Objective
Identify the most suitable medically relevant pretrained backbone for fine-tuning on cervical cancer cell classification.
Task Details
- Explore available medical domain-specific backbones (e.g., KimiaNet, RetCCL, HistAuCLR) and compare their performance with vision transformers.
- Compare model architectures, pretraining datasets, and relevance to histopathology.
- Evaluate performance on a small validation split or using transfer learning benchmarks.
- Determine the best fit with context to our use case
Deliverables
- Identification of the best medical domain specific backbone to use for fine tuning.
- Comprehensive reasoning along with proof(validation results) that can be mentioned in the research paper about why the selected model is the best choice.
Objective
Identify the most suitable medically relevant pretrained backbone for fine-tuning on cervical cancer cell classification.
Task Details
Deliverables