Farea Shaaf, Zakarya Farea Ghanem (2023) Convolutional neural network model for left ventricle segmentation to detect myocardial infarction. Doctoral thesis, Universiti Tun Hussein Onn Malaysia.

Preview	Text 24p ZAKARYA FAREA GHANEM FAREA SHAAF.pdf Download (640kB) | Preview
	Text (Copyright Declaration) ZAKARYA FAREA GHANEM FAREA SHAAF COPYRIGHT DECLARATION.pdf Restricted to Repository staff only Download (303kB) | Request a copy
	Text (Full Text) ZAKARYA FAREA GHANEM FAREA SHAAF WATERMARK.pdf Restricted to Registered users only Download (8MB) | Request a copy

Abstract

The quantification of the left ventricle (LV) by cardiac short-axis magnetic resonance images (MRI) is critical in the diagnosis of cardiovascular diseases. The LV clinical metrics are frequently extracted based on the LV segmentation and localization from short-axis MRI images. Manual LV segmentation and localization is tedious and time-consuming task for medical experts to diagnose cardiac pathologies. Therefore, a fully automated LV segmentation and localization technique is required to assist medical experts in working more efficiently. This study proposed a region-based convolutional network (Faster R-CNN) for the localization of LV from short-axis cardiac MRI images using a region proposal network (RPN) integrated with deep feature classification and regression. In addition, a fully convolutional network (FCN) architecture for automatic LV segmentation from short-axis MRI images was proposed. Several experiments were conducted in the training phase to compare the performance of the network and the U-Net model. The segmentation models were trained and tested on a public dataset, namely the evaluation of myocardial infarction from the delayed-enhancement cardiac MRI (EMIDEC) dataset. The dice metric, Jaccard index, sensitivity, and specificity were used to evaluate the segmentation network’s performance, with values of 0.93, 0.87, 0.98, and 0.94, respectively. Based on the experimental results, the proposed network outperforms the standard U-Net model and is an advanced fully automated method in terms of segmentation performance. The localization model was effective, with accuracy, precision, recall, and F1 score values of 0.91, 0.94, 0.95, and 0.95, respectively. This model also allows the cropping of the detected area of LV, which is vital to reduce the computational cost and time during segmentation and classification procedures. The proposed methods could be improved to be applicable in clinical practice for doctors to diagnose cardiac diseases from cardiac short-axis MRI images

Actions (login required)

View Item