EAR PATHOLOGIES USING DEEP LEARNING ON OTOSCOPIC IMAGES

Abstract

In this study, the performance of different deep learning architectures is comparatively analyzed for the classification of ear pathologies based on otoscopic images. The dataset included four basic classes: chronic otitis media, ear wax obstruction, myringosclerosis and normal ear structure. The images were normalized at 224×224-pixel resolution and made suitable for the model, and classification was performed using CNN, CNN-LSTM, DenseNet121, ResNet50 and EfficientNet architectures. During the training and validation phases, performance metrics such as accuracy, F1 score, precision, recall and loss values were calculated, and the class discrimination power of the models was evaluated with ROC curves and complexity matrices. According to the results, CNN+LSTM and DenseNet121 architectures showed the best performance with over 94% accuracy and high F1 score in both training and validation sets. Some transfer learning-based architectures such as EfficientNet and ResNet50 showed low generalization performance. This study demonstrates the effectiveness of deep learning-based models for computerized diagnosis of intra-ear diseases and provides an important basis for decision support systems to be developed in this field.

Keywords

• Deep Learning
• Ear Diseases
• Otoscopic Images
• Image Classification

OTOSKOPİK GÖRÜNTÜLER ÜZERİNDE DERİN ÖĞRENME KULLANARAK YAYGIN KULAK PATOLOJİLERİNİN SINIFLANDIRILMASI

Abstract

Bu çalışmada, otoskopik görüntülere dayalı kulak patolojilerinin sınıflandırılması için farklı derin öğrenme mimarilerinin performansı karşılaştırmalı olarak analiz edilmiştir. Veri kümesi dört temel sınıf içermektedir: kronik otitis media, kulak kiri tıkanıklığı, miringoskleroz ve normal kulak yapısı. Görüntüler 224×224 piksel çözünürlükte normalize edilerek modele uygun hale getirilmiş ve sınıflandırma CNN, CNN-LSTM, DenseNet121, ResNet50 ve EfficientNet mimarileri kullanılarak gerçekleştirilmiştir. Eğitim ve doğrulama aşamalarında doğruluk, F1 skoru, kesinlik, geri çağırma ve kayıp değerleri gibi performans metrikleri hesaplanmış, modellerin sınıf ayırt etme gücü ROC eğrileri ve karmaşıklık matrisleri ile değerlendirilmiştir. Sonuçlara göre, CNN+LSTM ve DenseNet121 mimarileri hem eğitim hem de doğrulama setlerinde %94'ün üzerinde doğruluk ve yüksek F1 skoru ile en iyi performansı göstermiştir. EfficientNet ve ResNet50 gibi bazı transfer öğrenme tabanlı mimariler düşük genelleme performansı göstermiştir. Bu çalışma, kulak içi hastalıkların bilgisayarlı teşhisi için derin öğrenme tabanlı modellerin etkinliğini göstermekte ve bu alanda geliştirilecek karar destek sistemleri için önemli bir temel sağlamaktadır.

Keywords

• Derin Öğrenme
• Kulak Hastalıkları
• Otoskopik Görüntüler
• Görüntü Sınıflandırma

References

1. J. Chan, K. Stephenson, Diagnosis and management of middle ear disease in children. Paediatrics and Child Health 33(12) (2023) 376-381.
2. T. Marom, O. Kraus, N. Habashi, S.O. Tamir, Emerging technologies for the diagnosis of otitis media. Otolaryngology–Head and Neck Surgery 160(3) (2019) 447-456.
3. A. Bone, Middle Ear. Pediatric Physical Examination-E-Book: Pediatric Physical Examination-E-Book (2023) 192.
4. H.M. Afify, K.K. Mohammed, A.E. Hassanien, Insight into automatic image diagnosis of ear conditions based on optimized deep learning approach. Annals of biomedical engineering 52(4) (2024) 865-876.
5. D. Song, T. Kim, Y. Lee, J. Kim, Image-based artificial intelligence technology for diagnosing middle ear diseases: a systematic review. Journal of Clinical Medicine 12(18) (2023) 5831.
6. F. Larrosa, L. Pujol, E. Hernández-Montero, Chronic otitis media. Medicina Clínica (English Edition), (2025) 106915.
7. R.G. Kashani, M.C. Młyńczak, D. Zarabanda, P. Solis-Pazmino, D.M. Huland, I.N. Ahmad, T.A. Valdez, Shortwave infrared otoscopy for diagnosis of middle ear effusions: A machine-learning-based approach. Scientific Reports 11(1) (2021) 12509.
8. A. Mahdavi, Diagnostic and imaging findings in inflammatory Opacifications of the middle ear: A review of the literature. The International Tinnitus Journal 27(2) (2023) 146-153.

9. M.A. Khan, S. Kwon, J. Choo, S.M. Hong, S.H. Kang, I.H. Park, S.J. Hong, Automatic detection of tympanic membrane and middle ear infection from oto-endoscopic images via convolutional neural networks. Neural Networks 126 (2020) 384-394.
10. M. Viscaino, J.C. Maass, P.H. Delano, M. Torrente, C. Stott, F. Auat Cheein, Computer-aided diagnosis of external and middle ear conditions: A machine learning approach. Plos one 15(3) (2020) e0229226.
11. A. K. Singh, A.S. Raghuvanshi, R. Mehta, A Soft Computing Approach for Efficient Diagnosis of Otitis Media Infection by Mucosal Disease Early Detection and Referrals. In 2024 2nd International Conference on Device Intelligence, Computing and Communication Technologies (DICCT) (2024) (pp. 687-692) IEEE.
12. Z. Cao, F. Chen, E.M. Grais, F. Yue, Y. Cai, D.W. Swanepoel, F. Zhao, Machine learning in diagnosing middle ear disorders using tympanic membrane images: a meta‐analysis, The Laryngoscope 133(4) (2023) 732-741.
13. M. A. H. Rony, K. Fatema, M. A. K. Raiaan, M. M. Hassan, S. Azam, A. Karim, A. Leach Artificial intelligence-driven advancements in otitis media diagnosis: a systematic review (2024) Ieee access.
14. M. Viscaino and F.A. Cheein , “Ear imagery database,” Figshare, (2020) [Online] Available: https://figshare.com/articles/dataset/Ear_imagery_database/11886630