Classification of High-Resolution Manometry Data According to Videofluoroscopic Parameters Using Pattern Recognition

Abstract

Objective

To determine if pattern recognition techniques applied to high-resolution manometry (HRM) spatiotemporal plots of the pharyngeal swallow can identify features of disordered swallowing reported on the Modified Barium Swallow Impairment Profile (MBSImP).

Study Design

Case series evaluating new method of data analysis.

Setting

University hospital.

Subjects and Methods

Simultaneous HRM and videofluoroscopy was performed on 30 subjects (335 swallows) with dysphagia. Videofluoroscopic studies were scored according to the MBSImP guidelines while HRM plots were analyzed using a novel program. Pattern recognition using a multilayer perceptron artificial neural network (ANN) was performed to determine if 7 pharyngeal components of the MBSImP as well as penetration/aspiration status could be identified from the HRM plot alone. Receiver operating characteristic (ROC) analysis was also performed.

Results

MBSImP parameters were identified correctly as normal or disordered at an average rate of approximately 91% (area under the ROC curve ranged from 0.902 to 0.981). Classifications incorporating two MBSImP parameters resulted in classification accuracies over 93% (area under the ROC curve ranged from 0.963 to 0.989).

Conclusion

Pattern recognition coupled with multiparameter quantitative analysis of HRM spatiotemporal plots can be used to identify swallowing abnormalities, which are currently assessed using videofluoroscopy. The ability to provide quantitative, functional data at the bedside while avoiding radiation exposure makes HRM an appealing tool to supplement and, at times, replace traditional videofluoroscopic studies.