Three-Dimensional Morphological Characterisation of Human Cortical Organoids Using a Customised Image Analysis Workflow

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript by Handcock et al. provides a novel and useful image analysis workflow to quantify cortical organoid morphology in a standardized way. The workflow developed and reported in this study will be useful for the field to assess cortical organoid morphology. However, there are some suggestions to further improve the manuscript as follows:

1) For Figures 4 and 5, it would be good to see representative images of the organoids with the image analysis overlaid (similar to previous figures), to show how the images were quantified for these specific measurements. Specifically, for Fig 4, it'll be good to see representative images of 4 vs. 6 month old organoid slices, and for Fig 5, it'll be good to see representative images of individual staining cell densities at 4 vs. 6 months. 

2) Would it be possible to show 3 dimensional image projections for further understanding of the image analysis workflow and specific localization patterns?

Minor Comments:

1) Please ensure table title and legend (for example, Table 1 and Table 2) and Figure title and legends (for all figures) are written out within the text. Currently, they appear to be embedded as part of the image file. 

2) Minor formatting errors: Line Number 198, please correct the minor grammatical errors as follows- (i) please remove the period (.) in the middle of the sentence, (ii) please change to data "are" instead of "is". Line Number 404, there is a period (.) before "Discussion".

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Translator        

The study presents a robust and flexible tool for image analysis that has the potential to improve reproducibility in the field and facilitate the evaluation of cortical organoid morphology. The developed workflow demonstrates impressive accuracy in nuclei segmentation, particularly in comparison to widely used tools such as CellProfiler, Imaris, and Cellpose. The analysis of nuclear size is used to classify nuclei as viable, non-viable, or clustered, enabling the evaluation of tissue health, such as distinguishing viable tissue regions from the necrotic core.

In addition to nuclei analysis, the authors developed a method for evaluating antibody staining of non-nuclear markers. This approach analyzes the peri-nuclear region of each cell to identify cells as positively or negatively stained. The pipeline was employed to describe the tissue composition of cortical organoids at four and six months of differentiation.

A key suggestion for improvement is to provide scripts and the test dataset in open-access repositories such as GitHub. Open access would enhance the pipeline's accessibility and increase the study's impact. Furthermore, without access to the script and test dataset, it is impossible to fully evaluate the workflow's efficiency, especially since the paper only presents one exemplary image from the 88 analyzed sections.

Additional comments include the following:

1. Introduction Focus:
   The introduction emphasizes the impact of hiPSC-derived cortical organoids, yet only half of the data is based on hiPSC-derived organoids, while the other half uses organoids derived from the H9 embryonic stem cell (ESC) line. As ESC lines cannot be used for personalized approaches, the emphasis on hiPSC-derived organoids seems excessive. The introduction also lacks sufficient information about current imaging and image analysis methods for complex tissue samples and organoids.

2. Expanded Application Suggestions:
   At line 60, the statement, “This imaging workflow would serve as an important exercise in standardization,” could be elaborated by suggesting specific applications, such as improving differentiation protocols.

3. Clarification of "High-Content Imaging":
   At lines 69–72, the authors state, “High-content imaging of organoid models offers a promising solution to accelerate tissue analysis, especially as these models become more structurally complex as they mature.” However, "high-content" encompasses a wide range of techniques. Including an overview of the data types crucial for their workflow would enhance clarity.

4. Imaging Methods and Workflow Requirements:
   At lines 72–79, the discussion contrasts classical tissue sectioning with 3D imaging methods such as two-photon imaging and tissue clearing. However, it remains unclear whether the authors’ image analysis method also requires tissue sectioning, potentially creating the misleading impression that their method supports 3D imaging of intact organoids.

5. Imaging Equipment Usage:
   The "Materials and Methods" section mentions two imaging setups: a Leica 165 THUNDER Imager DMI8 fluorescence microscope and an LSM900 scanning confocal microscope. It is unclear whether both datasets were used in the workflow or if specific tasks were assigned to each setup.

6. Tissue Boundary Identification:
   Figure 2C shows the identification of the “Necrotic core” and “Viable region.” However, the Results section 3.2 (lines 220–228) does not explain how this was achieved. Additionally, an example illustrating tissue artefacts, such as tears and folds, and their impact on tissue boundary identification is necessary.

7. Quantitative Support for Artefact Detection:
   The claim that artefact detection “helped filter out non-biological features, ensuring more reliable measurements” (lines 224–226) lacks supporting examples or metrics. Including quantitative parameters, such as the percentage of artefacts excluded compared to manual annotation or alternative methods, would strengthen the argument.

8. Comparison with Classical Quantitative Methods:
   In section 3.6.3, the developed method estimates the percentage of cells positive for GABA, MAP2, and S100b. It would be helpful to compare these results with classical quantitative methods, such as flow cytometry. A direct comparison using data from the same batch of organoids would be ideal. Alternatively, referencing previously published data could help estimate the accuracy of the classification.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript describes methodology for spatial resolution of cortical organoids based on immunohistology of stacked sections. It is aimed at providing a comparative method to assess organoids using a more standard equipment. The work is described clearly and results demonstrate feasibility of the approach.  It may provide a useful method towards a more standardizable means for assessing 3-dimensional cortical, and potentially other types on organoids of variable sizes. However, it would be improving the manuscript if the following points could be addressed by the authors:

1. Introduction: Although a few methods have been mentioned for 3-dimenstional image analysis of cell composition in organoids, the increasing use of spatial single cell analysis, or of 3D - proteomics based methods should also be mentioned.

2. Results and Discussion: The results shown in Figures 5 and 6 are interesting and are adequately discussed. However, a few points need to e clarified, or added to the figures/discussion:

- Figure 5: The reduced non-viable core was attributed to reduced cell density at month 6. However, was a correlation be performed between overall size of organoid and size of non-viable cell core, oder cell number? That data should be added.

- Figure 6C/D: There is a cluster of very low MAP2 positive cells in both months 4 and 6. Is it possible to define the region where these cells are localized in the organoid? If not, why is this not possible? It other words, is it possible to identify clusters of lower or higher density of specific cell types, or cell densities within an organoid, e.g. in the periphery, or radial layer (if present)?

- It should be discussed what kind of benchmark could be derived from this method, and how this can be used between labs (or is feasibility focused on only in-house application)?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf