DETECTION OF STANDARD PLANE FROM ULTRASOUND SCANS BY DEEP LEARNING METHODS FOR THE DIAGNOSIS OF DEVELOPMENTAL HIP DYSPLASIA

Öz

The term developmental dysplasia of the hip (DDH) describes a range of hip abnormalities affecting newborns where the femoral head and acetabulum are in improper alignment or grow abnormally, or both. The ultrasonographic evaluation technique rely on the capability of the ultrasonographer to pick up the accurate frame used for exact calculations. In our study we developed a new computer aided system that determines the exact frame from real time 2D ultrasound images and calculates the accuracy rate for each result. The deep learning architectures recently used in literature were utilized for these processes. In addition, transfer learning was carried out to increase the performance of the system using pretrained networks (SqueezeNet, VGG16, VGG19, ResNet50 and ResNet101). One of the best methods of object detection, You Only Look Once (YOLO) model, was used with pre-trained networks to determine DDH location. As a result of the study, the performance of the deep neural network model proposed with the help of these pre-trained networks was evaluated. When the obtained results were compared with expert opinions, frames (standard planes) in 605 of 676 (89.05%) test images were correctly detected. The accuracy rates for the used pre-trained networks were obtained as SqueezeNet 0.79, VGG16 0.95, VGG19 0.96, ResNet50 0.88 and ResNet101 0.93.

Anahtar Kelimeler

• DDH
• Standard Plane
• Deep Learning
• Transfer Learning
• YOLO

GELİŞİMSEL KALÇA DİSPLAZİSİ TANISINDA DERİN ÖĞRENME YÖNTEMLERİYLE ULTRASON TARAMALARINDAN STANDART DÜZLEM TESPİTİ

Öz

Gelişimsel kalça displazisi (GKD) terimi, femur başı ve asetabulumun yanlış hizada olduğu, anormal şekilde büyüdüğü veya her ikisinin birden olduğu yeni doğanları etkileyen bir dizi kalça anormalliği olarak tanımlanır. Ultrasonografik değerlendirme tekniği, ultrasonografi uzmanının kesin hesaplamalar için kullanılan doğru çerçeveyi(standart düzlem) seçme yeteneğine dayanır. Çalışmamızda, gerçek zamanlı 2B ultrason görüntülerinden standart düzlemi belirleyen ve her bir sonuç için doğruluk oranını hesaplayan yeni bir bilgisayar destekli sistemi geliştirilmiştir. Bu işlemler için literatürde son zamanlarda kullanılan derin öğrenme mimarilerinden yararlanılmıştır. Ayrıca önceden eğitilmiş ağlar (SqueezeNet, VGG16, VGG19, ResNet50 ve ResNet101) kullanılarak, sistemin performansını artırmak için transfer öğrenmesi gerçekleştirilmiştir. Nesne algılamanın en iyi yöntemlerinden biri olan You Only Look Once (YOLO) modeli, DDH konumunu belirlemek için önceden eğitilmiş ağlarla birlikte kullanılmıştır. Çalışma sonucunda önceden eğitilmiş bu ağlar yardımıyla önerilen derin sinir ağı modelinin performansı değerlendirilmiştir. Elde edilen sonuçlar uzman görüşleri ile karşılaştırıldığında 676 test görüntüsünün 605(%89,05) 'inde doğru kareler (standart düzlemler) doğru olarak tespit edilmiştir. Kullanılan önceden eğitilmiş ağlar için doğruluk oranları SqueezeNet 0.79, VGG16 0.95, VGG19 0.96, ResNet50 0.88 ve ResNet101 0.93 olarak elde edilmiştir.

Anahtar Kelimeler

• GKD
• Standart Düzlem
• Derin Öğrenme
• Transfer Öğrenme
• YOLO

Kaynakça

1. Chen, L., Cui, Y., Song, H., Huang, B., Yang, J., Zhao, D., & Xia, B. (2020). Femoral head segmentation based on improved fully convolutional neural network for ultrasound images. Signal, Image and Video Processing, 1-9.
2. Ciresan, D. C., Meier, U., Gambardella, L. M., & Schmidhuber, J. (2011). Convolutional neural network committees for handwritten character classification. Paper presented at the 2011 International Conference on Document Analysis and Recognition.
3. Cireşan, D. C., Meier, U., Gambardella, L. M., & Schmidhuber, J. (2010). Deep, big, simple neural nets for handwritten digit recognition. Neural computation, 22(12), 3207-3220.
4. Dezateux, C., & Rosendahl, K. (2007). Developmental dysplasia of the hip. The Lancet, 369(9572), 1541-1552.
5. Fukushima, K. (1980). A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position. Biol. Cybern., 36, 193-202.
6. Gamage, H., Wijesinghe, W., & Perera, I. (2019). Instance-based segmentation for boundary detection of neuropathic ulcers through Mask-RCNN. Paper presented at the International Conference on Artificial Neural Networks.
7. Golan, D., Donner, Y., Mansi, C., Jaremko, J., & Ramachandran, M. (2016). Fully automating Graf’s method for DDH diagnosis using deep convolutional neural networks. In Deep Learning and Data Labeling for Medical Applications (pp. 130-141): Springer.
8. Graf, R. (2006). Hip sonography: diagnosis and management of infant hip dysplasia: Springer Science & Business Media.

