During the partial differential equations course, we were asked to use PDEs to solve a real-world problem. Since we are studying systems and biomedical engineering, we decided that we shall solve a medical problem, after searching for all types of problems where PDEs can be used, we decided that we will focus on BrC. We decided to use PDEs to make artificial neural networks that can be used to detect BrC at its early stages.
- 95.2% validation accuracy.
- 97.1% training accuracy.
- 0.14 validation loss.
- 0.07 training loss.
- We are using small dataset offered by Baheya Foundation, so one of our constraints is the dataset.
- We will be looking to use DCGANs to generate a bigger dataset.
- We might be using transfer learning in the future to achieve better accuracy.
- We used our model to validate X-Ray breast images (The model is trained on Ultrasound images), and it showed positive results, further training can be done on X-Ray and MRI images in the future.
- We have built our BioGenesis Website, but the model deployment on web is yet to come once we reach better accuracy.