Researchers at KNUST, led by doctoral candidate Albert Dede, have developed an advanced AI model that significantly enhances the accuracy of breast cancer diagnosis, especially in low-resource healthcare settings. The model tackles the challenge of analyzing complex, high-resolution histopathology images by combining wavelet analysis and deformable convolutions—techniques that mimic how pathologists examine tissue slides.
Tested on the BreaKHis dataset, the model achieved impressive accuracy rates of 96.47% at the image level and 96.55% at the patient level, particularly excelling at high magnification (200×). Its low computational requirements make it ideal for hospitals with limited resources. The study is published in Applied Intelligence.
Source:https://www.knust.edu.gh/news/news-items/knust-researchers-develop-ai-model-enhance-breast-cancer-diagnosis
Tested on the BreaKHis dataset, the model achieved impressive accuracy rates of 96.47% at the image level and 96.55% at the patient level, particularly excelling at high magnification (200×). Its low computational requirements make it ideal for hospitals with limited resources. The study is published in Applied Intelligence.
Source:https://www.knust.edu.gh/news/news-items/knust-researchers-develop-ai-model-enhance-breast-cancer-diagnosis
