Nunes, NP; (2017) Novel Approaches to Ultrasound Based Evaluation and Management of Adnexal Masses. Doctoral thesis , UCL (University College London).

Preview

Text Nunes_MD Thesis Nunes 2017 Final.pdf - Submitted Version Download (3MB) | Preview

Abstract

Hypothesis and Aim: This thesis aims to assess current ultrasound tools and protocols for the diagnosis and management of ovarian tumours. I hypothesized that the IOTA (International Ovarian Tumour Analysis) tools would have less diagnostic accuracy when performed by a less experienced operator and that the IOTA LR models would therefore not be useful as a triage tool but that while the ‘simple rules’ (SR) model may be useful as a triage tool as it is most similar to pattern recognition. I also hypothesized that the SR modified protocol would offer fewer operations to women with benign disease as compared with Royal College of Obstetricians and Gynaecologists [RCOG] protocol using the risk of malignancy index [RMI]. Objectives: The aim of this thesis was to investigate and explore recently developed and currently utilised ultrasound based tools as methods of evaluation and means of determining the management of adnexal tumours. The novel approach was to investigate the use of these tools when performed by an average (level II) ultrasound operator as compared with the experts who developed them and to determine the management for asymptomatic postmenopausal women with incidentally detected adnexal tumours. Studies: Determining the accuracy of LR1/LR2 and ‘Simple Rules’ (SR) by an average operator compared with pattern recognition and evaluating the performance 4 of LR1/LR2 and SR using two reference standards, (histology and follow-up ultrasound scans) as a triage tool. Performing a meta-analysis of the SR model. Comparing the performance of two management protocols (Royal College of Obstetricians and Gynaecologists [RCOG] using the risk of malignancy index [RMI] and an SR-based protocol) for the likely intervention rates in asymptomatic postmenopausal women, and conducting a randomised controlled trial to compare the performance of those two management protocols (the RMI/RCOG guidance and the SR-based protocol) for the actual intervention rates in asymptomatic postmenopausal women. Findings: LR2 (average operator) had a similar sensitivity but the specificity was significantly lower, LR1/LR2 showed higher sensitivities and significantly lower specificities compared to the experts and pattern recognition and LR1/LR2 can be used as a triage tool once you accept the greater false positive rate. SR (average operator) showed similar sensitivity but a poorer specificity. When indeterminate tumours were assumed to be malignant there was a significantly increased sensitivity, decreased specificity and decreased diagnostic accuracy versus when pattern recognition was used, the sensitivity and diagnostic accuracy were maintained with increased specificity. SR performed well for the diagnosis of malignancy (meta-analysis) and SR works well as a triage tool, With the RMI/RCOG protocol, the likely intervention rate was 6 x more compared with SR and the RMI/RCOG protocol offered women surgery 9 x more often than SR and the actual surgical rate was 2.6 x more than SR (RCT). Conclusions: The IOTA LR1/LR2 models performed fairly well when used pre-operatively or as a 5 triage test by a less experienced operator. Comparatively, the SR model worked very well when the rules were applicable but this leaves the problem of deciding what to do with the 25% of women for whose ovarian tumours, the rules are not applicable. This thesis confirmed that the SRMP protocol had fewer indeterminately defined tumours and therefore offered surgery to fewer women with benign tumours when compared with the RCOG/RMI protocol.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as JSONCSVXML

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item