By: ayushij074 on June 4, 2024, 10 a.m.
Are we going to get formats of slides as .mha instead of .tiff and .svs for validation and testing data? As, Tiffslide, Openslide doesn't support these formats. And the library which is provided in example submission SimpleITK is taking too much time and too many resources to read a WSI, which Tiffslide is able to do within no time.
By: KhrystynaFaryna on June 4, 2024, 10:38 a.m.
Dear participant, Thank you for your question. The format of the validation and test slides will be exactly the same as the format for the training slides: multiresolution tiff. The .mha files and SimpleITK library are examples autogenerated by grand-challenge platform. You can use any library you prefer to read slides, just specify it with the correct version in requirements.txt and it will get pip installed within the docker. Do not hesitate to reach out in case of other questions. Looking forward to your submission.
Kind regards, Khrystyna
By: ayushij074 on June 5, 2024, 10:36 a.m.
Thanks for the answer. I have one more question regarding the submission format, As a prediction result of the algorithm are you guys expecting only follow up years/reoccurrence ? and not occurrence of event?
By: KhrystynaFaryna on June 10, 2024, 7:54 a.m.
Yes, the prediction format is years to recurrence only. The predicted time to recurrence should correlate with the risk of recurrence in survival models. A shorter predicted time to recurrence generally indicates a higher risk of recurrence, and vice versa. The challenge evaluation metric is C-index. Intuitively, this metric measures how well the model ranks patients based on their risk.
