1 0 13]]\n",
|
| 860 |
+
"=============================================\n",
|
| 861 |
+
"\n",
|
| 862 |
+
"Evaluating model_10.pth ...\n",
|
| 863 |
+
"\n",
|
| 864 |
+
"=== Classification Report ===\n",
|
| 865 |
+
" precision recall f1-score support\n",
|
| 866 |
+
"\n",
|
| 867 |
+
" FTC 0.88 0.93 0.90 15\n",
|
| 868 |
+
" PTC 0.53 0.80 0.64 10\n",
|
| 869 |
+
" MTC 0.67 0.40 0.50 10\n",
|
| 870 |
+
" Benign 1.00 0.87 0.93 15\n",
|
| 871 |
+
"\n",
|
| 872 |
+
" accuracy 0.78 50\n",
|
| 873 |
+
" macro avg 0.77 0.75 0.74 50\n",
|
| 874 |
+
"weighted avg 0.80 0.78 0.78 50\n",
|
| 875 |
+
"\n",
|
| 876 |
+
"\n",
|
| 877 |
+
"=== Confusion Matrix ===\n",
|
| 878 |
+
"[[14 0 1 0]\n",
|
| 879 |
+
" [ 1 8 1 0]\n",
|
| 880 |
+
" [ 1 5 4 0]\n",
|
| 881 |
+
" [ 0 2 0 13]]\n",
|
| 882 |
+
"=============================================\n",
|
| 883 |
+
"\n",
|
| 884 |
+
"Evaluating model_11.pth ...\n",
|
| 885 |
+
"\n",
|
| 886 |
+
"=== Classification Report ===\n",
|
| 887 |
+
" precision recall f1-score support\n",
|
| 888 |
+
"\n",
|
| 889 |
+
" FTC 0.74 0.93 0.82 15\n",
|
| 890 |
+
" PTC 0.73 0.80 0.76 10\n",
|
| 891 |
+
" MTC 0.75 0.60 0.67 10\n",
|
| 892 |
+
" Benign 1.00 0.80 0.89 15\n",
|
| 893 |
+
"\n",
|
| 894 |
+
" accuracy 0.80 50\n",
|
| 895 |
+
" macro avg 0.80 0.78 0.79 50\n",
|
| 896 |
+
"weighted avg 0.82 0.80 0.80 50\n",
|
| 897 |
+
"\n",
|
| 898 |
+
"\n",
|
| 899 |
+
"=== Confusion Matrix ===\n",
|
| 900 |
+
"[[14 0 1 0]\n",
|
| 901 |
+
" [ 1 8 1 0]\n",
|
| 902 |
+
" [ 2 2 6 0]\n",
|
| 903 |
+
" [ 2 1 0 12]]\n",
|
| 904 |
+
"=============================================\n",
|
| 905 |
+
"\n",
|
| 906 |
+
"Evaluating model_12.pth ...\n",
|
| 907 |
+
"\n",
|
| 908 |
+
"=== Classification Report ===\n",
|
| 909 |
+
" precision recall f1-score support\n",
|
| 910 |
+
"\n",
|
| 911 |
+
" FTC 1.00 0.73 0.85 15\n",
|
| 912 |
+
" PTC 0.75 0.90 0.82 10\n",
|
| 913 |
+
" MTC 0.60 0.60 0.60 10\n",
|
| 914 |
+
" Benign 0.88 1.00 0.94 15\n",
|
| 915 |
+
"\n",
|
| 916 |
+
" accuracy 0.82 50\n",
|
| 917 |
+
" macro avg 0.81 0.81 0.80 50\n",
|
| 918 |
+
"weighted avg 0.83 0.82 0.82 50\n",
|
| 919 |
+
"\n",
|
| 920 |
+
"\n",
|
| 921 |
+
"=== Confusion Matrix ===\n",
|
| 922 |
+
"[[11 0 3 1]\n",
|
| 923 |
+
" [ 0 9 1 0]\n",
|
| 924 |
+
" [ 0 3 6 1]\n",
|
| 925 |
+
" [ 0 0 0 15]]\n",
