The software, developed through the work of academics from the Faculty of Medicine and our department, detects cancer cells using artificial intelligence.
Following our university's specialization in Mechanical and Manufacturing Technologies, the Digital Transformation Laboratory established a team of academics and students from the Faculty of Medicine and the Department of Computer Engineering to conduct research on the use of artificial intelligence in the diagnosis of various cancers.
As part of the study, approximately 12,000 labeled blood samples taken from leukemia patients by the hematology and medical microbiology units were used to teach the AI about 10 different cancer types. Furthermore, a digital apparatus integrated with the microscope was developed, enabling the conversion of laboratory data into digital data.
Assoc. Prof. Dr. Ünsal Savcı, Head of the Department of Medical Microbiology at the Faculty of Medicine, stated that AI achieved a 90% success rate in diagnosis after three different trials. "We are now starting to diagnose acute and chronic leukemia," Savcı said. "While the success rate for some types of cancer in cells is nearly 100 percent, we have achieved an average of 90 percent success in distinguishing normal cells from cancer," he said. Savcı stated that with increased data, AI's successful diagnosis rate could also approach 100 percent. He added that it could be used as a decision-support mechanism, even if not a decision-making mechanism, particularly in hospitals and healthcare institutions lacking hematology specialists.
Our Department Head, Prof. Dr. Akif Akgül, explained that they are developing studies on AI-based disease detection in blood cells with doctors from the medical school. He continued:
"We identified 10 different classes with 12,000 labeled samples on approximately 1,200 datasets. While some of these classes have achieved close to 100 percent success, we have achieved an average success rate of 90 percent overall." "The development process with different algorithms is ongoing, and we aim to exceed 90 percent accuracy," he said.
Akgül explained that the study on disease detection in blood cells began with the idea of eliminating diagnostic problems in healthcare institutions where there are insufficient specialist physicians, and continued:
"We developed a project to solve these problems. We developed a device for microscopes. By attaching this device, cells can be imaged, and different classes can be identified using the images, determining the type of disease these classes represent. Furthermore, with an interface, this device can send messages or emails to specialist physicians, providing information as a decision support system, and doctors can then make a final diagnosis." Akgül pointed out that the mobile application they are working on will allow doctors to follow the AI's diagnostic recommendations on their mobile phones.
Prof. Dr. Akgül emphasized that by involving students, as well as academics, in the project, they have contributed to the students' learning about the use of AI and the project development process.
