Report of Clinical Bone Age Assessment using Deep Learning for an Asian population in Taiwan

Deep learning technique has made a tremendous impact on medical image processing and analysis. Typically, the procedure of medical image processing and analysis via deep learning technique includes image segmentation, image enhancement, and classification or regression. A challenge for supervised deep learning frequently mentioned is the lack of annotated training data. In this paper, we aim to address the problems of training transferred deep neural networks with limited amount of annotated data. We proposed a versatile framework for medical image processing and analysis via deep active learning technique. The framework includes (1) applying deep active learning approach to segment specific regions of interest (RoIs) from raw medical image by using annotated data as few as possible; (2) generative adversarial Network is employed to enhance contrast, sharpness, and brightness of segmented RoIs; (3) Paced Transfer Learning (PTL) strategy which means fine-tuning layers in deep neural networks from top to bottom step by step to perform medical image classification or regression tasks. In addition, in order to understand the necessity of deep-learning-based medical image processing tasks and provide clues for clinical usage, class active map (CAM) is employed in our framework to visualize the feature maps. To illustrate the effectiveness of the proposed framework, we apply our framework to the bone age assessment (BAA) task using RSNA dataset and achieve the state-of-the-art performance. Experimental results indicate that the proposed framework can be effectively applied to medical image analysis task.

Download Full-text

MON-LB017 The Effectiveness of Computed Assessment Using GP and TW3 Hybrid System

Journal of the Endocrine Society ◽

10.1210/jendso/bvaa046.2074 ◽

2020 ◽

Vol 4 (Supplement_1) ◽

Author(s):

Lindsey Yoojin Chung ◽

Kyu-chong Lee ◽

Kyung-Sik Ahn ◽

Jae Jun Lee ◽

Chang Ho Kang ◽

...

Keyword(s):

Deep Learning ◽

Clinical Practice ◽

Confidence Interval ◽

Hybrid Method ◽

Reading Time ◽

Bone Age ◽

Software System ◽

Age Assessment ◽

Bone Age Assessment ◽

Automated Software

Abstract Background: Bone age assessments (BAAs) is an important clinical modality to investigate endocrine, genetic and growth disorders in children. It is generally performed by radiological examination of the left hand by using either the Greulich-Pyle (GP) or the Tanner-Whitehouse (TW) method. However, both clinical procedures show several limitations, from significant intra- and inter-operator variability to examination effort of clinicians. To address these problems, several automated approaches have been proposed; nevertheless, some disparity still exists between automated BAAs and manual BAAs to be employed in clinical practice. To overcome this disparity, deep learning-based bone age assess software using GP and TW3 hybrid method has been developed. In this study, we evaluate the accuracy and efficiency of the new automated hybrid software system for bone age assessment and validate its feasibility in clinical practice. Materials and Methods: Greulich-Pyle (GP) and Tanner-Whitehouse (TW3) hybrid method-based deep-learning technique was used to develop the automated software system for bone age assessment. Total 102 radiographs from children with the chronological age of 4.9-17.0 years (mean age 10.9±2.3, 51 cases for females and 51 cases for males) were selected and bone age was estimated with this software. For validation of the automated software system, three human experts have manually performed BAAs at expert’s discretion based on GP method for accuracy estimation and one naïve radiologist performed BAAs with automated software system assist and BAAs reading time was recorded in each session for efficiency evaluation. The performance of automated software system was assessed by comparing mean absolute difference (MAD) between the system estimates and the experts manual BAAs.Results: The results of bone age assessment by human experts and automated software system showed no significant difference between the two groups. Each assessed average of bone age were 11.39 ± 2.74 and 11.35 ± 2.76, respectively. MAD was 0.39 years between automated software system BAAs and experts manual BAAs. The 95% confidence interval of the MAD was 0.33 years and 0.45 years. BAAs reading time was reduced from 56.81 sec (95% confidence interval 52.81 - 60.81 sec) in naïve manual BAAs to 31.72 sec (95% confidence interval 29.74 - 33.69 sec) in automated software system assisted BAAs and statistically significant (p < 0.001). MAD showed 0.42 years between naïve manual BAAs and the software-assisted BAAs (95% confidence interval 0.31-0.47 years).Conclusion: The newly developed GP and TW3 hybrid automated software system were reliable for bone age assessments with equivalent accuracy to human experts. Also, the automated system appeared to enhance efficiency by reducing reading times without compromising diagnostic accuracy.