|
| 926 |
+
"=============================================\n",
|
| 927 |
+
"\n",
|
| 928 |
+
"Evaluating model_13.pth ...\n",
|
| 929 |
+
"\n",
|
| 930 |
+
"=== Classification Report ===\n",
|
| 931 |
+
" precision recall f1-score support\n",
|
| 932 |
+
"\n",
|
| 933 |
+
" FTC 1.00 0.73 0.85 15\n",
|
| 934 |
+
" PTC 0.73 0.80 0.76 10\n",
|
| 935 |
+
" MTC 0.60 0.90 0.72 10\n",
|
| 936 |
+
" Benign 1.00 0.87 0.93 15\n",
|
| 937 |
+
"\n",
|
| 938 |
+
" accuracy 0.82 50\n",
|
| 939 |
+
" macro avg 0.83 0.82 0.81 50\n",
|
| 940 |
+
"weighted avg 0.87 0.82 0.83 50\n",
|
| 941 |
+
"\n",
|
| 942 |
+
"\n",
|
| 943 |
+
"=== Confusion Matrix ===\n",
|
| 944 |
+
"[[11 0 4 0]\n",
|
| 945 |
+
" [ 0 8 2 0]\n",
|
| 946 |
+
" [ 0 1 9 0]\n",
|
| 947 |
+
" [ 0 2 0 13]]\n",
|
| 948 |
+
"=============================================\n",
|
| 949 |
+
"\n",
|
| 950 |
+
"Evaluating model_14.pth ...\n",
|
| 951 |
+
"\n",
|
| 952 |
+
"=== Classification Report ===\n",
|
| 953 |
+
" precision recall f1-score support\n",
|
| 954 |
+
"\n",
|
| 955 |
+
" FTC 0.88 0.93 0.90 15\n",
|
| 956 |
+
" PTC 0.86 0.60 0.71 10\n",
|
| 957 |
+
" MTC 0.69 0.90 0.78 10\n",
|
| 958 |
+
" Benign 0.93 0.87 0.90 15\n",
|
| 959 |
+
"\n",
|
| 960 |
+
" accuracy 0.84 50\n",
|
| 961 |
+
" macro avg 0.84 0.82 0.82 50\n",
|
| 962 |
+
"weighted avg 0.85 0.84 0.84 50\n",
|
| 963 |
+
"\n",
|
| 964 |
+
"\n",
|
| 965 |
+
"=== Confusion Matrix ===\n",
|
| 966 |
+
"[[14 0 1 0]\n",
|
| 967 |
+
" [ 1 6 3 0]\n",
|
| 968 |
+
" [ 0 0 9 1]\n",
|
| 969 |
+
" [ 1 1 0 13]]\n",
|
| 970 |
+
"=============================================\n",
|
| 971 |
+
"\n",
|
| 972 |
+
"Evaluating model_15.pth ...\n",
|
| 973 |
+
"\n",
|
| 974 |
+
"=== Classification Report ===\n",
|
| 975 |
+
" precision recall f1-score support\n",
|
| 976 |
+
"\n",
|
| 977 |
+
" FTC 0.88 0.93 0.90 15\n",
|
| 978 |
+
" PTC 0.80 0.80 0.80 10\n",
|
| 979 |
+
" MTC 0.80 0.80 0.80 10\n",
|
| 980 |
+
" Benign 0.93 0.87 0.90 15\n",
|
| 981 |
+
"\n",
|
| 982 |
+
" accuracy 0.86 50\n",
|
| 983 |
+
" macro avg 0.85 0.85 0.85 50\n",
|
| 984 |
+
"weighted avg 0.86 0.86 0.86 50\n",
|
| 985 |
+
"\n",
|
| 986 |
+
"\n",
|
| 987 |
+
"=== Confusion Matrix ===\n",
|
| 988 |
+
"[[14 0 1 0]\n",
|
| 989 |
+
" [ 1 8 1 0]\n",
|
| 990 |
+
" [ 0 1 8 1]\n",
|
| 991 |
+
" [ 1 1 0 13]]\n",
|
