The use of CAD (Computer-Aided Detection) tools has been gaining popularity with the evolution of artificial intelligence technologies. However, for CAD systems to yield satisfactory results, it is crucial to have initially treated and labeled datasets so that the CAD system can perform automatic identifications when completed. ImageJ, GIMP, Itk-SNAP, 3DSlicer, and InVesalius are tools used for image processing to enable their use in machine learning models. However, they do not manage or control access to DICOM image repositories. This paper presents a tool for sharing DICOM data among research groups, allowing multiple expert analyses on the data. This, in turn, improves the quality of the database. The tool utilizes a web interface developed with the Flutter Web framework, which communicates with a MySQL database through a backend developed in Python Flask. The database stores both the DICOM files and the segmented and labeled data, as well as the ground truth, making it available for consumption by a machine learning model. Thus, the tool facilitates the process of labeling and segmentation of DICOM images. As a result, we were able to perform manual and accurate segmentations of various regions of a DICOM image in a practical manner, as well as save the original image with the segmentation and ground truth overlay. Another advantage was the ability to manage information, as it allows multiple specialists to access this dataset and perform segmentation on one or more images.