Predictive value of CT texture analysis in indeterminate ovarian mass- a multi-centre validation study.

Professor Lee, Elaine Yuen Phin   (Principal Investigator (PI))

Professor Ip Pun Ching Philip   (Co-Investigator)

Professor Tse Ka Yu   (Co-Investigator)

24

2022-01-22

996760

Predictive value of CT texture analysis in indeterminate ovarian mass- a multi-centre validation study.

Benign, Computed tomography, Malignant, Multi-centre, Ovarian mass, Texture analysis

Others - Medicine, Dentistry and Health

08192106

Health and Medical Research Fund - Full Grant

2020

On-going

Objectives: (1) To test the ability of CT texture analysis (CTTA) in differentiating malignant from benign ovarian masses and validate the algorithm derived from the diagnostic model from the derivation cohort that incorporates CTTA, age and CA-125 in multi-centre study in Hong Kong. (2) To validate the accuracy of CTTA in histological classification of epithelial ovarian carcinoma (EOC). Hypotheses to be tested: CTTA is a quantitative analysis that measures the heterogeneity of the greyscale and pixels distribution. We hypothesise CTTA could differentiate malignant from benign ovarian masses with high accuracy, and offer histological classification in EOC. Design and subjects: This will be multi-centre collaborations among health-authority hospitals in Hong Kong with retrospective accrual of patients with ovarian masses on CT with histological diagnosis. Study instruments: Contrast-enhanced CT and texture analysis software. Interventions: Pre-surgical imaging. Main outcome measures: (1) The discriminative power of the diagnostic algorithm incorporating CTTA and clinical parameters in determining the nature of the ovarian mass.(2) The accuracy of CTTA in histological classification of EOC. Data analysis: CT features will be compared between the benign and early-stage ovarian carcinoma and to determine the ability to discriminate between the two entities. A devised diagnostic algorithm (CTTA and clinical parameters) will be validated in external cohorts. For EOC, random forest method will classify histological subtypes by incorporating CTTA and classification performance will be evaluated. Expected results: To characterise indeterminate ovarian mass and to classify histological subtypes in EOC.