Comparitive Study of Uterine Cyst Characterisation Using Glcm and Wavelet Transform

Abstract — 50-75 % of the women in their late 30-50 years suffer from uterine cysts, 30 % of them are of reproductive age. Out of the different kinds of uterine cysts formed, misdiagnosis often happens between the conditions of adenomyosis and uterine fibroids. Though both the diseased conditions are found to have similar symptoms, the treatment offered to them differs thereby making it prominent in proper differentiation. Statistical features resulting in accurate characterization of the diseased conditions were identified. The features were extracted from both GLCM and wavelet transformed domain of the collected dataset of ultrasound images consisting of 25 uterinemyoma and 17 adenomyoma. The efficiency of the suggested statistical features in exact classification of the conditions was compared. The results show an efficacious uterine cyst characterization takes place in wavelet domain to the comparative study performed in spatial domain. Better classification of uterine fibroids and adenomyoma occurs in wavelet domain. With higher order of statistical features considered, still more accurate classification can be obtained in wavelet domain. Characteristic differentiation of the conditions of uterine fibroids and adenomyoma based on the features extracted from the ultrasound images makes it advantageous and simple method of uterine cyst identification compared to the current surgical procedures.