University of West Attica
Αυτοματοποιημένη ογκομέτρηση εγκεφάλου: σύγκριση λογισμικών και αξιολόγηση πρακτικής εφαρμογής τους
Abstract
dc:descriptionObjective and Aim:Brain volume analysis derived from MRI is gaining an increasingly important role in neurological diagnosis, offering the potential to introduce quantitative objectivity into clinical evaluations that are otherwise reliable but inherently subjective. This study aims to investigate: (a) the reliability of automated volumetric software, and (b) the feasibility of its practical clinical application in the evaluation of dementia. To address the first objective, volumetric measurements of dementia related brain structures obtained with NeuroQuant and volBrain were compared. For the second objective, Scheltens’ Medial Temporal Lobe Atrophy (MTA) scale scores were correlated with hippocampal and cerebrum volumes measured by volBrain, in an effort to enhance diagnostic accuracy and patient care. Materials and Methods: A total of 162 participants were included (56 in the NeuroQuant–volBrain reliability analysis and 106 in the MTA–volumetry correlation study). The cohort consisted of healthy volunteers and patients with mild cognitive impairment or multiple sclerosis, all examined with standardized MRI protocols (with minor variations tailored to each study objective). Using NeuroQuant and volBrain, volumes of the intracranial cavity, putamen, thalamus, amygdala, whole brain, cerebellum, white matter, and hippocampus were extracted for comparison (56 participants). Paired t test was used to determine significant differences between methods. Inter method reliability was assessed using the intraclass correlation coefficient (ICC), and linear regression evaluated potential linear relationships. In the second objective group (106 participants), two radiologists independently scored medial temporal lobe atrophy using Scheltens’ MTA scale, while volBrain provided automated volumetric measurements. Inter rater reliability was assessed with ICC and Cohen’s kappa. Normality was evaluated with the Shapiro–Wilk test. Spearman’s correlation, Kruskal–Wallis, and Dunn’s post hoc tests were used to compare MTA scores with volumetric data. Receiver Operating Characteristic (ROC) analysis assessed diagnostic performance, and ordinal logistic regression modeled the probability of each Scheltens’ score based on relative hippocampal volume. Results: Significant differences were found between NeuroQuant and volBrain for all brain structures except the hippocampus. ICC demonstrated satisfactory to excellent reliability for most segments, except for the thalamus and amygdala, where reliability was poor. Linear correlation was observed in most cases. Radiologist agreement on MTA scoring was almost perfect. Scheltens’ scale showed strong correlations with hippocampal volumetric measurements but only weak correlations with whole cerebrum volumes. Automated volumetry demonstrated 82% sensitivity and 100% specificity in identifying atrophy. Ordinal logistic regression confirmed overall model significance. Conclusion: The substantial differences observed between software packages for most brain structures raise important concerns regarding the reliability of automated volumetry as a purely quantitative tool. However, strong linear correlations and generally good inter method reliability suggest that each software provides consistent qualitative information about relative brain structure size. This study indicates that relative hippocampal volume is a sensitive biomarker for atrophy assessment and supports the development of an advanced, quantitative medial temporal lobe atrophy scale, offering clinicians a more objective and precise diagnostic tool.
Degree
thesis:*- Grantor dc:publisher
- University of West Attica
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Koussis, Panagiotis
- Κούσης, Παναγιώτης
Subjects
dc:subject × 10Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/61804