A female patient was admitted to Gazi University Faculty of Medicine Hospital Brain and Neurosurgery Outpatient Clinic with a complaint of non-specific headache and her doctor ordered a magnetic resonance imaging (MRI) scan.
The patient’s life was saved with early intervention after a tumor was detected by an artificial intelligence algorithm that analyzed real-time patient data while the patient was in the MRI device.
This situation entered the medical literature by being published in the “Journal of Neurosurgery: Case Lessons” as the first clinical case diagnosed by an artificial intelligence algorithm during the ongoing imaging of a patient during MRI, where the most appropriate treatment was quickly applied to the patient by instantly alerting the relevant physicians.
It Took 6 Months To Train The Algorithm
The details of the study were explained by Assoc. Prof. Dr. Emrah Çeltikçi, Lecturer at the Department of Neurosurgery, and Prof. Dr. Şeref Sağıroğlu, Project Coordinator of the Turkish Brain Project at Gazi University.
Brain tumor, pituitary and skull base surgery specialist Assoc. Prof. Dr. Çeltikçi said that it took 6 months to train the algorithm with information such as “stroke”, “bleeding” and “tumor” related to neurosurgery within the scope of the Turkish Brain Project, which started in February 2021.
Çeltikçi said that they then conducted studies on real data, completed the second version of the algorithm and started running the algorithm 24 hours a day on MRI devices at Gazi University Hospital.
In the meantime, Çeltikçi said that they included newly developed artificial intelligence models integrated into the hospital automation system where MRI images are recorded, so that the artificial intelligence algorithm was able to evaluate whether the brain MRI taken in 5-10 seconds while the patient was still in the MRI device contained an abnormal condition.
Stating that the artificial intelligence system was integrated into the video message service of a GSM operator with the support of the Digital Transformation Office, Assoc. Prof. Dr. Emrah Çeltikçi said, “Thus, when the algorithm detects an abnormality in the MRI, it is ensured that the algorithm sends a message to the doctors’ mobile phones through the special service developed and the anonymized image containing the abnormality is sent to the doctors. The artificial intelligence system colors the abnormal regions (segments) in MR sections containing abnormalities of the patient.”
Stating that the artificial intelligence algorithm has been sending alarms to doctors’ cell phones since the day it was established, Çeltikçi said the following about the patient who made medical history:
“It was in the evening. Our patient, who came to the outpatient clinic complaining of headache, was asked for an MRI. The artificial intelligence system detected an abnormality while our patient was still in the MRI device. When we looked at the phone, we saw a message that artificial intelligence detected brain edema and tumor in a serious part of our patient’s brain. Our doctors immediately took action and took the necessary measures. According to the biopsy we performed, our patient was diagnosed with a brain tumor and treatment was started.”
Çeltikçi stated that the number of MR imaging in Turkey is very high, and that radiologists need varying periods of time from hospital to hospital to evaluate these images, and added: “Artificial intelligence was able to show the abnormality requiring urgent intervention in a very short time. The aim of artificial intelligence is to work like an early alarm system.”
Çeltikçi reported that in their artificial intelligence studies, they aim to detect not only early alarm but also the name and genetic characteristics of the tumor.
“Our Patient Was The First Case In The World Literature”
Emrah Çeltikçi pointed out that artificial intelligence studies are being carried out in European countries and the USA and made the following assessment:
“We haven’t seen in the literature a real patient intervention detected by an algorithm that works in real time, which surprised us. Then we looked back and realized that the early warning systems developed in the US, for example, have not yet been approved by the FDA. Therefore, they do not have the capacity to test such a situation in real time with artificial intelligence. The bureaucratic obstacles there also do not allow this. In addition, there are problems in accessing data due to the private nature of the healthcare system in the US. This situation in the US was of course good for us. Our patient was actually the first person in the history of the world to be detected by artificial intelligence and intervened the day after his MRI was taken. We even published this. Our patient was the first case in the world literature. Not for experimental purposes, but a real patient entered an MRI device, artificial intelligence detected the tumor in his brain and notified the doctors and the patient was intervened. We are very happy and proud in that respect. After all, it is good to be doing something for the first time.”
Explaining that artificial intelligence enables early diagnosis of brain tumors so that the patient can be intervened urgently without waiting for radiologists to report the result of the MRI examination, Çeltikçi said that the artificial intelligence system developed especially for the early detection of the risk of “stroke” can be used in the emergency department of their hospital.
“It Was An Exemplary Project For Our Country”
Prof. Dr. Şeref Sağıroğlu, Director of the Center for Artificial Intelligence and Big Data Analytics and Security at Gazi University and coordinator of the Turkish Brain Project, said that the project is the first project in Turkey that is based on the philosophy of open data and can be used in real environments.
Explaining that up to 30 researchers, including Gazi University faculty members and researchers and the Presidency Digital Transformation Office team, worked and supported the project at different times, Sağıroğlu continued as follows:
“We anonymized and processed the data we obtained in accordance with the Personal Data Protection Law. Although it is very difficult to use MRI data in a research project, we succeeded. By using the data set we produced in our research and training artificial intelligence models, we developed a product that our doctors can use. We anonymized MRI data and made it available to researchers. This is an example for our country. We are extremely happy that one of our studies on early intervention to the patient as a result of tumor detection with artificial intelligence models from MRIs in real time was published in a world-renowned journal and early intervention entered the literature. We continue our work to produce newer solutions.”
Prof. Dr. Sağıroğlu also stated that they collaborated with a GSM operator to deliver the results of artificial intelligence to doctors’ mobile phones as an emergency alert.
Commitment To Free Installation In All Public Hospitals
Sağıroğlu reported that the artificial intelligence system developed within the scope of the project was committed to be installed free of charge in all public hospitals.
Stating that the project includes many innovations, Sarıoğlu concluded his words as follows:
“The Explainable Artificial Intelligence approach says ‘there is something abnormal here’ and explains its reasoning with a model. This system has increased the trust of our doctors. Doctors work closely with this system because they trust it. The system reported over 1000 abnormal conditions to our doctors. If every patient who comes to Gazi University Hospital has a brain-related MRI, the results are instantly sent to the doctors. Images from our 3 MRI devices are instantly analyzed by artificial intelligence. This gives confidence to our patients and enables our doctors to make rapid planning. The main reason why we voluntarily entered the project and want it to be installed free of charge in the public sector is to contribute to saving lives. We believe in the project wholeheartedly.”