Elasticity mapping for breast cancer diagnosis using tactile imaging and auxiliary sensor fusion

Student thesis: Doctoral Thesis

Abstract

Tactile Imaging (TI) is a technology utilising capacitive pressure sensors to image elasticity distributions within soft tissues such as the breast for cancer screening. TI aims to solve critical problems in the cancer screening pathway, particularly: low sensitivity of manual palpation, patient discomfort during X-ray mammography, and the poor quality of breast cancer referral forms between primary and secondary care facilities. TI is effective in identifying 'non-palpable', early-stage tumours, with basic differential ability that reduced unnecessary biopsies by 21% in repeated clinical studies. TI has its limitations, particularly: the measured hardness of a lesion is relative to the background hardness, and lesion location estimates are subjective and prone to operator error. TI can achieve more than simple visualisation of lesions and can act as an accurate differentiator and material analysis tool with further metric development and acknowledgement of error sensitivities when transferring from phantom to clinical trials. This thesis explores and develops two methods, specifically inertial measurement and IR vein imaging, for determining the breast background elasticity, and registering tactile maps for lesion localisation, based on fusion of tactile and auxiliary sensors. These sensors enhance the capabilities of TI, with background tissue elasticity determined with MAE < 4% over tissues in the range 9 kPa - 90 kPa and probe trajectory across the breast measured with an error ratio < 0.3%, independent of applied load, validated on silicone phantoms. A basic TI error model is also proposed, maintaining tactile sensor stability and accuracy with 1% settling times < 1.5s over a range of realistic operating conditions. These developments are designed to be easily implemented into commercial systems, through appropriate design, to maximise impact, providing a stable platform for accurate tissue measurements. This will allow clinical TI to further reduce benign referral rates in a cost-effective manner, by elasticity differentiation and lesion classification in future works.
Date of Award3 Sep 2021
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsUniversity of Strathclyde
SupervisorGordon Dobie (Supervisor) & Graeme West (Supervisor)

Cite this

'