Detection of corpora lutea in ovary stained tissue with Deep Learning

In the field of health sciences, the study of the various organic tissues in living beings is of great importance. This can give indications of the physiological and/or pathological state, since many tissue investigations are carried out for the purposes of diagnosing and treating various pathologies as early as possible. However, such investigations are complex processes, often involving manual methods and traditional techniques requiring many working hours in order to obtain conclusive results. This is the case with ovarian tissue histology, where the results of a morphometric evaluation are of great importance as it is the basis on which ovarian failure is identified, thus allowing access to diagnosis and treatment. However, thanks to developments in computer science, these processes are increasingly being optimized through the integration of intelligent systems that perform tasks autonomously, improving working times while delivering highly efficient and accurate results. These intelligent models are machine learning algorithms trained with sets of representative samples of a study population. These sets of samples are called datasets and are the most critical and crucial stage in process automation. This is why we have created a dataset of images of ovarian preparations with labels for a structure called corporea lutea. We have also trained a machine learning algorithm with this dataset as an input in a supervised learning to, finally, provide a useful tool for research.