| 992 |
+
"=============================================\n",
|
| 993 |
+
"\n",
|
| 994 |
+
"Evaluating model_16.pth ...\n",
|
| 995 |
+
"\n",
|
| 996 |
+
"=== Classification Report ===\n",
|
| 997 |
+
" precision recall f1-score support\n",
|
| 998 |
+
"\n",
|
| 999 |
+
" FTC 1.00 0.93 0.97 15\n",
|
| 1000 |
+
" PTC 0.56 0.90 0.69 10\n",
|
| 1001 |
+
" MTC 0.71 0.50 0.59 10\n",
|
| 1002 |
+
" Benign 1.00 0.87 0.93 15\n",
|
| 1003 |
+
"\n",
|
| 1004 |
+
" accuracy 0.82 50\n",
|
| 1005 |
+
" macro avg 0.82 0.80 0.79 50\n",
|
| 1006 |
+
"weighted avg 0.86 0.82 0.82 50\n",
|
| 1007 |
+
"\n",
|
| 1008 |
+
"\n",
|
| 1009 |
+
"=== Confusion Matrix ===\n",
|
| 1010 |
+
"[[14 0 1 0]\n",
|
| 1011 |
+
" [ 0 9 1 0]\n",
|
| 1012 |
+
" [ 0 5 5 0]\n",
|
| 1013 |
+
" [ 0 2 0 13]]\n",
|
| 1014 |
+
"=============================================\n",
|
| 1015 |
+
"\n",
|
| 1016 |
+
"Evaluating model_17.pth ...\n",
|
| 1017 |
+
"\n",
|
| 1018 |
+
"=== Classification Report ===\n",
|
| 1019 |
+
" precision recall f1-score support\n",
|
| 1020 |
+
"\n",
|
| 1021 |
+
" FTC 0.93 0.93 0.93 15\n",
|
| 1022 |
+
" PTC 0.88 0.70 0.78 10\n",
|
| 1023 |
+
" MTC 0.80 0.80 0.80 10\n",
|
| 1024 |
+
" Benign 0.88 1.00 0.94 15\n",
|
| 1025 |
+
"\n",
|
| 1026 |
+
" accuracy 0.88 50\n",
|
| 1027 |
+
" macro avg 0.87 0.86 0.86 50\n",
|
| 1028 |
+
"weighted avg 0.88 0.88 0.88 50\n",
|
| 1029 |
+
"\n",
|
| 1030 |
+
"\n",
|
| 1031 |
+
"=== Confusion Matrix ===\n",
|
| 1032 |
+
"[[14 0 1 0]\n",
|
| 1033 |
+
" [ 1 7 1 1]\n",
|
| 1034 |
+
" [ 0 1 8 1]\n",
|
| 1035 |
+
" [ 0 0 0 15]]\n",
|
| 1036 |
+
"=============================================\n",
|
| 1037 |
+
"\n",
|
| 1038 |
+
"Evaluating model_18.pth ...\n",
|
| 1039 |
+
"\n",
|
| 1040 |
+
"=== Classification Report ===\n",
|
| 1041 |
+
" precision recall f1-score support\n",
|
| 1042 |
+
"\n",
|
| 1043 |
+
" FTC 1.00 0.60 0.75 15\n",
|
| 1044 |
+
" PTC 0.90 0.90 0.90 10\n",
|
| 1045 |
+
" MTC 0.57 0.80 0.67 10\n",
|
| 1046 |
+
" Benign 0.88 1.00 0.94 15\n",
|
| 1047 |
+
"\n",
|
| 1048 |
+
" accuracy 0.82 50\n",
|
| 1049 |
+
" macro avg 0.84 0.82 0.81 50\n",
|
| 1050 |
+
"weighted avg 0.86 0.82 0.82 50\n",
|
| 1051 |
+
"\n",
|
| 1052 |
+
"\n",
|
| 1053 |
+
"=== Confusion Matrix ===\n",
|
| 1054 |
+
"[[ 9 0 5 1]\n",
|
| 1055 |
+
" [ 0 9 1 0]\n",
|
| 1056 |
+
