中耳炎是兒童很常見的一種疾病，根據流行病學研究資料顯示，1歲以前的嬰幼兒約有60%以上感染過中耳炎；3歲以前高達80%以上的兒童有中耳炎病史；6歲以前，幾乎所有兒童都至少感染過1次中耳炎。臺灣2006年的健保資料顯示，兒童急性中耳炎的年發生率高達64.5人次/千人，年復發率更高達33.5%。對兒童而言中耳炎是嚴重疾病，早期診斷及治療非常重要。臨床發現，許多家長把中耳炎誤以為是感冒，且中耳炎僅半數患者會有發燒症狀。由於中耳炎無法從外觀辨識，若小孩還不會表達，家人常會忽視症狀，甚至非專科醫師也很容易誤判，從而延誤治療黃金時期，後果可能導致患者聽力受損甚至完全喪失，影響其語言發展及學習能力，嚴重時更可能擴大感染到其他部位。針對高盛行率、高復發率的中耳炎帶來的家庭醫療成本與臨床診治缺口，本計畫將多工稀疏表示技術應用於耳鏡數位影像分類演算法，結合雲端計算、儲存體和網絡整合工具等基礎設施配置伺服器與分析軟體，利用雲端服務的特性和優勢提供家庭和醫療院所一套可靠且易於使用的中耳炎輔助診斷系統。 ;Otitis Media (OM) is a common pediatric disease. According to epidemiological research data, the prevalence of OM infection in children before the age of 1 is more than 60%; up to 80% of children had OM before the age of 5 and almost all children are infected with OM at least once before the age of 6. Up to 64.5 per thousand children have acute OM annually in Taiwan based on the National Health Insurance data in 2006. Moreover, the annual recurrence rate is as high as 33.5%.Since Otitis Media can develop some serious complications for the children, early diagnosis and treatment of OM is very important. Many parents, however, easily considered OM as common cold. In particular, OM can hardly be determined with visual inspection, and only half of the patients with OM develop fever. For the children have OM who cannot verbally describe their other symptoms, the parents often ignore the symptoms related to OM and the physicians with the specialty other than ENT could misdiagnose the disease based on the symptoms. The golden window for OM treatment is, therefore, easily missed and the patients may consequently develop hearing impairment or loss that further affect the language development and the learning capability or even became systemic infection of the patients if the infection spread out.In order to save the medical spending of a family and fix the health care deficits from the high prevalence and recurrence of OM and its complications, an image system with diagnosis aiding algorithm is developed based on the multi-task sparse representation technique and served as a solution for the diagnostic classification of the digital images captured by the smart otoscope in this project. In combination of cloud computing and data storage server, we aims to develop a reliable and useful OM diagnosis aiding system to provide better care of OM for the family and the hospital.