The recognition of severe thoracic aortic dissection using conventional chest radiography in conjunction with a whale optimized bilateral residual convolutional neural network

Abstract

Author(s): Vikram Shete, Anand Gudur, Malathi H, Nipun Setia, Devanshu J. Patel, Sudhanshu Dev

Findings might help decide on a treatment that involves the individual being treated, a haematologist, a doctor who treats pregnant women, and a neonatologist. Severe Thoracic Aortic Dissection (STAD) is a potentially fatal disorder caused by bleeding from the damaged inner layer of the aorta, which divides the adventitial and intimal layers. The diagnosis of this illness is challenging. Although they are often used for early screening or diagnosis, Chest X-Rays (CXR) are not very accurate in diagnosing conditions. Recently, Deep Learning (DL) has been used successfully to several medical image processing applications. We suggest to develop the accuracy of the STAD diagnosis based on chest x-rays by utilizing DL approaches. With the use of the Whale Optimized Bilateral Residual Convolutional Neural Network (WOBRCNN), the major thoracic aortic dissection was identified. The results showed that the WO-BRCNN accuracy was 96.19%, with 94.93% precision, 97.89% recall, 95.71% F1-score, and 93.42% specificity. This type of studies might be useful for the early identification of typical descending aorta dissection associated with diverse carcinomas like fibrosarcoma’s, leiomyosarcomas, histiocytomas and myxomas. Further investigation is required to enhance the precision of diagnosis by aorta segmentation.

Share this article