Towards Glaucoma Detection Using Intraocular Pressure Monitoring Christophe Gisler *† , Antonio Ridi *† , Mil` ene Fauquex ‡ , Dominique Genoud ‡ and Jean Hennebert *† * Department of Informatics, Faculty of Science University of Fribourg Boulevard de P´ erolles 90, CH-1700 Fribourg, Switzerland Email: christophe.gisler@unifr.ch † Institue of Complex Systems University of Applied Sciences Western Switzerland, Fribourg Boulevard de P´ erolles 80, CH-1705 Fribourg, Switzerland Email: jean.hennebert@hefr.ch ‡ Institut Informatique de Gestion University of Applied Sciences Western Switzerland, Valais Techno-Pˆ ole 3, CH-3960 Sierre, Switzerland Email: dominique.genoud@hevs.ch Abstract—Diagnosing the glaucoma is a very difficult task for healthcare professionals. High intraocular pressure (IOP) remains the main treatable symptom of this degenerative disease which leads to blindness. Nowadays, new types of wearable sensors, such as the contact lens sensor Triggerfish R  , provide an automated recording of 24-hour profile of ocular dimensional changes related to IOP. Through several clinical studies, more and more IOP-related profiles have been recorded by those sensors and made available for elaborating data-driven experiments. The objective of such experiments is to analyse and detect IOP pattern differences between ill and healthy subjects. The potential is to provide medical doctors with analysis and detection tools allowing them to better diagnose and treat glaucoma. In this paper we present the methodologies, signal processing and machine learning algorithms elaborated in the task of automated detection of glaucomatous IOP- related profiles within a set of 100 24-hour recordings. As first convincing results, we obtained a classification ROC AUC of 81.5%. Keywords-Glaucoma diagnosis; biomedical signal processing; machine learning; I. I NTRODUCTION According to the World Health Organization (WHO), glaucoma is the second most frequent cause of blindness in the world. More than 80 million people worldwide suffer from this asymptomatic and painless disease of the eye. Development of glaucoma usually comes along with an increase of intraocular pressure (IOP) which gradually damages the optic nerve at the back of the eye. This leads to a progressive and irreversible loss of vision for affected patients. Thus, a high IOP is a major risk factor of glaucoma. However, IOP is subject to variations depending on indi- viduals’ body circadian cycles and activities, such as phys- ical activity, food intake, stress, emotions and sleep periods throughout day and night. Measuring it punctually with usual tonometers is not sufficient to capture its continuous changes throughout a 24 hour cycle. Understanding how IOP evolves and detecting patterns that truly characterize a glaucomatous profile has become a necessity for healthcare professionals to detect glaucoma early and prevent the loss of vision. The contact lens sensor (CLS) Triggerfish R  , developed by the swiss company Sensimed [1], is a sensor that provides such an automated recording of 24-hour profile of ocular dimensional changes related to IOP. This company is now leading clinical studies over the world to record IOP-related profiles and build a growing database dedicated to research on glaucoma. In this paper, we report on a research work done in collaboration with Sensimed for the application of signal analysis and machine learning techniques on 24-hour profiles data recorded from healthy and ill subjects suffering from glaucoma. This paper is organized as follows. Section II gives an overview of the existing related work. Section III gives a description of the data acquisition system, i.e. the CLS Triggerfish R  and the constituted database of IOP- related profiles. Section IV presents the statistical and physiological features which were extracted. In Section V, we presents the machine learning techniques as well as the evaluation protocol that were applied to perform the automated detection of glaucomatous profiles. Finally, in Section VI, we present and discuss the obtained results. II. RELATED WORK Machine learning (ML) applied to medical data for diagnosis diseases is not a new topic. Different pattern recognition techniques have been applied for the detection of various diseases like Alzheimer’s, Parkinson’s, diabetes, just to name a few. Cruz et al. [2] made a survey on ML International Conference of Soft Computing and Pattern Recognition 978-1-4799-5934-1/14/$31.00 ©2014 IEEE 255