" [ 0 1 8 1]\n",
|
| 1057 |
+
" [ 0 0 0 15]]\n",
|
| 1058 |
+
"=============================================\n",
|
| 1059 |
+
"\n",
|
| 1060 |
+
"Evaluating model_19.pth ...\n",
|
| 1061 |
+
"\n",
|
| 1062 |
+
"=== Classification Report ===\n",
|
| 1063 |
+
" precision recall f1-score support\n",
|
| 1064 |
+
"\n",
|
| 1065 |
+
" FTC 0.88 0.93 0.90 15\n",
|
| 1066 |
+
" PTC 0.80 0.80 0.80 10\n",
|
| 1067 |
+
" MTC 0.78 0.70 0.74 10\n",
|
| 1068 |
+
" Benign 0.93 0.93 0.93 15\n",
|
| 1069 |
+
"\n",
|
| 1070 |
+
" accuracy 0.86 50\n",
|
| 1071 |
+
" macro avg 0.85 0.84 0.84 50\n",
|
| 1072 |
+
"weighted avg 0.86 0.86 0.86 50\n",
|
| 1073 |
+
"\n",
|
| 1074 |
+
"\n",
|
| 1075 |
+
"=== Confusion Matrix ===\n",
|
| 1076 |
+
"[[14 0 1 0]\n",
|
| 1077 |
+
" [ 1 8 1 0]\n",
|
| 1078 |
+
" [ 1 1 7 1]\n",
|
| 1079 |
+
" [ 0 1 0 14]]\n",
|
| 1080 |
+
"=============================================\n",
|
| 1081 |
+
"\n",
|
| 1082 |
+
"Evaluating model_2.pth ...\n",
|
| 1083 |
+
"\n",
|
| 1084 |
+
"=== Classification Report ===\n",
|
| 1085 |
+
" precision recall f1-score support\n",
|
| 1086 |
+
"\n",
|
| 1087 |
+
" FTC 0.93 0.93 0.93 15\n",
|
| 1088 |
+
" PTC 1.00 0.40 0.57 10\n",
|
| 1089 |
+
" MTC 0.75 0.90 0.82 10\n",
|
| 1090 |
+
" Benign 0.74 0.93 0.82 15\n",
|
| 1091 |
+
"\n",
|
| 1092 |
+
" accuracy 0.82 50\n",
|
| 1093 |
+
" macro avg 0.86 0.79 0.79 50\n",
|
| 1094 |
+
"weighted avg 0.85 0.82 0.80 50\n",
|
| 1095 |
+
"\n",
|
| 1096 |
+
"\n",
|
| 1097 |
+
"=== Confusion Matrix ===\n",
|
| 1098 |
+
"[[14 0 0 1]\n",
|
| 1099 |
+
" [ 1 4 2 3]\n",
|
| 1100 |
+
" [ 0 0 9 1]\n",
|
| 1101 |
+
" [ 0 0 1 14]]\n",
|
| 1102 |
+
"=============================================\n",
|
| 1103 |
+
"\n",
|
| 1104 |
+
"Evaluating model_3.pth ...\n",
|
| 1105 |
+
"\n",
|
| 1106 |
+
"=== Classification Report ===\n",
|
| 1107 |
+
" precision recall f1-score support\n",
|
| 1108 |
+
"\n",
|
| 1109 |
+
" FTC 0.93 0.93 0.93 15\n",
|
| 1110 |
+
" PTC 0.75 0.30 0.43 10\n",
|
| 1111 |
+
" MTC 0.56 1.00 0.71 10\n",
|
| 1112 |
+
" Benign 1.00 0.87 0.93 15\n",
|
| 1113 |
+
"\n",
|
| 1114 |
+
" accuracy 0.80 50\n",
|
| 1115 |
+
" macro avg 0.81 0.78 0.75 50\n",
|
| 1116 |
+
"weighted avg 0.84 0.80 0.79 50\n",
|
| 1117 |
+
"\n",
|
| 1118 |
+
"\n",
|
| 1119 |
+
"=== Confusion Matrix ===\n",
|
| 1120 |
+
"[[14 0 1 0]\n",
