Teijiro Isokawa

Abstract Recent advances in prototissue construction have highlighted the importance of organizing multiple artificial cells with distinct function and intercellular communication capabilities. Previous approaches, including the use of optical tweezers, phoresis driven by the magneto-Archimedes effect, and 3D-printing, enabled the aggregation of liposome-based artificial cells, but they lacked precise spatial organization of individual artificial cells for large-scale production. Microfluidic devices (MFDs) offer potential for high-throughput prototissue fabrication but have faced challenges in achieving precise ordering of fragile liposomes with diameters exceeding 1 µm. In this study, we demonstrated an MFD incorporating U-shaped microstructures designed for liposome trapping that enables stochastic liposome capture and release, with this process being influenced by differences in the mechanical properties of the liposomal membrane. Based on these findings, we developed a two-step loading strategy, involving the sequential introduction of two liposome types, and successfully arranged them within the trapping microstructures of the MFD. Furthermore, DNA modification of the liposomes enhanced their spatial proximity within the MFD, enabling DNA hybridization and strand displacement reactions between the adjacent liposomal membranes. This MFD platform provides a statistical assay for constructing and analyzing minimal prototissues composed of two heterogeneous micrometer-sized liposomes.

Abstract While AI-based diagnostic support using whole slide images of pathological specimens has advanced in recent years, the development of systems that work with small biopsy specimens remains underexplored. In this study, we evaluated 1.5-mm-diameter colorectal cancer tissue microarray cores to estimate histological grade, to predict the resection-assigned pathological T (pT) category from microscopic morphology, and to estimate pathological TNM stage. We fine-tuned multiple ImageNet-pretrained convolutional neural network (CNN) backbones (ResNet, WideResNet, DenseNet, MobileNet, and EfficientNet) and evaluated individual models and ensemble learning. A maximum area under the curve for two-class classification of tumor histological grade was 0.991 across CNNs. A Nemenyi multiple-comparison test indicated significant differences in performance across tumor grading task definitions. Furthermore, the models predicted early versus advanced pT category (resection-assigned) with a maximum AUC of 0.888, and the Nemenyi test again indicated differences across task definitions. Early versus advanced pathological stage was separated with a maximum AUC of 0.721. Gradient-weighted class activation mapping visualization confirmed that the CNN model focused on histological features characteristic of Grade 3 tumors, such as tumor budding and poorly differentiated clusters, validating its classification performance. These results suggest that pathologist-selected biopsy-scale region of interests (ROIs) enable robust grading and pT-associated risk stratification with modest computational requirements, although they do not directly measure anatomical depth of invasion. They also suggest the potential of CNN-based diagnostic support systems as practical tools for treatment planning that can be implemented with limited computational resources, including constrained image data and storage capacity.