Scientists have developed mathematical models to accurately predict breast cancer!
Release date: 2018-05-21
A team of researchers from Kaunas University of Technology (KTU) is developing mathematical methods for diagnosing breast cancer. By using deep learning, researchers try to teach computers to identify areas of malignancy, which can partially automate the breast cancer diagnosis process while increasing accuracy.
Image source: KTU
In 2014, approximately 93,500 people in Europe died of breast cancer, most of them women (92,500). Among women, breast cancer caused 3.7% of deaths. According to WHO data, more than 1 million people worldwide are diagnosed with breast cancer each year. The international medical professional team warned that the incidence of this cancer is increasing year by year, and the incidence rate in Lithuania has increased by 75% in the past 15 years.
Early diagnosis is key to better treating patients. “Diagnostics in the diagnosis of cancer often relies on visual information – analyzing tissue images to determine the degree of malignancy of the lesion. This process is time consuming and may also be misdiagnosed, which is fatal to cancer patients. Developed for diagnosis The mathematical model, we want to automate the diagnostic process to minimize the rate of misdiagnosis,†said Dr. Tomas Lesmantas, postdoctoral researcher at KTU.
To diagnose breast cancer, he introduced a neural network method founded by the British scientist Geoffrey Hinton, the father of deep learning. Dr. Iešmantas and his postdoctoral coordinator, Professor Robertas Alzbutas, analyzed microscopic images of more than 100 breast tissue provided by the University of Porto, including four types: non-tumor tissue, non-malignant tumor tissue, non-invasive cancer and invasive cancer. The goal is to design a mathematical model to distinguish the above four organizations.
“The initial results are very encouraging – our accuracy rate is 85%,†KTU researchers said.
He will present their results at the 15th Annual Conference on Image Analysis and Recognition in Portugal. Iešmantas said that although the application of mathematical methods in the medical field has been expanded in recent years, researchers are training computers to diagnose lung damage, identify lymph node metastases, and brain tumor localization, but in the short term, tumor diagnosis is unlikely to be fully automated. of.
“These studies are not just at the theoretical level, and some of the research methods are already in clinical use. Although digitalization is unlikely to replace human judgment, I believe that automated computer diagnostics will become more common and will help us to be more accurate. Diagnose certain cancers."
Reference materials:
Mathematical methods for diagnosing breast cancer
Source: Bio Valley
Garlic Granules,Granulated Garlic,Granulated Garlic Powder,Granulated Garlic To Garlic Powder
shandong changrong international trade co.,ltd. , https://www.changronggarliccn.com