Computer-Assisted Image Analysis of Human Ovarian Follicles: Imaging Physiologic Selection

Abstract

Antral ovarian folliculogenesis involves recruitment of a cohort of small follicles, physiological selection of a dominant follicle, and ovulation. The mechanism of selection has not been precisely determined. Identification of the timing of preovulatory selection is a key component in understanding natural and peri-menopausal ovarian function, ovarian suppression for contraception, and improvement of ovarian stimulation protocols. Morphologic characteristics obtained by ultrasonography cannot be precisely quantitated by the human eye. Computer-assisted image analysis overcomes subjective human evaluation of ultrasonographic images.

The objectives of this research were to assess ultrasound image attributes of human dominant (DF) and 1st subordinate (SF1) ovarian follicles during natural menstrual cycles and following discontinuation of conventional and continuous oral contraceptives (OC). We utilized sophisticated computer algorithms to elucidate an association between image attributes and physiologic status of follicles. Transvaginal ultrasonographic images obtained in 2 previous studies were used to quantify changes that occur in ovarian follicles.

We detected quantitative differences between the dominant and largest subordinate follicles of ovulatory and major anovulatory follicular waves, as well as during the first wave following OC discontinuation. Differences in ultrasonographic image attributes were associated with the physiological status of follicles. Evidence of follicular dominance in follicles which develop during major ovulatory waves or following OC discontinuation can be detected prior to the time of selection manifest by differences in dominant and subordinate follicle diameters. In addition, differences in quantitative image attributes were detected between ovulatory and anovulatory DF. Follicles that develop following conventional and continuous OC administration schemes exhibit the same image characteristics.

Further research is necessary to elucidate the exact correlation of follicle image attributes during all stages of development with histological characteristics, prediction of the timing of DF selection and the effects of different OC formulations on follicle development during and following OC cessation. Computer-assisted image analysis of ultrasound images has the potential to develop into a diagnostic, prognostic, and research tool for the in vivo evaluation of ovarian physiology and pathology and elucidate biologically important times such as physiologic selection, ovulation of DF and characterization of abnormal follicles (i.e., follicular cysts, luteinized unovulated follicles).