|
| 1121 |
+
" [ 1 3 6 0]\n",
|
| 1122 |
+
" [ 0 0 10 0]\n",
|
| 1123 |
+
" [ 0 1 1 13]]\n",
|
| 1124 |
+
"=============================================\n",
|
| 1125 |
+
"\n",
|
| 1126 |
+
"Evaluating model_4.pth ...\n",
|
| 1127 |
+
"\n",
|
| 1128 |
+
"=== Classification Report ===\n",
|
| 1129 |
+
" precision recall f1-score support\n",
|
| 1130 |
+
"\n",
|
| 1131 |
+
" FTC 0.64 0.93 0.76 15\n",
|
| 1132 |
+
" PTC 1.00 0.40 0.57 10\n",
|
| 1133 |
+
" MTC 0.86 0.60 0.71 10\n",
|
| 1134 |
+
" Benign 0.82 0.93 0.88 15\n",
|
| 1135 |
+
"\n",
|
| 1136 |
+
" accuracy 0.76 50\n",
|
| 1137 |
+
" macro avg 0.83 0.72 0.73 50\n",
|
| 1138 |
+
"weighted avg 0.81 0.76 0.74 50\n",
|
| 1139 |
+
"\n",
|
| 1140 |
+
"\n",
|
| 1141 |
+
"=== Confusion Matrix ===\n",
|
| 1142 |
+
"[[14 0 0 1]\n",
|
| 1143 |
+
" [ 4 4 1 1]\n",
|
| 1144 |
+
" [ 3 0 6 1]\n",
|
| 1145 |
+
" [ 1 0 0 14]]\n",
|
| 1146 |
+
"=============================================\n",
|
| 1147 |
+
"\n",
|
| 1148 |
+
"Evaluating model_5.pth ...\n",
|
| 1149 |
+
"\n",
|
| 1150 |
+
"=== Classification Report ===\n",
|
| 1151 |
+
" precision recall f1-score support\n",
|
| 1152 |
+
"\n",
|
| 1153 |
+
" FTC 0.94 1.00 0.97 15\n",
|
| 1154 |
+
" PTC 0.67 0.80 0.73 10\n",
|
| 1155 |
+
" MTC 0.89 0.80 0.84 10\n",
|
| 1156 |
+
" Benign 1.00 0.87 0.93 15\n",
|
| 1157 |
+
"\n",
|
| 1158 |
+
" accuracy 0.88 50\n",
|
| 1159 |
+
" macro avg 0.87 0.87 0.87 50\n",
|
| 1160 |
+
"weighted avg 0.89 0.88 0.88 50\n",
|
| 1161 |
+
"\n",
|
| 1162 |
+
"\n",
|
| 1163 |
+
"=== Confusion Matrix ===\n",
|
| 1164 |
+
"[[15 0 0 0]\n",
|
| 1165 |
+
" [ 1 8 1 0]\n",
|
| 1166 |
+
" [ 0 2 8 0]\n",
|
| 1167 |
+
" [ 0 2 0 13]]\n",
|
| 1168 |
+
"=============================================\n",
|
| 1169 |
+
"\n",
|
| 1170 |
+
"Evaluating model_6.pth ...\n",
|
| 1171 |
+
"\n",
|
| 1172 |
+
"=== Classification Report ===\n",
|
| 1173 |
+
" precision recall f1-score support\n",
|
| 1174 |
+
"\n",
|
| 1175 |
+
" FTC 0.92 0.80 0.86 15\n",
|
| 1176 |
+
" PTC 0.70 0.70 0.70 10\n",
|
| 1177 |
+
" MTC 0.57 0.80 0.67 10\n",
|
| 1178 |
+
" Benign 1.00 0.87 0.93 15\n",
|
| 1179 |
+
"\n",
|
| 1180 |
+
" accuracy 0.80 50\n",
|
| 1181 |
+
" macro avg 0.80 0.79 0.79 50\n",
|
| 1182 |
+
"weighted avg 0.83 0.80 0.81 50\n",
|
| 1183 |
+
"\n",
|
| 1184 |
+
"\n",
|
| 1185 |
