[ 1 ] [ 2 ] [ 3 ] [ 4 ] [ 5 ] [ 6 ] [ 7 ] [ 8 ] [ 9 ] [ 10 ] [ 11 ] [ 12 ]

Journal of Information Science and Engineering, Vol. 38 No. 2, pp. 463-477

Automatic Classification of Uroflow Patterns via the Grading-based Approach

1Department of Computer Science and Engineering
National Sun Yat-sen University
Kaohsiung, 80424 Taiwan
E-mail: changyi@mail.cse.nsysu.edu.tw; chichihlin1991@gmail.com

2Division of Urology
Taipei Tzu Chi Hospital, and Buddhist Tzu Chi University
New Taipei City, 23142 Taiwan
E-mail: urolyang@gmail.com

3Department of Urology, College of Medicine
Chang Gung University
Kaohsiung, 33302 Taiwan
E-mail: chuang82@ms26.hinet.net

4Department of Information Communication
Asia University
Taichung, 41354 Taiwan
E-mail: shenjh@asia.edu.tw

5Department of Medical Research
China Medical University Hospital, China Medical University
Taichung, 40447 Taiwan

6Department of Information Management
Cheng Shiu University
Kaohsiung, 833301 Taiwan
E-mail: lichiaen@gmail.com

Automatic classification of uroflowmetry curves into different patterns can help urologists to make an accurate diagnosis of the lower urinary track function in real time and increase the agreement of interpretation among urologists. In this paper, we propose a grading-based approach to the automatic classification of uroflowmetry curves by considering 87 cases of medical data, which are confirmed by the consensus of two highly experienced urologists. The interpretation of uroflowmetry is usually subjective and empirical. In this study, the same results identified by both urologists were 87 cases out of 160. To avoid the disadvantage of visual interpretation, our approach integrates the urologist’s experiences and different weights for different conditions, including several new conditions, the raising angle and the number of significant drops. Moreover, the objective view of classification is useful for teaching urologists to watch for voiding dysfunctions. From our experimental study with statistical analysis comparing the results of our approach with two urologists’ observations, we have shown that the agreement of normal / abnormal types is very good.

Keywords: classification, grading, urodynamic study, uroflow pattern, uroflowmetry curve

  Retrieve PDF document (JISE_202202_11.pdf)