This project focuses on developing a model for detecting breast cancer using computer vision techniques. The model analyzes mammogram images to detect masses and lumps in the breast, which aids in early diagnosis and improves treatment chances. We used Roboflow for dataset collection and preprocessing, along with YOLOv8 for model building.
The project utilizes the CBIS-DDSM dataset (The Curated Breast Imaging Subset of DDSM). This dataset contains mammogram images with annotations indicating the presence of masses.
Dataset link: CBIS-DDSM Dataset
Various preprocessing techniques were applied to improve the model's performance:
Below is an image showing the result before and after applying preprocessing techniques. You can see how the preprocessing improved the clarity and quality of the image.
The YOLOv8 model was used to detect masses in the images. The model was trained on the CBIS-DDSM dataset and fine-tuned using advanced techniques like deep data augmentation.
The model achieved high accuracy in detecting masses, which was evaluated using a Confusion Matrix, demonstrating a significant improvement in performance after preprocessing techniques were applied.
The confusion matrix shows the model's classification performance, highlighting the distribution of correct and incorrect classifications.
This project demonstrates how computer vision techniques can be effectively applied to breast cancer detection. Using advanced preprocessing methods and powerful models like YOLOv8, high accuracy in mass detection can be achieved, supporting early diagnosis.
To review the project's source code, visit my GitHub repository.
This project is licensed under the MIT License.