+
"=== Confusion Matrix ===\n",
|
| 1186 |
+
"[[12 0 3 0]\n",
|
| 1187 |
+
" [ 1 7 2 0]\n",
|
| 1188 |
+
" [ 0 2 8 0]\n",
|
| 1189 |
+
" [ 0 1 1 13]]\n",
|
| 1190 |
+
"=============================================\n",
|
| 1191 |
+
"\n",
|
| 1192 |
+
"Evaluating model_7.pth ...\n",
|
| 1193 |
+
"\n",
|
| 1194 |
+
"=== Classification Report ===\n",
|
| 1195 |
+
" precision recall f1-score support\n",
|
| 1196 |
+
"\n",
|
| 1197 |
+
" FTC 0.87 0.87 0.87 15\n",
|
| 1198 |
+
" PTC 0.57 0.80 0.67 10\n",
|
| 1199 |
+
" MTC 0.56 0.50 0.53 10\n",
|
| 1200 |
+
" Benign 1.00 0.80 0.89 15\n",
|
| 1201 |
+
"\n",
|
| 1202 |
+
" accuracy 0.76 50\n",
|
| 1203 |
+
" macro avg 0.75 0.74 0.74 50\n",
|
| 1204 |
+
"weighted avg 0.79 0.76 0.77 50\n",
|
| 1205 |
+
"\n",
|
| 1206 |
+
"\n",
|
| 1207 |
+
"=== Confusion Matrix ===\n",
|
| 1208 |
+
"[[13 0 2 0]\n",
|
| 1209 |
+
" [ 1 8 1 0]\n",
|
| 1210 |
+
" [ 0 5 5 0]\n",
|
| 1211 |
+
" [ 1 1 1 12]]\n",
|
| 1212 |
+
"=============================================\n",
|
| 1213 |
+
"\n",
|
| 1214 |
+
"Evaluating model_8.pth ...\n",
|
| 1215 |
+
"\n",
|
| 1216 |
+
"=== Classification Report ===\n",
|
| 1217 |
+
" precision recall f1-score support\n",
|
| 1218 |
+
"\n",
|
| 1219 |
+
" FTC 1.00 0.87 0.93 15\n",
|
| 1220 |
+
" PTC 0.73 0.80 0.76 10\n",
|
| 1221 |
+
" MTC 0.58 0.70 0.64 10\n",
|
| 1222 |
+
" Benign 0.93 0.87 0.90 15\n",
|
| 1223 |
+
"\n",
|
| 1224 |
+
" accuracy 0.82 50\n",
|
| 1225 |
+
" macro avg 0.81 0.81 0.81 50\n",
|
| 1226 |
+
"weighted avg 0.84 0.82 0.83 50\n",
|
| 1227 |
+
"\n",
|
| 1228 |
+
"\n",
|
| 1229 |
+
"=== Confusion Matrix ===\n",
|
| 1230 |
+
"[[13 0 2 0]\n",
|
| 1231 |
+
" [ 0 8 2 0]\n",
|
| 1232 |
+
" [ 0 2 7 1]\n",
|
| 1233 |
+
" [ 0 1 1 13]]\n",
|
| 1234 |
+
"=============================================\n",
|
| 1235 |
+
"\n",
|
| 1236 |
+
"Evaluating model_9.pth ...\n",
|
| 1237 |
+
"\n",
|
| 1238 |
+
"=== Classification Report ===\n",
|
| 1239 |
+
" precision recall f1-score support\n",
|
| 1240 |
+
"\n",
|
| 1241 |
+
" FTC 0.83 1.00 0.91 15\n",
|
| 1242 |
+
" PTC 0.67 0.80 0.73 10\n",
|
| 1243 |
+
" MTC 0.71 0.50 0.59 10\n",
|
| 1244 |
+
" Benign 1.00 0.87 0.93 15\n",
|
| 1245 |
+
"\n",
|
| 1246 |
+
" accuracy 0.82 50\n",
|
| 1247 |
+
" macro avg 0.80 0.79 0.79 50\n",
|
| 1248 |
+
"weighted avg 0.83 0.82 0.81 50\n",
