Breast Cancer Detection & Classification using Transfer Learning
style="display:inline;">I affirm that all information submitted through this FYP application is correct and complete as to my best knowledge. I further agree that Ignite can approve, reject, defer or cancel this FYP application without mentioning any reason at any stage of NGIRI 2019. Information ca
2025-06-28 16:30:42 - Adil Khan
Project Title
Breast Cancer Detection & Classification using Transfer Learning
Project Area of Specialization Artificial IntelligenceProject Summary- Breast Cancer is the most common disease among women all over the world.
- 6,90,000 cases have been reported in Pakistan between 2017 – 2018.
- Traditional Process is Long and Expensive. Because of these factors, there's also a lack of follow-up on Patients.
- Machine Learning aided detection and classification can help the Radiologists filter out the Digital Mammograms of the patients preemptively.
- And it can also serve as a double-check for the Radiologist so that no diagnosis is made inaccurately.
- To precisely detect Breast Cancer in Digital Mammograms (if any)
- To accurately classify cancerous tissues & effected masses as Benign or Malignant.
- To accurately generate computable features for further analysis of the Digital Mammograms.
- We intend to use Transfer Learning, a machine learning technique, to detect and classify breast cancer in digital mammograms.
- As ML can be very resource intensive, we intend to train our models in the cloud.
- We also want to build our final application based on the cloud infrastructure, so that medical experts can use it from anywhere, anytime without requiring any specs-heavy hardware systems.
- Server-side implementation of the back-end and ML engine can enable us to gather Digital Mammography and Breast Cancer data from many medical experts from different areas to retrain and better our models and have a better statistical understanding of the problem in Pakistan.
- Building a Precise Breast Cancer Detector
- Generation of a Meaningful Visual Vocabulary for Breast Cancer Classification
- Robust Multi - Class Classifier for Classification of Malicious Tissues
- ML aided Classification for Breast Cancer, Averse to Over-fitting
- Bag of Features is intended to be used for Feature Generation and Categorization of Digital Mammograms.
- Support Vector Machines are intended to be used for the Detection and Classification of Breast Cancer in Digital Mammograms.
- The final application is intended to be deployed in the cloud to remove any specs-intensive hardware requirements by the medical professionals (the intended user) and their ease of access.
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 70000 | |||
| Amazon Web Services Access | Equipment | 1 | 70000 | 70000 |