Region of Interest-Based HEVC Video Transmission for Telemedicine in Healthcare
This paper proposes a Region of Interest-based HEVC modification for telemedicine that utilizes image segmentation and background blurring to significantly reduce video bitrate while maintaining high quality, thereby enabling efficient, uninterrupted data transmission over limited bandwidth.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the modern world of healthcare, the ability to see a patient clearly from miles away is often a matter of life and death. This is the promise of telemedicine, a system that allows doctors to diagnose and treat individuals without them ever needing to leave their homes. However, this visual connection relies entirely on the internet, and for many people in remote areas or during global health crises, that connection is fragile. Video streams require a significant amount of data to travel smoothly. When the internet connection is slow or unstable, the video freezes, pixelates, or cuts out entirely, leaving the doctor unable to see the patient's condition. To solve this, engineers use video compression, a process that shrinks video files by removing unnecessary information. The challenge has always been a trade-off: compress a video too much to save space, and the image becomes too blurry to be useful for medical diagnosis. The goal is to keep the picture sharp where it matters most while making the rest of the image smaller, a balance that standard video technology has struggled to achieve efficiently.
A team of researchers from universities in Pakistan, Malaysia, and Taiwan has developed a new method to tip this balance in favor of the doctor and the patient. They focused on a specific type of video coding known as HEVC, which is the current standard for high-quality video transmission. While this standard is already quite good at shrinking video files, it treats every part of the screen the same way. It does not know that a doctor's face or a patient's wound is more important than the wall behind them. The researchers modified this system to make it "smart" about what it is looking at. They built a computer program that can instantly identify the people in the video—the patient, the doctor, and any medical staff—and separate them from the background. Once the system knows who is who, it treats the people with extreme care, preserving their image in high definition, while it deliberately blurs and simplifies the background. This allows the video file to become much smaller without losing the clinical details that a doctor needs to see.
The core of this work lies in a technique called semantic segmentation, which is essentially a way for a computer to understand the meaning of an image rather than just its colors. The researchers trained a deep learning network, a type of artificial intelligence, to recognize human shapes in various positions, whether a person is standing, sitting, or lying down. This network acts as a filter that scans every frame of the video and draws an invisible map around the people. This map tells the video encoder exactly where the "region of interest" is. The encoder then applies a special rule: it uses a high-quality setting for the people inside that map and a low-quality setting for everything outside of it. By sacrificing the clarity of the wall, the floor, or the furniture, the system frees up enough data space to keep the human subjects crystal clear, even on a very slow internet connection.
To test if this approach actually works, the team ran extensive simulations using video sequences that mimicked real telemedicine scenarios. They compared their new method against the standard, unmodified video system. The results were striking. In a test involving a patient, the standard system required a data speed of 2,929.3 kilobits per second to transmit the video. Their new method managed to send the same video with the patient clearly visible using only 735.2 kilobits per second. This represents a massive reduction in the amount of data needed, effectively allowing the video to flow over connections that were previously too weak to support it. When they tested the system with a doctor, the difference was even more pronounced, dropping the required data speed from over 8,000 kilobits per second down to roughly 2,757 kilobits per second.
The researchers also measured the quality of the images using a standard metric that compares the original video to the compressed version. They found that while the background became intentionally blurry, the quality of the people in the video remained exceptionally high. In one specific test, their method achieved a quality score of 49.77 while using a data rate of 2,552 kilobits per second. The standard system, trying to keep the whole image sharp, needed a much higher data rate of 5,939 kilobits per second to reach a slightly lower quality score of 49.61. This proves that the new method can deliver a clearer picture of the patient using less than half the data of the traditional approach. The system was also tested against other existing methods that tried to solve similar problems, and it consistently outperformed them in both speed and image quality.
This work suggests a practical path forward for healthcare systems that operate in difficult environments. The researchers demonstrated that by focusing only on what is medically relevant, it is possible to maintain a high-quality visual link even when the internet is poor. The system is designed to be flexible, capable of handling one-on-one consultations or even connecting a single patient to a panel of multiple doctors. While the current tests were conducted using specific video datasets and simulations, the results indicate that this approach could significantly reduce the cost and complexity of telemedicine. By ensuring that the video does not freeze or degrade when the connection is weak, this technology could help doctors reach patients who are currently cut off from care due to distance or infrastructure limitations. The study concludes that this intelligent way of compressing video is a viable solution for the future of remote medical care, ensuring that the most important part of the screen—the human being—remains clear and visible.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.