New Delhi: Breast cancer is the most common form of the the disease reported in women living in metropolitan India. And contrary to popular belief, it is reported in men as well. While it is largely treatable, doctors stress the importance of regular screenings and timely treatment for the same. Earlier this year, researchers at Northeastern University unveiled an artificial intelligence tool created to diagnose prostate cancer faster and more accurately. Researchers of the same group developed another AI tool for the diagnosis of breast cancer and that too with 99.72% accuracy.
Breast cancer conventionally accounts for 30% of the cancer cases reported in women every year. In 2024, the American Cancer Society found that breast cancer accounts for 30% of the female cancer cases reported globally every year. In 2024 itself, 42500 women are likely to succumb to the tumour. The study’s findings were published in Cancers journal. These projects are largely an effort to form an online framework for doctors to access AI tech for various types of cancer. This is likely to redefine digital pathology.
The AI tool would evaluate high-resolution images and evaluate historical data on how to identify cancer patterns and perform diagnoses. The system cannot miss a tumour in a biopsy and it does not get exhausted either after evaluating 10-20 people. Ideally, this would allow doctors to not just treat patients faster and accurately but also form AI models to diagnose uncommon and rare cancers with little patient data.
For the same project, researchers took advantage of the dataset publicly available with pictures of benign and malignant breast tissues found on the Breast Cancer Histopathological Database. With this data, experts build a learning model for increased accuracy and low risk of errors and train them on breast tissue image data. This, experts said, was like taking diagnosis of multiple doctors and then voting to pick the best decision.