NephroNet: A Novel Program for Identifying Renal Cell Carcinoma and Generating Synthetic Training Images with Convolutional Neural Networks and Diffusion Models 
Currently in development. Progress:
Part 1: Classifying
- Split trainng, validation, and testing samples
- Generate 20,000 patches per directory (Data Processing)
- Model Training (ResNet-18 @ 40 Epochs)
- Testing on Whole-Slide Images
- Implement thresholding with a grid search
- Visualization of predictions
- Final Testing & Confusion Matrix
Part 2: Generating
- Data Preprocessing
- Stable Diffusion (online, no modifiers)
- Dreambooth Text-to-Image (CompVis trained on DHMC dataset, fine-tuned UNet and later fine-tuned text encoder)
- Textual Inversion (will try on several tokens such as types of RCC)
- Text-to-Image (if time + money allows)
- Unconditional Image Generation (if time + money allows)
More to come soon (including results, research paper, code, etc).