Summary: Researchers at RUDN University, in collaboration with scientists from China and Saudi Arabia, have successfully developed a machine learning model that significantly improves the accuracy of breast cancer diagnosis through histological images. By incorporating additional attention modules into convolutional neural networks, the model achieved an accuracy rate of nearly 100%. This technological breakthrough is expected to reduce the burden on doctors, improve the treatment and diagnosis of breast cancer, and enhance the overall capabilities of medical image analysis.
In the field of medical diagnostics, accurate and timely diagnosis greatly influences the prognosis for patients with breast cancer. However, subjective factors and sample quality can often lead to incorrect diagnoses based on histology results. To address this issue, a team of mathematicians at RUDN University explored the potential of machine learning to more precisely recognize cancer within histological images.
Their approach involved testing various convolutional neural networks integrated with dual convolutional attention modules. These additional modules were designed to enhance the network’s ability to detect cancerous formations within the images. The model was trained and evaluated using the BreakHis dataset, which encompassed almost 10 thousand histological images obtained from 82 patients.
Among the tested models, the one that yielded the most promising results was a composition of the DenseNet211 convolutional network with attention modules. This model achieved an impressive accuracy rate of 99.6%. During their research, the mathematicians also observed that the recognition of cancerous formations was influenced by scale. Consequently, they emphasized the necessity of considering an appropriate approximation technique for real-world applications.
According to Ammar Muthanna, Ph.D., the Director of the Scientific Center for Modeling Wireless 5G Networks at RUDN University, the attention modules greatly improved the overall performance of the model, enhancing feature extraction and allowing the model to focus on critical areas of the images. Muthanna emphasizes the significance of attention mechanisms in analyzing medical images, stating that this breakthrough technology will not only alleviate the workload on doctors but also enhance the accuracy of tests, ultimately benefiting breast cancer treatment and diagnosis.
The source of the article is from the blog scimag.news