Download Full-text

Deep Learning and Classic Machine Learning Approach for Automatic Bone Age Assessment

2019 4th Asia-Pacific Conference on Intelligent Robot Systems (ACIRS) ◽

10.1109/acirs.2019.8935965 ◽

2019 ◽

Cited By ~ 1

Author(s):

Ari Wibisono ◽

Mei Silviana Saputri ◽

Petrus Mursanto ◽

Joachim Rachmad ◽

Alberto ◽

...

Keyword(s):

Machine Learning ◽

Deep Learning ◽

Bone Age ◽

Learning Approach ◽

Age Assessment ◽

Machine Learning Approach ◽

Bone Age Assessment

Download Full-text

Data Enhancement and Deep Learning for Bone Age Assessment using The Standards of Skeletal Maturity of Hand and Wrist for Chinese

10.1109/embc46164.2021.9630226 ◽

2021 ◽

Author(s):

Yu Lu ◽

Xi Zhang ◽

Liwen Jing ◽

Xianghua Fu

Keyword(s):

Deep Learning ◽

Skeletal Maturity ◽

Bone Age ◽

Age Assessment ◽

Bone Age Assessment

Download Full-text

Deep Learning Framework for Automatic Bone Age Assessment

10.1109/embc46164.2021.9629650 ◽

2021 ◽

Author(s):

Chaitanya Mehta ◽

Bibi Ayeesha ◽

Ayesha Sotakanal ◽

Nirmala S. R ◽

Shrinivas D Desai ◽

...

Keyword(s):

Deep Learning ◽

Bone Age ◽

Age Assessment ◽

Learning Framework ◽

Bone Age Assessment

Download Full-text

Deep Learning Approach Applied to Prediction of Bone Age Based on Computed Tomography Orthopedic Image Processing

Journal of Medical Imaging and Health Informatics ◽

10.1166/jmihi.2020.3053 ◽

2020 ◽

Vol 10 (5) ◽

pp. 1242-1248

Author(s):

Gefei Tan ◽

Daoshun Wang

Keyword(s):

Computed Tomography ◽

Mathematical Model ◽

Deep Learning ◽

Image Recognition ◽

Bone Age ◽

Research Progress ◽

Age Assessment ◽

Black And White ◽

Significant Difference ◽

Bone Age Assessment

Objective: Deep learning and neural network models are new research directions in the field of machine learning and artificial intelligence. Deep learning has made breakthroughs in image recognition and speech recognition applications, and has also shown unique advantages in face recognition and information retrieval, and has been widely used. Methods: Thin-layer computed tomography (CT) scan and multiplanar reconstruction (MPR) and volume reconstruction (VR) techniques were used to perform CT thin-slice scan and volume of the bilateral clavicle sternum at 548 number of l5~25 years old. Reproduction (volume rendering, VR) and three-dimensional image recombination, measuring and calculating the longest diameter of the sternal end of the bilateral clavicle, the longest diameter of the metaphysis and its length ratio, the area of the epiphysis, the area of the metaphysis and its area ratio, etc. We establish a mathematical model of bone age inference, and then substitute 50 training samples into the mathematical model to test the accuracy of the model. Results: There was a statistically significant difference between the male and female sex ratios in the same age group (P < 0.05). The established mathematical model shows that the developmental law of the sternal skeletal bone is highly correlated with the biological age. The accuracy of all models is 70.5% (±1.0 years old) and 82.5% (±1.5 years). Skeletal X-ray images show different gradation changes in black and white, with black-and-white contrast and hierarchical image features. Based on the advantages of deep learning in image recognition, we combine it with bone age assessment research to build a forensic bone age automation. Conclusions: This paper harnesses the basic concepts of deep learning and its network structure, and expounds the research progress of deep learning in image recognition in different research fields at home and abroad in recent years, as well as the advantages and application prospects of deep learning in bone age assessment.

Download Full-text