Rehman, Mujeeb Ur ORCID: https://orcid.org/0000-0002-4228-385X, Najam, Shaheryar, Khalid, Sohail, Shafique, Arslan ORCID: https://orcid.org/0000-0001-7495-2248, Alqahtani, Fehaid ORCID: https://orcid.org/0000-0001-9564-6653, Baothman, Fatmah ORCID: https://orcid.org/0000-0003-0344-1007, Shah, Syed Yaseen, Abbasi, Qammer H. ORCID: https://orcid.org/0000-0002-7097-9969, Imran, Muhammad Ali ORCID: https://orcid.org/0000-0003-4743-9136 and Ahmad, Jawad ORCID: https://orcid.org/0000-0001-6289-8248 (2021) Infrared Sensing Based Non-Invasive Initial Diagnosis of Chronic Liver Disease Using Ensemble Learning. IEEE Sensors Journal, 21 (17). pp. 19395-19406.

Full text not available from this repository.

Abstract

The liver is a vital human body organ and its functionality can be degraded by several diseases such as hepatitis, fatty liver disease, and liver cancer and so forth. Hence, the early diagnosis of liver diseases is extremely crucial for saving human lives. With the rapid development of multimedia technology, it is now possible to design and implement a non-invasive system that can chronic liver diseases. For this purpose, machine learning and Artificial Intelligence (AI) have been used within the past few years. In this regard, digital image processing supported by AI methods has been implemented in the diagnosis of diseases that also showed high reliability. Therefore, in this paper, an iris feature-based non-invasive technique is proposed by incorporating a novel machine-learning algorithm. The experimental setup involved data set for the models' training included 879 subjects from Pakistan, of which 453 subjects have chronic liver disease and 426 are healthy. The iris images were collected using an infrared camera that consists of a lens, a thermal sensor and digital electronics processing. The lens focuses on the infrared energy on the sensor, using distinctive forms of features twenty-two physiological and thirty-three iris features. The designed classification model for a non-invasive system combined eleven different classifiers and used cross-validation techniques for comparing the results. The overall performance of the model was analyzed using five parameters: accuracy, precision, F-score, specificity, and sensitivity. The results confirmed that the proposed non-invasive model is capable of predicting chronic liver diseases with 98% of accuracy.

Item Type:	Article
Status:	Published
DOI:	10.1109/JSEN.2021.3091471
School/Department:	School of Science, Technology and Health
URI:	https://ray.yorksj.ac.uk/id/eprint/8169

University Staff: Request a correction | RaY Editors: Update this record