Retrieval with Clustering in a Case-Based Reasoning System for Radiotherapy Treatment Planning

Mini EURO Conference on Improving Healthcare: new challenges, new approaches

物理学;医药卫生

Khussainova, Gulmira^1 ; Petrovic, Sanja^1 ; Jagannathan, Rupa^1

Operations Management and Information System Division, Nottingham University Business School, University of Nottingham, Nottingham

NG8 1BB, United Kingdom^1

关键词: Artificial intelligence tools;    Case based reasoning systems;    Casebased reasonings (CBR);    Clustering methods;    Medical physicists;    Radiotherapy treatment planning;    Retrieval mechanisms;    Trial-and-error process;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/616/1/012013/pdf
DOI  :  10.1088/1742-6596/616/1/012013

学科分类：卫生学

来源: IOP

【 摘 要 】

Radiotherapy treatment planning aims to deliver a sufficient radiation dose to cancerous tumour cells while sparing healthy organs in the tumour surrounding area. This is a trial and error process highly dependent on the medical staff's experience and knowledge. Case-Based Reasoning (CBR) is an artificial intelligence tool that uses past experiences to solve new problems. A CBR system has been developed to facilitate radiotherapy treatment planning for brain cancer. Given a new patient case the existing CBR system retrieves a similar case from an archive of successfully treated patient cases with the suggested treatment plan. The next step requires adaptation of the retrieved treatment plan to meet the specific demands of the new case. The CBR system was tested by medical physicists for the new patient cases. It was discovered that some of the retrieved cases were not suitable and could not be adapted for the new cases. This motivated us to revise the retrieval mechanism of the existing CBR system by adding a clustering stage that clusters cases based on their tumour positions. A number of well-known clustering methods were investigated and employed in the retrieval mechanism. Results using real world brain cancer patient cases have shown that the success rate of the new CBR retrieval is higher than that of the original system.