Article · AI translation
IT Home reported on November 15 that on April 10, 2019, the Event Horizon Telescope (EHT) Collaboration released the first image of a black hole in human history. The black hole in the image is at the center of a galaxy called M87, about 55 million light-years from Earth.
The image’s iconic ring structure quickly became one of the most recognizable images in the history of modern science.
The image shows the “shadow” of a black hole, surrounded by a halo of light produced by swirling hot gas. This sight fully matches the predictions of Einstein’s general theory of relativity and supports a long-held scientific view: supermassive black holes are common at the centers of galaxies.
However, scientists have since disagreed over whether the original image accurately depicted the details and structure of the black hole’s halo.
IT Home noted that a new study published in Nature used an entirely new technique to reanalyze the original observational data. The results not only confirmed the original study’s conclusions but also revealed the structure of the black hole’s halo from a new perspective.
This independent study further strengthens confidence in this major achievement of modern science and brings the black hole image, captured from tens of millions of light-years away, into clearer focus once again.
A Telescope as Large as Earth
Imaging the supermassive black hole at the center of the M87 galaxy is an extremely challenging task. It is like trying to photograph a doughnut on the Moon from Earth. To obtain an image with sufficient clarity and detail requires a telescope with a diameter comparable to that of the entire Earth.
To achieve this, the Event Horizon Telescope project linked radio telescopes around the world and combined the signals they received to create a virtual telescope the size of Earth, enabling observations at the highest resolution then available to humanity.
This achievement ushered in a new era of black hole imaging research, finally allowing humanity to observe directly the extreme environment around a black hole.
In the original published study, 4 independent imaging teams within the Event Horizon Telescope Collaboration took part. Although all the teams obtained a similar ring structure, their depictions of the black hole’s detailed features differed somewhat.
Researchers noted that such differences are not uncommon: human judgment during imaging and algorithmic bias can both affect the final result.
However, one study also fundamentally questioned whether the observed object really had a ring structure.
Some differences are understandable. After all, combining data from a limited number of widely spaced telescopes to build a virtual telescope spanning the entire Earth presents many technical challenges in itself.
Other factors can also affect the observations. Each radio telescope collecting data may introduce errors because of its own instruments or the effects of Earth’s atmosphere.
Astronomers normally estimate and correct these errors during data analysis. But because the entire process is highly complex, some residual errors may still lead to incorrect images.
To address this problem, the new independent study developed a new data-processing method called “closure invariants.”
Traditional data-analysis methods usually process measurements from 2 telescopes as a pair, whereas closure invariants take a different approach, combining measurements from 3 telescopes at a time.
This method can eliminate errors caused by atmospheric conditions and instruments and extract structural information about the observed object more clearly.
Reconstructing the Black Hole’s True Appearance
To make full use of the data provided by closure invariants, the researchers designed a machine-learning imaging method called GenDIReCT, which can generate images consistent with the actual measurements.
To minimize human bias, the researchers trained the model on images of terrestrial objects such as cats, dogs, and cars. During training, therefore, the model did not know what a black hole looked like.
The researchers also subjected the model to a series of rigorous validation tests, including imaging tests on a large number of non-ring structures and validation using real observational data from other galaxies, such as Centaurus A.
After confirming that the model was sufficiently accurate, the researchers fixed the entire training and testing procedure to prevent human factors from influencing the final images.
On December 4, 2025, the researchers finally used this thoroughly validated imaging method to analyze the Event Horizon Telescope’s observational data for the M87 galaxy.
The results showed that the black hole’s ring structure was consistent across all datasets.
The halo’s appearance and diameter indicate a black hole with a mass about 6.5 billion times that of the Sun, viewed almost directly face-on. This viewing angle is somewhat like looking straight down into the eye at the center of a hurricane.
Verifying the First Black Hole Image from a New Perspective
A direct comparison of the study’s images with the image previously released by the Event Horizon Telescope Collaboration found that their halo diameters and overall orientations matched exactly.
Of course, some details still differ, such as the width of the halo and the surface brightness along the ring’s edge. But their overall structures are highly consistent.
The study presents the halo structure in this iconic black hole image from a new perspective and also explains and resolves some disagreements in earlier research.
To show that the study’s success did not depend entirely on advanced machine-learning technology, the researchers also developed a supplementary imaging method. It uses traditional statistical techniques and is entirely different from the machine-learning approach.
The results showed that, using the newly generated closure-invariant data from the study, this traditional method also confirmed the black hole’s ring structure.
Overall, the study provides strong independent verification of the structure of the supermassive black hole at the center of the M87 galaxy.
The technique could play a larger role in the future. Beyond astronomy, it could be used in other fields that need to combine data from multiple sensors to reconstruct unknown structures, such as medical image analysis and seismic imaging.
Reference:
https://www.nature.com/articles/s41586-026-11241-9
Advertising notice: External links in this article (including but not limited to hyperlinks, QR codes, and passcodes) are provided to convey more information and save time in selecting sources. The results are for reference only. All IT Home articles containing external links include this notice.