Automatic texture and intensity based ovarian classification.

Detection of ovarian follicle/cysts and monitoring its growth is vital in infertility treatment in women. Ultrasound imaging technique is used for recognition of ovarian follicles and cysts. Profound variation in location, shape, size and color is seen in follicles. Human interpretation of follicles gives a chance for misinterpretation and false diagnosis. Follicle recognition becomes a challenging task due to the non-homogenous nature of the follicles and presence of speckle noise. To overcome this problem, computer assisted recognition of ovarian follicle and cysts followed by ovarian classification were proposed. Discrete wavelet transform (dwt) was used for despeckling. Texture and intensity based segmentation methods were used for automatic recognition. Classification of ovary was done based on ovarian morphology. This novel method serves as a decision support system for the medical expert. The efficiency of the proposed texture and intensity based ovarian classification (TIOC) method was demonstrated using various performance indices like sensitivity, specificity, accuracy, precision, Mathew's correlation coefficient and receiver operating characteristic curve. The resultant images obtained from the TIOC method was compared with the control images and existing methods for validation.