Taylor, Paul Martin;
(1998)
Computer assisted decision making for image understanding in medicine.
Doctoral thesis (Ph.D.), University College London (United Kingdom).
![]() |
Text
Computer_assisted_decision_mak.pdf Download (12MB) |
Abstract
This thesis considers the different kinds of computer system which assist in the interpretation of medical images. These systems contain information represented in two very different ways: as images and as symbolic knowledge. If a decision aid is to provide access to all the information that might assist a radiologist, it must be able to employ information represented in images and as symbolic knowledge. Image processing is needed to provide the descriptions required by knowledge-based systems. Symbolic representations are needed to relate image data to the decisions radiologists take. This thesis sets out a design for decision aids which combine image data and symbolic representations. The approach is based on a model of decision making, a symbolic decision procedure which constructs arguments, or lines of reasoning, about possible solutions to a problem. An extension to the decision procedure provides a model of three generic tasks in image interpretation: detection, classification and measurement. The extended decision procedure is implemented as a program which allows the symbolic decison procedure to draw on information obtained from processing images. A generic architecture, based on the extended decision procedure, has been used in the implementation of two prototypes: one to assist in the interpretation of breast X-rays or mammograms and one to assist in the use of CT for the management of abdominal tumours. The mammography system has been evaluated as an aid to the differential diagnosis of microcalcifications. An ROC analysis showed that radiographers who had been trained to interpret mammograms performed better on a differential diagnosis task when they had access to the information supplied by the system: the information was based in part on a symbolic representation of facts about microcalcifications and in part on the processing of digitised mammograms.
Type: | Thesis (Doctoral) |
---|---|
Qualification: | Ph.D. |
Title: | Computer assisted decision making for image understanding in medicine |
Open access status: | An open access version is available from UCL Discovery |
Language: | English |
Additional information: | Thesis digitised by ProQuest. |
Keywords: | (UMI)AAI10016126; Health and environmental sciences; Decision making; Medicine |
URI: | https://discovery-pp.ucl.ac.uk/id/eprint/10102032 |
Archive Staff Only
![]() |
View Item |