Endocrine Abstracts

Objective: The introduction of imaging technologies into routine clinical practice has increased the pituitary incidentalomas cases detection by magnetic resonance imaging (MRI) of the brain. Identification of pituitary formations requires additional hormonal examination and dynamic MRI pituitary examination for differential diagnosis and treatment strategy for patients. An important clinical problem is to determine the malignancy, predictors of potential growth and hormonal activity of pituitary incidentalomas from the results of MRI studies.

Materials and methods: A preview of retrospective study patients with pituitary incidentalomas detected by pituitary MRI examination at Republican clinical hospital of medical rehabilitation in 2021–2022 years is presented. All patients underwent pituitary MRI without gadolinium using at the beginning. In doubtful cases, repeated MRI was performed with gadolinium. Sagittal and coronal scans were examined. A pituitary macroadenoma was diagnosed with a one size of 10 mm or more. Laboratory, clinical and MRI data were analyzed using a neural network. Our neural network architecture consists of two consecutive parts. Detection model presented as a one-stage object detection neural network (Retina Net) to produce bounding boxes of the pituitary. Diagnosis model: ResNet-34 used to predict probabilities of pituitary macro/microadenoma using output of previous neural network (segmented slices). Derived probabilities was combined with hormonal tests via Gradient Boosting on Decision Trees algorithm.

Results: We studied 310 patients with pituitary incidentalomas, average age 40.9±15.91 years. A pituitary macroadenoma was diagnosed in 24 (7.7%) patients, average age 50.8±14.09 years, male:female rate was 1:2 (male – 8 (33.3%), female – 16 (66.7%)). 286 patients with pituitary incidentalomas less 10 mm, average age 36.2±13.55 years. Pituitary macroadenoma was significantly more frequently diagnosed in older patients compared to younger ones (P<0.05). Pituitary incidentalomas less than 10 mm were more frequently diagnosed in young patients (P<0.05).

Conclusion: The results of the study indicate an increased prevalence of pituitary macroadenomas in women and associated with an increase in age. It is necessary to clarify the plan for long-term monitoring of incidentalomas in young people for timely detection of growth.