9. Hareendranathan, A. R., Zonoobi, D., Mabee, M., Cobzas, D., Punithakumar, K., Noga, M., & Jaremko, J. L. (2017). Toward automatic diagnosis of hip dysplasia from 2D ultrasound. Paper presented at the 2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017).
10. Iandola, F. N., Han, S., Moskewicz, M. W., Ashraf, K., Dally, W. J., & Keutzer, K. J. a. p. a. (2016). SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and< 0.5 MB model size.
11. Irene, K., Haidi, H., Faza, N., & Chandra, W. (2019). Fetal Head and Abdomen Measurement Using Convolutional Neural Network, Hough Transform, and Difference of Gaussian Revolved along Elliptical Path (Dogell) Algorithm. arXiv preprint arXiv:1911.06298.
12. Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. Paper presented at the Advances in neural information processing systems.
13. Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017). Imagenet classification with deep convolutional neural networks. Communications of the ACM, 60(6), 84-90.
14. LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11), 2278-2324.
15. Matlab. (2020). Image Labeler. Retrieved from https://www.mathworks.com/help/vision/ug/get-started-with-the-image-labeler.html
16. Paserin, O., Mulpuri, K., Cooper, A., Abugharbieh, R., & Hodgson, A. J. (2018). Improving 3D ultrasound scan adequacy classification using a three-slice convolutional neural network architecture. CAOS, 2, 152-156.
17. Paserin, O., Mulpuri, K., Cooper, A., Hodgson, A. J., & Abugharbieh, R. (2017). Automatic near real-time evaluation of 3D ultrasound scan adequacy for developmental dysplasia of the hip. In Computer Assisted and Robotic Endoscopy and Clinical Image-Based Procedures (pp. 124-132): Springer.
18. Paserin, O., Mulpuri, K., Cooper, A., Hodgson, A. J., & Garbi, R. (2018). Real time RNN based 3D ultrasound scan adequacy for developmental dysplasia of the hip. Paper presented at the International Conference on Medical Image Computing and Computer-Assisted Intervention.
19. Qassim, H., Verma, A., & Feinzimer, D. (2018). Compressed residual-VGG16 CNN model for big data places image recognition. Paper presented at the 2018 IEEE 8th Annual Computing and Communication Workshop and Conference (CCWC).
20. Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real-time object detection. Paper presented at the Proceedings of the IEEE conference on computer vision and pattern recognition.
21. Redmon, J., & Farhadi, A. (2017). YOLO9000: better, faster, stronger. Paper presented at the Proceedings of the IEEE conference on computer vision and pattern recognition.
22. Rhu, M., Gimelshein, N., Clemons, J., Zulfiqar, A., & Keckler, S. W. (2016). vDNN: Virtualized deep neural networks for scalable, memory-efficient neural network design. Paper presented at the 2016 49th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO).
23. Schams, M., Labruyère, R., Zuse, A., & Walensi, M. (2017). Diagnosing developmental dysplasia of the hip using the Graf ultrasound method: risk and protective factor analysis in 11,820 universally screened newborns. European Journal of Pediatrics, 176(9), 1193-1200.
24. Shaw, B. A., & Segal, L. S. (2016). Evaluation and referral for developmental dysplasia of the hip in infants. Pediatrics, 138(6).
25. Simonyan, K., & Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556.
26. Szegedy, C., Ioffe, S., Vanhoucke, V., & Alemi, A. (2016). Inception-v4, inception-resnet and the impact of residual connections on learning. arXiv preprint arXiv:1602.07261.
27. Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., . . . Rabinovich, A. (2015). Going deeper with convolutions. Paper presented at the Proceedings of the IEEE conference on computer vision and pattern recognition.
28. Tang, M., Zhang, Z., Cobzas, D., Jagersand, M., & Jaremko, J. L. (2018). Segmentation-by-detection: A cascade network for volumetric medical image segmentation. Paper presented at the 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018).
29. Van Rijthoven, M., Swiderska-Chadaj, Z., Seeliger, K., van der Laak, J., & Ciompi, F. (2018). You only look on lymphocytes once.
30. Weiss, K., Khoshgoftaar, T. M., & Wang, D. (2016). A survey of transfer learning. Journal of Big data, 3(1), 9.
31. Wong, A., Famuori, M., Shafiee, M. J., Li, F., Chwyl, B., & Chung, J. (2019). YOLO nano: A highly compact you only look once convolutional neural network for object detection. arXiv preprint arXiv:1910.01271.
32. Yang, S., Zusman, N., Lieberman, E., & Goldstein, R. Y. (2019). Developmental dysplasia of the hip. Pediatrics, 143(1).