Label-free, High-Resolution 3D Imaging and Machine Learning Analysis of Intestinal Organoids via Low-Coherence Holotomography

We aim to establish a real-time, label-free imaging methods when monitoring biophysical changes in live organoids during development and in response to drugs. This protocol facilitates streamlined scale of imaging and non-invasive drug testing in live organoids supported by AI-driven mutation and quantitative texture selection for biomedical research. We plot to further integrate label-free 3D imaging and a year analysis for high resolution, non-invasive organoids studies in disease modeling and precise medicine.

To begin, aspirate the spent medium from each well of a 48-well plate containing the extracellular matrix. Add 200 microliters of cell recovery solution to each well and incubate at four degrees Celsius for 30 minutes. Using a pipette, gently collect the organoid suspension and transfer it into a microcentrifuge tube.

Centrifuge the tube at 150g for three minutes, then carefully remove the supernatant. To disassociate the pellet mechanically, resuspended in 200 microliters of culture medium and pipette the suspension up and down 20 to 30 times using a P200 pipette before centrifuging it for three minutes. After centrifugation and removal of the supernatant, add medium and fresh extracellular matrix, or ECM, in a one-to-four ratio to the pellet and gently pipette to mix thoroughly.

Dispense 15 microliters of the mixture per dome into each-well of a 48-well plate. Place the plate upside down in a 37 degrees Celsius, 5%carbon dioxide incubator for one hour to allow polymerization of the ECM. After polymerization, add 200 microliters of fresh culture medium to each well, and fill the empty outer wells with PBS.

To prepare the sample for imaging, dispense 15 microliters of organoid ECM dome onto a number 1.5 coverslip bottom imaging dish and incubate it at room temperature for one minute. Place the plate upside down in a 37 degrees Celsius, 5%carbon dioxide incubator for one hour. Then, gently add enough culture medium to fully submerge the organoids.

At five days post-passage, wash the sample two to three times with PBS immediately before imaging. Next, for imaging, turn on the environmental controller. The chamber controller unit will automatically set the temperature to 37 degrees Celsius and carbon dioxide to 5%Press the Door button to open the door.

Add water to form a thin layer inside the chamber well. Place the imaging dish in the vessel holder, insert it into the imaging chamber, and secure it using a pin to prevent movement. Close the door to prevent external light interference.

Now, launch the TomoStudio X software and log in. Click Start to open the main window, and then click Add Project in the top left corner and assign the experiment. Confirm that the correct medium type is selected for appropriate refractive index usage.

Click on the desired-well in the panel and then click Create at the top to register the well as a specimen. Then, click ROI Setup in the top right corner to define the region of interest in the dish. Once set, click Run Experiment in the bottom right corner to open the Image Acquisition window.

Click Load Vessel in the corner to display a bright field image. Adjust the Z position using the Z and Z buttons to bring the image into focus. In the Single Imaging tab, adjust the ROI size.

Capture organoids in a 160 micrometer by 160 micrometer field of view and acquire stacks up to 140 micrometers deep. Navigate to the Time Lapse Imaging tab to set up long-term imaging and set the desired duration and interval time. Click the Scan icon to capture the current ROI location.

Click the BF button to adjust the intensity and exposure values for bright-field imaging. Move the ROI box in the Preview panel to select the ROI. Once the desired ROI is selected, click Add Point at the bottom, the imaging point list will be created.

Now, click Acquire to begin imaging and to acquire the raw image data. Launch the HTX Processing Server by clicking the Desktop icon. Drag and drop raw image files to the HTX Processing Server.

Click Process to generate a TCF file from the raw image file. Launch TomoAnalysis Viewer by clicking the Desktop icon. Load the processed TCF files by dragging and dropping them into the viewer window.

Double-click a file thumbnail to open the ROI Tomogram. Examine the 2D view by navigating through the X, Y, Z planes using zoom, pan, and scroll controls. Click the MIP rendering view icon on the left to switch to 3D rendering mode.

Navigate the 3D view by rotating, zooming, and panning the image. For the machine learning-based image segmentation, export the TCF image file to HDF5 format to ensure the data is in a multidimensional format, compatible with ilastik. Open ilastik and navigate to New Project.

Select Pixel Classification under the Segmentation Workflow section and save the project in a designated folder. To load the HGF5 file, navigate to the Input Data tab, click Add New File, select the appropriate H5 dataset, and verify the correct image channel assignments. Now, navigate to the Feature Selection tab to choose features like color, intensity, edge, and texture to optimize segmentation.

In the Training tab, label the organoid and non-organoid regions using different-color brush strokes. Click Live Update to preview segmentation results and make adjustments, if necessary. Once done, go to the Prediction Export tab.

Select Simple Segmentation as the source to export labeled predictions and click Export All. Set the export format to H5 or TIFF based on the analysis requirements. For the quantitative analysis, open the Supplementary Coding File 2.

Designate the appropriate folder paths for the mask file and the corresponding TCF file within the script. Run the code to initiate quantitative analysis. The code will calculate the organoid volume, protein density, and total protein content for each dataset.

This figure illustrates the high-resolution, three-dimensional refractive index reconstructions used to visualize the overall morphology of small intestinal organoids. The rendered reconstructions reveal distinct structural differences between vehicle-treated and cisplatin-treated organoids across all three optical section depths. Cisplatin-treated organoids showed significantly higher volume, lower protein density, and higher total protein contents, compared to vehicle-treated organoids at 10 minutes post-treatment, indicating early structural swelling and altered protein composition.

Time lapse imaging over 24 hours showed that vehicle-treated organoids maintained structural integrity, while cisplatin-treated organoids experienced progressive structural degradation, including crypt collapse and increased cell dissociation, suggesting time-dependent cytotoxic damage. Quantitative tracking confirmed that vehicle-treated organoids increased in volume and protein contents over time, while cisplatin-treated organoids showed a time-dependent decline in both parameters, indicating that cisplatin suppressed growth and induced cellular degradation. Protein density remained stable in vehicle-treated organoids, but progressively decreased in cisplatin-treated organoids over the 24-hour period, suggesting a breakdown in cellular structure and increased extracellular space due to exfoliation.