|
| 1249 |
+
"\n",
|
| 1250 |
+
"\n",
|
| 1251 |
+
"=== Confusion Matrix ===\n",
|
| 1252 |
+
"[[15 0 0 0]\n",
|
| 1253 |
+
" [ 0 8 2 0]\n",
|
| 1254 |
+
" [ 2 3 5 0]\n",
|
| 1255 |
+
" [ 1 1 0 13]]\n",
|
| 1256 |
+
"=============================================\n"
|
| 1257 |
+
]
|
| 1258 |
+
}
|
| 1259 |
+
]
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"cell_type": "markdown",
|
| 1263 |
+
"source": [
|
| 1264 |
+
"## Best model is:``model_17.pth``"
|
| 1265 |
+
],
|
| 1266 |
+
"metadata": {
|
| 1267 |
+
"id": "4oPsGo7TDShu"
|
| 1268 |
+
}
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"cell_type": "code",
|
| 1272 |
+
"source": [
|
| 1273 |
+
"import torch\n",
|
| 1274 |
+
"import torch.nn as nn\n",
|
| 1275 |
+
"from sklearn.metrics import confusion_matrix, classification_report\n",
|
| 1276 |
+
"import seaborn as sns\n",
|
| 1277 |
+
"import matplotlib.pyplot as plt\n",
|
| 1278 |
+
"\n",
|
| 1279 |
+
"# === Config ===\n",
|
| 1280 |
+
"model_path = \"model_17.pth\"\n",
|
| 1281 |
+
"classes = [\"FTC\", \"PTC\", \"MTC\",\"Benign\"] # adjust if you have 4 classes\n",
|
| 1282 |
+
"device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
|
| 1283 |
+
"\n",
|
| 1284 |
+
"# === Recreate model architecture ===\n",
|
| 1285 |
+
"model = CheckpointedAlexNet(num_classes=len(classes)).to(device)\n",
|
| 1286 |
+
"\n",
|
| 1287 |
+
"# === Load state dict ===\n",
|
| 1288 |
+
"model.load_state_dict(torch.load(model_path, map_location=device))\n",
|
| 1289 |
+
"model.eval()\n",
|
| 1290 |
+
"\n",
|
| 1291 |
+
"# === Collect predictions ===\n",
|
| 1292 |
+
"all_preds, all_labels = [], []\n",
|
| 1293 |
+
"with torch.no_grad():\n",
|
| 1294 |
+
" for imgs, labels in test_loader:\n",
|
| 1295 |
+
" imgs, labels = imgs.to(device), labels.to(device)\n",
|
| 1296 |
+
" outputs = model(imgs)\n",
|
| 1297 |
+
" _, preds = torch.max(outputs, 1)\n",
|
| 1298 |
+
" all_preds.extend(preds.cpu().numpy())\n",
|
| 1299 |
+
" all_labels.extend(labels.cpu().numpy())\n",
|
| 1300 |
+
"\n",
|
| 1301 |
+
"# === Confusion Matrix ===\n",
|
| 1302 |
+
"cm = confusion_matrix(all_labels, all_preds)\n",
|
| 1303 |
+
"\n",
|
| 1304 |
+
"plt.figure(figsize=(6, 5))\n",
|
| 1305 |
+
"sns.heatmap(cm, annot=True, fmt=\"d\", cmap=\"Blues\",\n",
|
| 1306 |
+
" xticklabels=classes, yticklabels=classes)\n",
