Κατεύθυνση Ιατρική Φυσική-Ακτινοφυσική
Library of the School of Health Sciences

2024-04-30

2024

Gazetis Angelos-Sotirios

Ευστράτιος Καραβασίλης, Επίκουρος Καθηγητής, Τμήμα Ιατρικής, Δημοκρίτειο Πανεπιστήμιο Θράκης
Αδάμ Αδαμόπουλος, Καθηγητής, Τμήμα Ιατρικής, Δημοκρίτειο Πανεπιστήμιο Θράκης
Αθανασία Κοτίνη, Καθηγήτρια, Τμήμα Ιατρικής, Δημοκρίτειο Πανεπιστήμιο Θράκης

Ανάλυση υφής εικόνων Μαγνητικής Τομογραφίας για την εξαγωγή προβλεπτικών δεικτών της έκβασης ασθενών με ισχαιμικό εγκεφαλικό επεισόδιο

Greek

Texture analysis of MRI images for the extraction of predictive indicators of outcome in patients with ischemic stroke

Radiomics analysis is an important branch of medical imaging that focuses on the extraction and evaluation of multiple quantitative features from medical images. This analysis technique is also applied to CT and MRI images of people with different pathologies including ischemic stroke patients in order to provide timely, accurate diagnosis and personalized treatment. In this paper we applied this method to MRI images with the purpose of creating a model to predict the clinical status of patients with ischemic stroke. The imaging data collection was performed at the General University Hospital of Alexandroupolis using the Phillips manufactured 1.5T MRI scanner. Image post-processing was performed with ITK-Snap software for segmentation of the pathological area and Pyradiomics software for pre-processing and extraction of texture features. Potential linear correlations between the extracted radiomic features and the NIHSS clinical index in the whole sample were then explored. Because of the impossibility of model generation in the whole sample, 2 subgroups were created based on the median NIHSS output and the patients' mild or severe symptomatology (NIHSS threshold = 5) to test for linear correlation in the 2 subgroups. Statistical analyses were performed using SPSS software. Linear model fit control conditions were met only in the NOT MILD subgroup. The radiomic feature highlighted by the linear model was the LeastAxisLenght. The radiomic features that separated the subgroups of patients using the median NIHSS output value as the separation criterion were Maximum2DDiameterSlice, SurfaceVolumeRatio and InterquartileRange. In conclusion, the promising potential of texture analysis combined with clinical indicators for the prognosis and clinical outcome of patients with ischemic stroke is highlighted.

Health Sciences

Radiomics analysis, Ischemic stroke, Texture feature, Magnetic resonance image

No

0

Yes

63

87

File access is restricted only to the intranet of UoA.

https://pergamos.lib.uoa.gr/uoa/dl/object/3397811