Hello, @vqdang @simongraham I'm sorry to bother you. I would like to realize if it is possible to improve classification accuracy of model by removing the copped annotations touching with borders of patchs like the picture below ?
It's a good question, but we haven't experimented with this. It may be worth exploring. It may also affect the stats! Probably actually makes more sense to assess performance for entire nuclei!
Hello, @vqdang @simongraham I'm sorry to bother you. I would like to realize if it is possible to improve classification accuracy of model by removing the copped annotations touching with borders of patchs like the picture below ?