Clinical decision support system for early detection and diagnosis of dementia

Bennasar, Mohamed 2014. Clinical decision support system for early detection and diagnosis of dementia. PhD Thesis, Cardiff University. Item availability restricted.

Preview	PDF - Accepted Post-Print Version Download (5MB) | Preview
	PDF - Supplemental Material Restricted to Repository staff only Download (81kB)

Abstract

Dementia is a syndrome caused by a chronic or progressive disease of the brain, which affects memory, orientation, thinking, calculation, learning ability and language. Until recently, early diagnosis of dementia was not a high priority, since the related diseases were considered untreatable and irreversible. However, more effective treatments are becoming available, which can slow the progress of dementia if they are used in the early stages of the disease. Therefore, early diagnosis is becoming more important. The Clock Drawing Test (CDT) and Mini Mental State Examination (MMSE) are well-known cognitive assessment tests. A known obstacle to the wider usage of the CDT assessments is the scoring and interpretation of the results. This thesis introduces a novel diagnostic Clinical Decision Support System (CDSS) based on CDT which can help in the diagnosis of three stages of dementia. It also introduces the advanced methods developed for the interpretation and analysis of CDTs. The data used in this research consist of 604 clock drawings produced by dementia patients and healthy individuals. A comprehensive catalogue of 47 visual features within CDT drawings is proposed to enhance the sensitivity of the CDT in diagnosing the early stages of dementia. These features are selected following a comprehensive analysis of the available data and the most common CDT scoring systems reported in the medical literature. These features are used to build a new digitised dataset necessary for training and validating the proposed CDSS. In this thesis, a novel feature selection method is proposed for the study of CDT feature significance and to define the most important features in diagnosing dementia. iii A new framework is also introduced to analyse the temporal changes in the CDT features corresponding to the progress of dementia over time, and to define the first onset symptoms. The proposed CDSS is designed to differentiate between four cognitive function statuses: (i) normal; (ii) mild cognitive impairment or mild dementia; (iii) moderate or severe dementia; and (vi) functional. This represents a new application of the CDT, as it was previously used only to detect the positive dementia cases. Diagnosing mild cognitive impairment or early stage dementia using CDT as a standalone tool is a very challenging task. To address this, a novel cascade classifier is proposed, which benefits from combining CDT and MMSE to enhance the overall performance of the system. The proposed CDSS diagnoses the CDT drawings and places them into one of three cognitive statuses (normal or functional, mild cognitive impairment or mild dementia, and moderate or severe dementia) with an accuracy of 78.34 %. Moreover, the proposed CDSS can distinguish between the normal and the abnormal cases with accuracy of 89.54 %. The achieved results are good and outperform most of CDT scoring systems in discriminating between normal and abnormal cases as reported in existing literature. Moreover, the system shows a good performance in diagnosing the CDT drawings into one of the three cognitive statuses, even comparing well with the performance of dementia specialists. The research has been granted ethical approval from the South East Wales Research Ethics Committee to employ anonymised copies of clock drawings and copies of Mini Mental State Examination made by patients during their examination by the memory team in Llandough hospital, Cardiff

Item Type:	Thesis (PhD)
Status:	Unpublished
Schools:	Engineering
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science T Technology > TA Engineering (General). Civil engineering (General)
Uncontrolled Keywords:	Dementia; Feature selection; Cascade classification; Dimensionality reduction; Clinical decision support system; Mild cognitive impairment.
Date of First Compliant Deposit:	30 March 2016
Last Modified:	15 Oct 2020 01:35
URI:	https://orca.cardiff.ac.uk/id/eprint/73073

Citation Data

Actions (repository staff only)

Edit Item