Automated ROI Detection in Left Hand X-ray Images using CNN and RNN

Automatic segmentation of the area of interest in medical image processing is a very important but difficult problem. Deep learning algorithms can help clinicians and radiologists determine diagnosis and treatment plans. We propose and evaluate a probabilistic approach for automated region of interest ROIs detection using convolutional neural networks (CNNs). The proposed algorithm is simple and can be divide into regions and features can be extracted for the divided regions. We also propose a preprocessing algorithm based on CNN and RNN to automatically classify ROIs that are finely adjusted through image standardization based on TW3. The result is 20%-40% more accurate than those obtained using the conventional method. In addition, input image sensitivity is approximately 40% greater and the specificity was equal to or greater than 96%.

[1]  E. Brandser,et al.  Effect of knowledge of chronologic age on the variability of pediatric bone age determined using the Greulich and Pyle standards. , 2001, AJR. American journal of roentgenology.

[2]  Simone Palazzo,et al.  Deep learning for automated skeletal bone age assessment in X‐ray images , 2017, Medical Image Anal..

[3]  Cordelia Schmid,et al.  Multi-region Two-Stream R-CNN for Action Detection , 2016, ECCV.

[4]  Seung-Won Shin,et al.  Image Enhancement with Rotating Kernel Transformation Filter Generated by Bresenham's Algorithm , 2012 .

[5]  Youngbok Cho,et al.  The Kirsch-Laplacian edge detection algorithm for predicting iris-based desease , 2017, 2017 IEEE 21st International Conference on Computer Supported Cooperative Work in Design (CSCWD).

[6]  Dimitris N. Metaxas,et al.  Multi-Instance Multi-Stage Deep Learning for Medical Image Recognition , 2017, Deep Learning for Medical Image Analysis.

[7]  Kaiming He,et al.  Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[9]  Rong Zhang,et al.  Lesion detection of endoscopy images based on convolutional neural network features , 2015, 2015 8th International Congress on Image and Signal Processing (CISP).

[10]  P. M. Garamendi,et al.  Forensic Age Estimation on Digital X‐ray Images: Medial Epiphyses of the Clavicle and First Rib Ossification in Relation to Chronological Age *,† , 2011, Journal of forensic sciences.

[11]  Sameem Abdul Kareem,et al.  A Fuzzy Inference System for Skeletal Age Assessment in Living Individual , 2017, Int. J. Fuzzy Syst..

[12]  Emmanuel Salinas,et al.  Bone age detection via carpogram analysis using convolutional neural networks , 2017, Symposium on Medical Information Processing and Analysis.

[13]  Cheol-Su Kim,et al.  Color Compensation of an Underwater Imaging System Using Electromagnetic Wave Propagation , 2016, J. Inform. and Commun. Convergence Engineering.

[14]  Nico Karssemeijer,et al.  Deep learning-based assessment of tumor-associated stroma for diagnosing breast cancer in histopathology images , 2017, 2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017).

[15]  Barry-John Theobald,et al.  Automated Bone Age Assessment Using Feature Extraction , 2012, IDEAL.

[16]  Myungjin Cho,et al.  Three-Dimensional Automatic Target Recognition System Based on Optical Integral Imaging Reconstruction , 2016, J. Inform. and Commun. Convergence Engineering.

[17]  V. De Sanctis,et al.  Are the new automated methods for bone age estimation advantageous over the manual approaches? , 2014, Pediatric endocrinology reviews : PER.

[18]  H. K. Huang,et al.  Computer-assisted bone age assessment: image preprocessing and epiphyseal/metaphyseal ROI extraction , 2001, IEEE Transactions on Medical Imaging.

[19]  Bin Liang,et al.  Using Convolutional Neural Networks and Transfer Learning for Bone Age Classification , 2017, 2017 International Conference on Digital Image Computing: Techniques and Applications (DICTA).

[20]  Shyam Lal,et al.  Fully automatic segmentation of phalanges from hand radiographs for bone age assessment , 2019, Comput. methods Biomech. Biomed. Eng. Imaging Vis..

[21]  R. Malina,et al.  Tanner–Whitehouse Skeletal Ages in Male Youth Soccer Players: TW2 or TW3? , 2018, Sports Medicine.

[22]  Somjit Arch-int,et al.  Ontology Mapping and Rule-Based Inference for Learning Resource Integration , 2016, J. Inform. and Commun. Convergence Engineering.

[23]  Soo Young Kim,et al.  Comparison of the Greulich-Pyle and Tanner Whitehouse (TW3) Methods in Bone age Assessment , 2008 .