By Ashwini Sakharkar 6 Dec, 2024
Collected at: https://www.techexplorist.com/ai-predicts-mini-organs-development-early-stage/94221/
Organoids – tiny, lab-grown tissues that replicate the function and architecture of organs – are revolutionizing biomedical research. Sometimes known as “mini-organs,” they are crucial for regenerative medicine, drug discovery, and basic research.
Researchers from Kyushu University and Nagoya University in Japan have now created a model that utilizes artificial intelligence (AI) to predict the early-stage development of organoids. This model is faster and more accurate than evaluations made by expert researchers, which could enhance the efficiency and reduce the expenses associated with culturing organoids.
In their study, the researchers focused on predicting the growth of hypothalamic-pituitary organoids. These organoids simulate the functions of the pituitary gland, including the secretion of adrenocorticotropic hormone (ACTH), an essential hormone for the regulation of stress, metabolism, blood pressure, and inflammation. A deficiency in ACTH can result in fatigue, loss of appetite, and other potentially life-threatening complications.
“In our lab, our studies on mice show that transplanting hypothalamic-pituitary organoids has the potential to treat ACTH deficiency in humans,” says corresponding author Hidetaka Suga, Associate Professor of Nagoya University’s Graduate School of Medicine.
One major challenge faced by researchers is the ability to assess whether organoids are maturing correctly. Derived from stem cells suspended in liquid, organoids are highly responsive to slight changes in their environment, leading to inconsistencies in their growth and overall quality.
The researchers discovered that a key indicator of successful development is the widespread presence of a protein known as RAX during an early stage of maturation, which frequently results in organoids that secrete a substantial amount of ACTH later on.
“We can track development by genetically modifying the organoids to make the RAX protein fluoresce,” says Suga. “However, organoids intended for clinical use, like transplantation, can’t be genetically modified to fluoresce. So, our researchers must judge instead based on what they see with their eyes: a time-consuming and inaccurate process.”
“Deep-learning models are a type of AI that mimics the way the human brain processes information, allowing them to analyze and categorize large amounts of data by recognizing patterns,” explains Hirohiko Niioka, Professor of the Data-Driven Innovation Initiative at Kyushu University.
Researchers in Nagoya obtained both fluorescent and bright-field images—which display the organoids’ appearance under regular white light without any fluorescence—of organoids with fluorescent RAX proteins after 30 days of development.
They meticulously categorized 1500 bright-field images into three distinct quality levels based on the fluorescent images: A (wide RAX expression, high quality), B (medium RAX expression, medium quality), and C (narrow RAX expression, low quality). Under the guidance of Niioka, two state-of-the-art deep-learning models—EfficientNetV2-S and Vision Transformer, developed by Google—were trained using 1200 of these images (400 from each category) to predict quality classification.
Building upon this foundation, Niioka skillfully combined the two models into a robust ensemble model, significantly enhancing its performance. The research team then tested the optimized ensemble model using the remaining 300 images (100 from each classification), achieving a classification accuracy of 70% for the bright-field images of the organoids.
In comparison, researchers predicted the category of the same bright-field images; their accuracy was less than 60%.
“The deep-learning models outperformed the experts in all respects: in their accuracy, their sensitivity, and in their speed,” says Niioka.
The next step involved evaluating whether the ensemble model could accurately classify bright-field images of organoids that lacked genetic modifications to induce RAX fluorescence. The scientists evaluated the trained ensemble model using bright-field images of hypothalamic-pituitary organoids that did not contain fluorescent RAX proteins at 30 days of development.
Through staining methods, they found that the organoids classified as A (high quality) indeed exhibited elevated RAX expression at 30 days. As these organoids were further cultured, they subsequently demonstrated significant ACTH secretion. In contrast, the organoids designated as C (low quality) displayed low levels of RAX and, consequently, lower ACTH levels.
“Our model can, therefore, predict at an early stage of development what the final quality of the organoid will be, based solely on visual appearance,” says Niioka. “As far as we know, this is the first time in the world that deep-learning has been used to predict the future of organoid development.”
Next, the researchers plan to improve the accuracy of the deep-learning model by utilizing a more extensive dataset. However, even with its existing accuracy, the model holds significant implications for ongoing organoid studies.
“We can quickly and easily select high-quality organoids for transplantation and disease modeling and reduce time and costs by identifying and removing organoids that are developing less well,” concludes Suga. “It’s a game-changer.”
Journal reference:
- Tomoyoshi Asano, Hidetaka Suga, Hirohiko Niioka, Hiroshi Yukawa, Mayu Sakakibara, Shiori Taga, Mika Soen, Tsutomu Miwata, Hiroo Sasaki, Tomomi Seki, Saki Hasegawa, Sou Murakami, Masatoshi Abe, Yoshinori Yasuda, Takashi Miyata, Tomoko Kobayashi, Mariko Sugiyama, Takeshi Onoue, Daisuke Hagiwara, Shintaro Iwama, Yoshinobu Baba & Hiroshi Arima. A deep learning approach to predict differentiation outcomes in hypothalamic-pituitary organoids. Communications Biology, 2024; DOI: 10.1038/s42003-024-07109-1
Leave a Reply