Marlee Long

FIDI and AIDOC partner to apply artificial intelligence to radiology

Time is of the essence when it comes to the life of a patient. Using AI to prioritize service in cases of emergency is the goal of the tool that the Image Diagnostic Studies and Research Institute (FIDI), which manages image diagnostic services in the public health system, has incorporated in the cranial tomography services performed at the Mandaqui Hospital. The unit is the first one to obtain this technology, which was created by the Israeli company AIDOC, which identifies injuries in cranial tomography through artificial intelligence, aiding and signaling through the system that the patient should be prioritized.

The partnership between FIDI and AIDOC, a startup that focuses on applying artificial intelligence to radiology, with the support of the São Paulo State Health Office, brings to Brazil the first artificial intelligence research project for image diagnostics in the public health sector. The expectation is to offer qualified and quick service in cases of trauma, cerebrovascular accident, and other situations that lead to brain damage, in order to reduce its secondary effects and even save lives. Since the implementation of the project, at the beginning of this year, 92% of the cases with bleedings were duly detected and prioritized by the algorithm, reducing the report turnaround time by 36 minutes.

The software was developed by AIDOC two years ago, and it uses specific algorithms to detect the abnormalities even before the radiologist has access to the images.

In November 2017, the tool obtained the CE Mark, a certificate from the European Union which attests to the applicability and safety of medical devices. This week, the software was authorized by the Food and Drug Administration (FDA), a governmental entity in the USA which regulates food (both human and animal), food supplements, medications (human and animal), cosmetics, medical equipment, biological materials, and products derived from human blood, becoming the first company to be approved by the American government for the screening of patients with AI.

The goal is not to replace human capital in the analysis and reporting process since all the exams are analyzed and reported by the radiologist physician, but to provide even more precision and agility to the professional. “Every technological development is used for the benefit of the patient. With this tool, we streamlined the service for patients with more severe conditions, which aren’t always detectable before the analysis of the exam,” says Igor dos Santos, radiologist physician and Chief of Innovation in FIDI.

Putting the tool into action

The process has the following flow: when performing an exam, the images go to the FIDI server, which identifies if it is a cranial tomography, hides the patient’s data, and sends it to AIDOC’s server on the cloud. The images are sent back to the FIDI server with highlights of the abnormalities if there are any. The exam is re-identified with the patient’s data and is then forwarded to FIDI’s report center.

At the report center, the system automatically prioritizes the exam with the most serious condition, allowing the radiologist physician to analyze and report quicker so that the patient can be treated earlier. “It is important to highlight that the images that go to the AIDOC server are still simultaneously available on FIDI’s report center, that is, the use of the tool does not hinder the work that we already do,” dos Santos explains.

To incorporate the technology, FIDI and AIDOC validated the use of the tool by analyzing over 3 thousand images to confirm whether the areas highlighted by the software matched the reports made by FIDI without the tool. The results were significant, giving them the confidence to start the project, which is being performed at the Mandaqui Hospital and will be expanded to other units. The hospital was chosen due to the high occurrence of encephalic trauma and because it provides more modern CT scanners –  as of 64 channels – to receive such technology. “We are confident with this innovation and the possibility of leading the first artificial intelligence research project applied to image diagnostics in the Brazilian public health service,” Igor stated.

International acknowledgment

In November 2017, a team of Brazilian doctors made up by three FIDI radiologists achieved third place in a contest by the Radiological Society of North America (RSNA), in the first edition of the 2017 Pediatric Bone Age Challenge. With over 300 contestants from all over the world, Nitamar Abdala, the Foundation’s Counselor, and the radiologists Felipe Kitamura and Igor Santos, responsible for the Innovation Department at FIDI, developed along with experts from the Goiás Federal University (UFG) an algorithm that identifies, in milliseconds, the bone age in hand radiography exams. The acknowledgment was given during the RSNA Annual Meeting in Chicago (USA).


Read the original article in Portuguese here.

Explore the Latest AI Insights, Trends and Research

Marlee Long