B148: Optimising Ambulatory Chemotherapy Services: A System for Monitoring Drug Delivery in Elastomeric Pumps

Dahlia Salman1,Bruna Gomes1,Angelo Di Paolo1,David Wertheim1,Reem Kayyali1,Shereen Nabhani-Gebara1

1Kingston University, London, UK

Presenting date: Tuesday 3 November
Presenting time: 13.10-14.00


Ambulatory Chemotherapy (AC), using small portable elastomeric pumps, allows patients to receive chemotherapy outside the hospital. AC services have positively impacted patients’ quality of life, satisfaction, staff workload and NHS costs [1-2].

Despite those advantages, accuracy and reliability of elastomeric pumps have been a limiting factor associated with adoption of AC [3]. A recent two-phase audit, carried out by the authors at three gastrointestinal medical day units, showed that in 40% of disconnection cases flow was lower than expected with some patients arriving at hospital prior to complete delivery of the chemotherapy dose. 58% of these cases had an estimated volume remaining which exceeded 10mL (total infusion volume=120mL). Inaccurate flow and infusion duration could result in unpredictable end of infusion time and affect patient safety, treatment efficacy and patient outcome (if pumps are disconnected before the full dose has been infused). An automated elastomeric pump monitoring system may thus help to predict end of infusion time. The aim of this study was to develop image analysis software for monitoring elastomeric pump performance.


Software was developed using MATLAB (The MathWorks Inc., USA) to monitor the remaining volume in Baxter FOLFusor-SV2.5 pumps from images acquired with a mobile phone. 156 colour images (8 MP) were taken using an iPhone 4s (Apple Inc, USA). 


Two images were collected for each of 6 background colours and each of these images was analysed three times. With red as background colour, the mean (sd) difference between known and measured volumes was -1.14 (2.91) mL.


In conclusion, this study showed good agreement between the known and measured remaining volume. By using images taken with a minimum of 2 time points, prediction of end of infusion time should now be possible. In addition the system could help to identify infusion problems e.g. lack of flow due to blockage.