MANILA — The Ateneo Laboratory for Intelligent Visual Environments (ALIVE) developed the YOLO 11n, an AI software that detects odontogenic sinusitis through dental X-rays with 98.2% accuracy.

Dr. Patricia Angela Abu, ALIVE head, made the technological breakthrough with her colleagues from Taiwan’s Chang Gung Memorial Hospital, National Cheng Kung University, Chung Yuan Christian University, and Ming Chi University of Technology with their findings published in the journal “Bioengineering”.

Dr. Abu told ABS-CBN News that the You Only Look Once (YOLO) model is an existing application for image detection used in different areas and fields.

It uses images as input data to fine-tune the machine for a specific task. In the case of Abu’s team of researchers, they trained the machine to detect ondontogenic sinusitis.

Odontogenic sinusitis is a frequently misdiagnosed dental condition that could spread to a patient's face, eyes, and brain if left unchecked.

Journalists urged to train on AI

Complications related to the upper teeth cause such a diagnosis, with symptoms consisting of nasal congestion, foul-smelling nasal discharge, and occasional tooth pain, similar to those of ordinary general sinusitis.

Treatment is often delayed as traditional diagnosis requires collaboration between dentists and otolaryngologists. About a third of patients experience noticeable dental pain, which general practitioners often overlook.

According to a press release from Ateneo de Manila University, using YOLO 11n makes diagnosis of faster as the model swiftly and accurately distinguishes the affected area in real time with only a single pass of the X-ray image

Its usage also provides various benefits as it is cost-effective to resource-limited areas, lessens radiation exposure as the need for CT scans reduces, and prevents further complications as it allows for immediate intervention from healthcare providers.

Abu shared that it took them less than 100 x-ray images to train the machine, to which she said is still not as definitive. She also shared that it was difficult to obtain sample images for the tech due to various reasons, including data privacy and non-disclosure agreements.

She said that while AI is a promising tool in the field of medicine, human intervention is still needed before finalizing a diagnosis.

PH seen boosting cybersecurity with DICT revamp

“Yeah, it's promising. It shows there are many biomedical imaging journals that make use of machine learning and deep learning to detect, recognize, categorize diseases, diagnosis, and many other things biomedical related,” she said.

While they are currently in the training and testing phase, the researcher said that they are aiming for its commercialization in the future.

“It would be ideal if we can actually already have it used by medical doctors. But of course, it also needs to pass through certain metrics if a machine or a product or a tool is something that meets the standards and can be used already for clinical diagnosis,” she said.

RELATED VIDEO