Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 Dec 14:13:1080273.
doi: 10.3389/fphar.2022.1080273. eCollection 2022.

A novel deep learning segmentation model for organoid-based drug screening

Xiaowen Wang  1 , Chunyue Wu  2 , Shudi Zhang  1 , Pengfei Yu  1 , Lu Li  2 , Chunming Guo  2 , Rui Li  3
Affiliations

A novel deep learning segmentation model for organoid-based drug screening

Xiaowen Wang et al. Front Pharmacol. .

Abstract

Organoids are self-organized three-dimensional in vitro cell cultures derived from stem cells. They can recapitulate organ development, tissue regeneration, and disease progression and, hence, have broad applications in drug discovery. However, the lack of effective graphic algorithms for organoid growth analysis has slowed the development of organoid-based drug screening. In this study, we take advantage of a bladder cancer organoid system and develop a deep learning model, the res-double dynamic conv attention U-Net (RDAU-Net) model, to improve the efficiency and accuracy of organoid-based drug screenings. In this RDAU-Net model, the dynamic convolution and attention modules are integrated. The feature-extracting capability of the encoder and the utilization of multi-scale information are substantially enhanced, and the semantic gap caused by skip connections has been filled, which substantially improved its anti-interference ability. A total of 200 images of bladder cancer organoids on culture days 1, 3, 5, and 7, with or without drug treatment, were employed for training and testing. Compared with the other variations of the U-Net model, the segmentation indicators, such as Intersection over Union and dice similarity coefficient, in the RDAU-Net model have been improved. In addition, this algorithm effectively prevented false identification and missing identification, while maintaining a smooth edge contour of segmentation results. In summary, we proposed a novel method based on a deep learning model which could significantly improve the efficiency and accuracy of high-throughput drug screening and evaluation using organoids.

Keywords: RDAU-Net model; bladder cancer organoid; deep learning; drug screening; organoid segmentation.

PubMed Disclaimer

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Figures

FIGURE 1
FIGURE 1
Main steps of the automated drug screening assessment in organoids. The steps in the blue dotted line show the traditional, manual method, while in our method these steps are performed automatically by the computer.
FIGURE 2
FIGURE 2
Steps for making labels. (A) Original image of organoids on the third day. (B) Manual labeling of organoids. (C) Label generation by LabelMe software for model learning. The white regions are the _target organoids, and the black region is the background.
FIGURE 3
FIGURE 3
Comparison before and after non-local mean filtering. (A) Original image of organoids. (B) Image of organoids after preprocessing using the non-local mean filter. It can be observed that the background of the processed image is smoother.
FIGURE 4
FIGURE 4
Schematic of the RADU-Net model structure.
FIGURE 5
FIGURE 5
Schematic of the dynamic convolution module (Chen et al., 2020).
FIGURE 6
FIGURE 6
Schematic of the res-double dynamic convolution module.
FIGURE 7
FIGURE 7
Schematic of the CA module (Hou et al., 2021).
FIGURE 8
FIGURE 8
Schematic of the AG module (Oktay et al., 2018).
FIGURE 9
FIGURE 9
Heatmap with or without attention module. (A) Original image of organoids. (B) Heatmap without attention. (C) Heatmap with attention. The model with added attention is able to better focus on the _target region.
FIGURE 10
FIGURE 10
Segmentation results of different models for 1-, 3-, 5-, and 7-day images of organoids.
FIGURE 11
FIGURE 11
Comparison of segmentation indicators (precision, IoU, and DSC) of five models. Among the five models, RDAU-Net can achieve the best segmentation indicators.
FIGURE 12
FIGURE 12
Comparison of segmentation indicators (precision, IoU, and DSC) of ablation models. Each of the added modules can improve the performance of the U-Net.
FIGURE 13
FIGURE 13
Segmentation results of the RDAU-Net model compared with manual annotation. (A) Manual labeling of organoids. (B) RDAU-Net model labeling of organoids (the red markers indicate _targeted organoids). The comparison between Figure 13A and Figure 13B shows that the result of our model labeling is already very close to that of manual labeling.
FIGURE 14
FIGURE 14
Drug screening evaluation by the RDAU-Net model. (A) Area statistics of the organoid in the images (only a small part is shown here). (B) Violin plot shows the area changes of organoids in CTR, RA, and 14 groups at days 1, 3, 5 and 7. The area data statistics of the organoids of different treatments on the same day can clearly show the difference between the size of the organoids (adding a quartile distribution map, the dotted line in the middle indicates the median of the data in this group).
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Badrinarayanan Vijay, Kendall Alex, Cipolla Roberto. (2017). Segnet: A deep convolutional encoder-decoder architecture for image segmentation. IEEE Trans. Pattern Anal. Mach. Intell. 12, 2481–2495. 10.1109/TPAMI.2016.2644615 - DOI - PubMed
    1. Ballard David H., Boyer Christen J., Steven Alexander J. (2019). Organoids—Preclinical models of human disease. N. Engl. J. Med. 20, 1981–1982. 10.1056/NEJMc1903253 - DOI - PMC - PubMed
    1. Barker N., van Es J. H., Kuipers J., Kujala P., van den Born M., Cozijnsen M., et al. (2007). Identification of stem cells in small intestine and colon by marker gene Lgr5. Nature 449 (7165), 1003–1007. 10.1038/nature06196 - DOI - PubMed
    1. Bejnordi, Ehteshami B., Johannes van Diest P., van Ginneken B., Karssemeijer N., Litjens G., et al. (2017). Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast cancer. Jama 22, 2199–2210. 10.1001/jama.2017.14585 - DOI - PMC - PubMed
    1. Buades Antoni, Coll Bartomeu, Morel J-M. IEEE.A non-local algorithm for image denoising. Proceedings of the 2005 IEEE computer society conference on computer vision and pattern recognition (CVPR'05). San Diego, CA, USA June 2005
  NODES
twitter 2