Nanomedicine, that is, the use of nanoparticles for medical purposes, is a rapidly expanding field, particularly for the treatment of tumors. It could allow for targeted and localized action, unlike conventional treatment methods such as radiotherapy or chemotherapy. One of the major obstacles to the clinical application of nanomedicine lies in the fact that nanoparticles struggle to penetrate deep into tumors, which prevents effective treatment. To investigate this issue, tumor models are used, such as the multicellular spheroid. The spheroid offers the advantage of possessing a three-dimensional architecture comparable to that of a tumor, unlike conventional 2D cultures. Moreover, microfluidics proves to be a versatile tool that enables high-throughput nanoparticle penetration assays, a necessary characteristic given the numerous parameters likely to alter this penetration. In particular, microfluidics allows precise control of the physicochemical environment surrounding the tumor models used. The mechanical constraints applied within a chip can be representative of characteristics encountered in vivo, such as the pressure induced by tumor growth or the shear stress generated by blood circulation. Among the factors influencing penetration efficiency are, on the one hand, the properties of the nanoparticles used, such as their size, charge, shape, or even their type. On the other hand, the in vitro models used can also influence penetration efficiency. Although the spheroid has its advantages, it does not reproduce the tumor microenvironment, which is composed of tumor-associated cell types such as stromal cells (fibroblasts) and immune cells (dendritic cells, macrophages), as well as an extracellular matrix rich in collagen, hyaluronic acid, and proteoglycans. All these components of the tumor microenvironment can alter nanoparticle penetration. During this thesis, we developed a microfluidic chip capable of immobilizing five spheroids in parallel while exposing them to physicochemical perturbations. The immobilization of the spheroids within the chip relies on the principle of micropipette aspiration. This feature makes it possible to apply mechanical stress to the spheroids through aspiration, while exposing them to a change of medium that generates shear stress and also enables their exposure to nanoparticles. The monitoring of nanoparticle exposure and penetration is carried out by confocal microscopy. Regarding the system used, we chose to study the penetration of cubosomes in spheroids derived from the A338 RFP (murine pancreatic cancer) and S180 GFP (murine sarcoma) cell lines. Cubosomes are self-assembled lipid nanoparticles with a cubic crystalline structure, capable of encapsulating many types of compounds (hydrophilic, hydrophobic, amphiphilic). The penetration of these innovative particles proved difficult in the models studied: nanoparticle accumulation remains localized in the outermost layers of the spheroids, whether with or without mechanical stress. This accumulation could be explained by a strong internalization of the cubosomes by individual cells; once internalized, the particles do not appear to be able to exit the cells. Nevertheless, the microfluidic chip developed proves to be a versatile tool for studying nanoparticle penetration in model biological objects.