Recently, many types of 3D culture systems have been developed to preserve the physicochemical environment and biological characteristics of the original tumors better than the conventional 2D monolayer culture system. There are various types of models belonging to this culture, such as the culture based on non-adherent and/or scaffold-free matrices to form the tumors. Agarose mold has been widely used to facilitate tissue spheroid assembly, as it is essentially non-biodegradable, bio-inert, biocompatible, low-cost, and low-attachment material that can promote cell spheroidization. As no studies have been carried out on the development of a fluorescent bicellular tumoroid mimicking ductal carcinoma in situ (DCIS) using human cell lines, our objective was to detail the practical approaches developed to generate this model, consisting of a continuous layer of myoepithelial cells (MECs) around a previously formed in situ breast tumor. The practical approaches developed to generate a bi-cellular tumoroid mimicking ductal carcinoma in situ (DCIS), consisting of a continuous layer of myoepithelial cells (MECs) around a previously formed in situ breast tumoroid. Firstly, the optimal steps and conditions of spheroids generation using a non-adherent agarose gel were described, in particular, the appropriate medium, seeding density of each cell type and incubation period. Next, a lentiviral transduction approach to achieve stable fluorescent protein expression (integrative system) was used to characterize the different cell lines and to track tumoroid generation through immunofluorescence, the organization of the two cell types was validated, specific merits and drawbacks were compared to lentiviral transduction. Two lentiviral vectors expressing either EGFP (Enhanced Green Fluorescent Protein) or m-Cherry (Red Fluorescent Protein) were used. Various rates of a multiplicity of infection (MOI) and multiple types of antibodies (anti-p63, anti-CK8, anti-Maspin, anti-Calponin) for immunofluorescence analysis were tested to determine the optimal conditions for each cell line. At MOI 40 (GFP) and MOI 5 (m-Cherry), the signals were almost homogeneously distributed in the cells which could then be used to generate the DCIS-like tumoroids. Images of the tumoroids in agarose molds were captured with a confocal microscope Micro Zeiss Cell Observer Spinning Disk or with IncuCyte (R) to follow the progress of the generation. Measurement of protumoral cytokines such as IL-6, IL8 and leptin confirmed their secretion in the supernatants, indicating that the properties of our cells were not altered. Finally the advantages and disadvantages of each fluorescent approach were discussed. This model could also be used for other solid malignancies to study the complex relationship between different cells such as tumor and myoepithelial cells in various microenvironments (inflammatory, adipose and tumor, obesity, etc.). Although, this new model is well established to monitor drug screening applications and perform pharmacokinetic and pharmacodynamic analyses.