Search

Miguel Angel Zapata,1 Dídac Royo-Fibla,1 Octavi Font,1 José Ignacio Vela,2,3 Ivanna Marcantonio,2,3 Eduardo Ulises Moya-Sánchez,4,5 Abraham Sánchez-Pérez,5 Darío Garcia-Gasulla,4 Ulises Cortés,4,6 Eduard Ayguadé,4,6 Jesus Labarta4,6 1Optretina, Barcelona, Spain; 2Ophthalmology Department, Hospital de la Santa Creu I de Sant Pau, Barcelona 08041, Spain; 3Universitat Autònoma de Barcelona (UAB), Campus de la UAB, Barcelona, Spain; 4Barcelona Supercomputing Center (BSC), Barcelona, Spain; 5Universidad Autónoma de Guadalajara - Postgrado en Ciencias Computacionales, Guadalajara, Mexico; 6Universitat Politècnica de Catalunya - BarcelonaTECH, Campus Nord, Barcelona, SpainCorrespondence: Miguel Angel ZapataOptretina, C/ Las Palmas 11, 08195 Sant Cugat del Vallès, Barcelona, SpainTel +34 655809682Email mazapata@optretina.comPurpose: To assess the performance of deep learning algorithms for different tasks in retinal fundus images: (1) detection of retinal fundus images versus optical coherence tomography (OCT) or other images, (2) evaluation of good quality retinal fundus images, (3) distinction between right eye (OD) and left eye (OS) retinal fundus images,(4) detection of age-related macular degeneration (AMD) and (5) detection of referable glaucomatous optic neuropathy (GON).Patients and Methods: Five algorithms were designed. Retrospective study from a database of 306,302 images, Optretina’s tagged dataset. Three different ophthalmologists, all retinal specialists, classified all images. The dataset was split per patient in a training (80%) and testing (20%) splits. Three different CNN architectures were employed, two of which were custom designed to minimize the number of parameters with minimal impact on its accuracy. Main outcome measure was area under the curve (AUC) with accuracy, sensitivity and specificity.Results: Determination of retinal fundus image had AUC of 0.979 with an accuracy of 96% (sensitivity 97.7%, specificity 92.4%). Determination of good quality retinal fundus image had AUC of 0.947, accuracy 91.8% (sensitivity 96.9%, specificity 81.8%). Algorithm for OD/OS had AUC 0.989, accuracy 97.4%. AMD had AUC of 0.936, accuracy 86.3% (sensitivity 90.2% specificity 82.5%), GON had AUC of 0.863, accuracy 80.2% (sensitivity 76.8%, specificity 83.8%).Conclusion: Deep learning algorithms can differentiate a retinal fundus image from other images. Algorithms can evaluate the quality of an image, discriminate between right or left eye and detect the presence of AMD and GON with a high level of accuracy, sensitivity and specificity.Keywords: artificial intelligence, retinal diseases, screening, retinal fundus image

In Europe senior citizens are a fast growing part of population, increasing proportion of disabled persons and that of persons with reduced quality of life. The concept of disability itself is not always precise and quantifiable. To improve agreement on it, the World Health Organization (WHO) developed the clinical test WHO Disability Assessment Schedule, (WHO-DASII) that includes physical, mental, and social well-being, as a generic measure of functioning. From the medical point of view, the purpose of this work is to extract knowledge about performance of the WHO-DASII using a sample of patients from an italian hospital. This Knowledge Discovery problem has been faced by using clustering based on rules, an hybrid AI and Statistics technique introduced by Gibert (1994), which combines some Inductive Learning (from AI) with clustering (from Statistics) to extract knowledge from certain complex domains in form of tipical profiles. In this paper, the results of applying this technique to the WHO-DASII results is presented together with a comparison of other more classical analysis approaches. [ABSTRACT FROM AUTHOR]

This paper presents and motivates the use of a given knowledge representation designed to facilitate the knowledge sharing from heterogeneous sources for problem solving in environmental processes using an Agent based technology. In order to test the feasibility of this representation (Domino Model), a common problem occurring in the wastewater treatment process has been selected (filamentous bulking). [ABSTRACT FROM AUTHOR]

Describes the implementation of the proposed Hybrid Supervisory System (HSS) into a real wastewater treatment plant (WWTP). Factors for consideration when transferring the HSS to another facility; Problems in implementing the HSS.

Discusses how knowledge management can improve environmental decision support systems (EDSS). Goals of knowledge management; Synergies between knowledge management and environmental systems; Description of a general model for EDSS.