Characterization of adnexal lesions using photoacoustic imaging to improve sonographic O-RADS risk assessment

Objective: To assess the impact of photoacoustic imaging (PAI) on the assessment of ovarian/adnexal lesion(s) of different risk categories using the sonographic ovarian-adnexal imaging-reporting-data system (O-RADS) in women undergoing planned oophorectomy. Method: This prospective study enrolled women with ovarian/adnexal lesion(s) suggestive of malignancy referred for oophorectomy. Participants underwent clinical ultrasound (US) examination followed by coregistered US and PAI prior to oophorectomy. Each ovarian/adnexal lesion was graded by two radiologists using the US O-RADS scale. PAI was used to compute relative total hemoglobin concentration (rHbT) and blood oxygenation saturation (%sO₂) colormaps in the region of interest. Lesions were categorized by histopathology into malignant ovarian/adnexal lesion, malignant Fallopian tube only and several benign categories, in order to assess the impact of incorporating PAI in the assessment of risk of malignancy with O-RADS. Malignant and benign histologic groups were compared with respect to rHbT and %sO₂ and logistic regression models were developed based on tumor marker CA125 alone, US-based O-RADS alone, PAI-based rHbT with %sO₂, and the combination of CA125, O-RADS, rHbT and %sO_2. Areas under the receiver-operating-characteristics curve (AUC) were used to compare the diagnostic performance of the models. Results: There were 93 lesions identified on imaging among 68 women (mean age, 52 (range, 21–79) years). Surgical pathology revealed 14 patients with malignant ovarian/adnexal lesion, two with malignant Fallopian tube only and 52 with benign findings. rHbT was significantly higher in malignant compared with benign lesions. %sO₂ was lower in malignant lesions, but the difference was not statistically significant for all benign categories. Feature analysis revealed that rHbT, CA125, O-RADS and %sO₂ were the most important predictors of malignancy. Logistic regression models revealed an AUC of 0.789 (95% CI, 0.626–0.953) for CA125 alone, AUC of 0.857 (95% CI, 0.733–0.981) for O-RADS only, AUC of 0.883 (95% CI, 0.760–1) for CA125 and O-RADS and an AUC of 0.900 (95% CI, 0.815–0.985) for rHbT and %sO₂ in the prediction of malignancy. A model utilizing all four predictors (CA125, O-RADS, rHbT and %sO₂) achieved superior performance, with an AUC of 0.970 (95% CI, 0.932–1), sensitivity of 100% and specificity of 82%. Conclusions: Incorporating the additional information provided by PAI-derived rHbT and %sO₂ improves significantly the performance of US-based O-RADS in the diagnosis of adnexal lesions.