|
| 1307 |
+
"plt.xlabel(\"Predicted\")\n",
|
| 1308 |
+
"plt.ylabel(\"True\")\n",
|
| 1309 |
+
"plt.title(\"Confusion Matrix - model_17.pth\")\n",
|
| 1310 |
+
"plt.tight_layout()\n",
|
| 1311 |
+
"plt.show()\n",
|
| 1312 |
+
"\n",
|
| 1313 |
+
"# === Classification report ===\n",
|
| 1314 |
+
"print(\"Classification Report:\\n\")\n",
|
| 1315 |
+
"print(classification_report(all_labels, all_preds, target_names=classes))\n"
|
| 1316 |
+
],
|
| 1317 |
+
"metadata": {
|
| 1318 |
+
"colab": {
|
| 1319 |
+
"base_uri": "https://localhost:8080/",
|
| 1320 |
+
"height": 733
|
| 1321 |
+
},
|
| 1322 |
+
"id": "P35hNwjvDXFU",
|
| 1323 |
+
"outputId": "5316101a-156b-4dc0-ee7a-114997260275"
|
| 1324 |
+
},
|
| 1325 |
+
"execution_count": 8,
|
| 1326 |
+
"outputs": [
|
| 1327 |
+
{
|
| 1328 |
+
"output_type": "display_data",
|
| 1329 |
+
"data": {
|
| 1330 |
+
"text/plain": [
|
| 1331 |
+
"<Figure size 600x500 with 2 Axes>"
|
| 1332 |
+
],
|
| 1333 |
+
"image/png": "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\n"
|
| 1334 |
+
},
|
| 1335 |
+
"metadata": {}
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"output_type": "stream",
|
| 1339 |
+
"name": "stdout",
|
| 1340 |
+
"text": [
|
| 1341 |
+
"Classification Report:\n",
|
| 1342 |
+
"\n",
|
| 1343 |
+
" precision recall f1-score support\n",
|
| 1344 |
+
"\n",
|
| 1345 |
+
" FTC 0.93 0.93 0.93 15\n",
|
| 1346 |
+
" PTC 0.88 0.70 0.78 10\n",
|
| 1347 |
+
" MTC 0.80 0.80 0.80 10\n",
|
| 1348 |
+
" Benign 0.88 1.00 0.94 15\n",
|
| 1349 |
+
"\n",
|
| 1350 |
+
" accuracy 0.88 50\n",
|
| 1351 |
+
" macro avg 0.87 0.86 0.86 50\n",
|
| 1352 |
+
"weighted avg 0.88 0.88 0.88 50\n",
|
| 1353 |
+
"\n"
|
| 1354 |
+
]
|
| 1355 |
+
}
|
| 1356 |
+
]
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"cell_type": "markdown",
|
| 1360 |
+
"source": [
|
| 1361 |
+
"## Final Report\n",
|
| 1362 |
+
"\n",
|
| 1363 |
+
"Benign: perfect classification (15/15)\n",
|
| 1364 |
+
"\n",
|
| 1365 |
+
"FTC: only one misclassified\n",
|
| 1366 |
+
"\n",
|
| 1367 |
+
"PTC: 2 misclassified (one as FTC, one as Benign)\n",
|
| 1368 |
+
"\n",
|
| 1369 |
+
"MTC: also strong, only a few mislabels"
|
| 1370 |
+
],
|
| 1371 |
+
"metadata": {
|
| 1372 |
+
"id": "gAzSMUGnEF5u"
|
| 1373 |
+
}
|
| 1374 |
+
}
|
| 1375 |
+
]
|
| 1376 |
+
}
|