Undiagnosed polycystic diabetes based ovarian syndrome (PCOS) [Diabetes] can lead to infertility in females. Using a non-invasive transvaginal ultrasound machine, doctors can get a better look at the ovaries to determine if a woman has PCOS. Diabetes based ovarian images are distinguished primarily by the number and size of follicles present. By counting the follicles and measuring their diameters manually, PCOS is diagnosed. There are many steps to this procedure and error prone. The various studies that have been carried out since 2019 are examined in this publication. Gaussian low pass filter (GLPF), homogeneous region increasing mean (HRGMF) and contourlet transform are just some of the pre-processing methods that are examined. The segmentation methods that are discussed include Watershed transformation, edge-based approach, optimal thresholding, scanline thresholding, and active contour methods. Similarly, some of the classification methods discussed include the K-NN, linear discriminant, fuzzy, and SVM classifiers. Deep learning methods are used in this paper to identify follicles and PCOS diagnoses in ovarian ultrasound pictures. A comprehensive statistical analysis is presented in this review, which was carried out by gathering data from 30 publications released between the years 2019 to 2021. Some of the previous performances of the work are noted and compared. Finally, this paper offered ideas for future research methods.