Background: The growth and drug response of tumors are influenced by their stromal composition, both in vivo and 3D-cell culture models. Cell-type inherent features as well as mutual relationships between the different cell types in a tumor might affect drug susceptibility of the tumor as a whole and/or of its cell populations. However, a lack of single-cell procedures with sufficient detail has hampered the automated observation of cell-type-specific effects in three-dimensional stroma-tumor cell co-cultures. Methods: Here, we developed a high-content pipeline ranging from the setup of novel tumor-fibroblast spheroid co-cultures over optical tissue clearing, whole mount staining, and 3D confocal microscopy to optimized 3D-image segmentation and a 3D-deep-learning model to automate the analysis of a range of cell-type-specific processes, such as cell proliferation, apoptosis, necrosis, drug susceptibility, nuclear morphology, and cell density. Results: This demonstrated that co-cultures of KP-4 tumor cells with CCD-1137Sk fibroblasts exhibited a growth advantage compared to tumor cell mono-cultures, resulting in higher cell counts following cytostatic treatments with paclitaxel and doxorubicin. However, cell-type-specific single-cell analysis revealed that this apparent benefit of co-cultures was due to a higher resilience of fibroblasts against the drugs and did not indicate a higher drug resistance of the KP-4 cancer cells during co-culture. Conversely, cancer cells were partially even more susceptible in the presence of fibroblasts than in mono-cultures. Conclusion: In summary, this underlines that a novel cell-type-specific single-cell analysis method can reveal critical insights regarding the mechanism of action of drug substances in three-dimensional cell culture models.

Advances in imaging of microscopic structures are supported and complemented by adaptive visualization tools. These tools enable researchers to precisely capture and analyze complex three-dimensional structures of different kinds such as crystals, microchannels and electronic or biological material. In this contribution, we focus on 3D cell cultures. The new possibilities can play a particularly important role in biomedical research, especially here in the study of 3D cell cultures such as spheroids in the field of histology. By applying advanced imaging techniques, detailed information about the spatial arrangement and interactions between cells can be obtained. These insights help to gain a better understanding of cellular organization and function and have potential implications for the development of new therapies and drugs. In this context, this study presents a multi-modal light sheet microscope designed for the detection of elastic and inelastic light scattering, particularly Rayleigh scattering as well as the Stokes Raman effect and fluorescence for imaging purposes. By combining multiple modalities and stitching their individual results, three-dimensional objects are created combining complementary information for greater insight into spatial and molecular information. The individual components of the microscope are specifically selected to this end. Both Rayleigh and Stokes Raman scattering are inherent molecule properties and accordingly facilitate marker-free imaging. Consequently, altering influences on the sample by external factors are minimized. Furthermore, this article will give an outlook on possible future applications of the prototype microscope.

Biomedical research increasingly relies on 3D cell culture models and AI-based analysis can potentially facilitate a detailed and accurate feature extraction on a single-cell level. However, this requires for a precise segmentation of 3D cell datasets, which in turn demands high-quality ground truth for training. Manual annotation, the gold standard for ground truth data, is too time-consuming and thus not feasible for the generation of large 3D training datasets. To address this, we present a novel framework for generating 3D training data, which integrates biophysical modeling for realistic cell shape and alignment. Our approach allows the in silico generation of coherent membrane and nuclei signals, that enable the training of segmentation models utilizing both channels for improved performance. Furthermore, we present a new GAN training scheme that generates not only image data but also matching labels. Quantitative evaluation shows superior performance of biophysical motivated synthetic training data, even outperforming manual annotation and pretrained models. This underscores the potential of incorporating biophysical modeling for enhancing synthetic training data quality.

Deep learning-based segmentation and classification are crucial to large-scale biomedical imaging, particularly for 3D data, where manual analysis is impractical. Although many methods exist, selecting suitable models and tuning parameters remains a major bottleneck in practice. Hence, we introduce the 3D data Analysis Optimization Pipeline, a method designed to facilitate the design and parameterization of segmentation and classification using two Bayesian Optimization stages. First, the pipeline selects a segmentation model and optimizes postprocessing parameters using a domain-adapted syntactic benchmark dataset. To ensure a concise evaluation of segmentation performance, we introduce a segmentation quality metric that serves as the objective function. Second, the pipeline optimizes design choices of a classifier, such as encoder and classifier head architectures, incorporation of prior knowledge, and pretraining strategies. To reduce manual annotation effort, this stage includes an assisted class-annotation workflow that extracts predicted instances from the segmentation results and sequentially presents them to the operator, eliminating the need for manual tracking. In four case studies, the 3D data Analysis Optimization Pipeline efficiently identifies effective model and parameter configurations for individual datasets.