Lung Cancer Prediction from 3D CT Scan.
- Post by: admin
- December 27, 2021
- Comments off
There is a unique opportunity in Pakistan for using AI in health care. In 2017, a competition hosted by Kaggle called Data Science Bowl 2017 (DSB17) was held, which released, for the rst time, a dataset of thoracic CT scans with cancer annotations. This provides an opportunity for researchers to develop AI systems that can learn to predict cancer. Capitalizing on these advancements, we want to introduce AI based disease diagnosis in Pakistan to alleviate the strain on our overloaded health care system. The doctors and radiologist manually inspect CT scans for malignant nodules and check for cancer. This process is tedious, time consuming and has a false positive rate of above 97%. An automated lung cancer prediction system would bypass the need for manual inspection and yield accurate and instant predictions. Lung cancer is the least survivable cancer with a survival rate of 20% and early detection could possibly double the survival rate. The need for an automated early lung cancer detection is dire all over the world and the need is accentuated even more for a country like Pakistan where public health is not a priority concern. In pursuit of this endeavor, we designed an AI based prototype solution leveraging state of the art deep learning strategies to reduce false positives in nodule detection which exhibited highly competitive results.
