Aifeng Zhang PhD, Presenter: Nothing to Disclose

Arkadiusz Gertych PhD, Abstract Co-Author: Nothing to Disclose

Sinchai Tsao MS, Abstract Co-Author: Developer, SurgMatix, Inc

Brent Julius Liu PhD, Abstract Co-Author: Nothing to Disclose

Han K. Huang DSc, Abstract Co-Author: Nothing to Disclose

A computer-aided-diagnosis (CAD) method for bone age assessment (BAA) has been developed based on phalangeal regions of interest (ROI). In order to improve the accuracy for young children, carpal ROI was incorporated. In consideration of the missing feature problem and imprecise nature of relationship between bone growths and chronological age, a rule-based fuzzy inference system was developed to assess the bone age. Two goals of this paper are to develop a fuzzy system for carpal ROI, and a modular and extensible model integrating carpal ROI with the existing phalangeal ROIs analysis. Fuzzy system design, training and testing have been accomplished and BAA results obtained from the fuzzy system have been evaluated with our data collection of 1,400 normals.

A digital hand atlas has been developed with 1,400 normal children hand images of Caucasian, African American, Hispanic, and Asian. Carpal bone analysis of each hand image yields six morphological features, which were taken as the input into the carpal fuzzy sub-system. During the training process, the output was set as the subject’s chronological age since the normality of children was ensured. An automatic training algorithm was used to generate Gaussian membership functions for inputs and output and then rules for each sub-system, one for each category by race and gender. To assess the bone age of a given hand image, two steps are involved: A fuzzy bone age BA1 was obtained from the existing phalangeal sub-system and BA2 from carpal sub-system; BA1 and BA2 were then integrated into the final fuzzy BA which yields a precise bone age through the defuzzification procedure using center of gravity method.

The integrated fuzzy system takes the features from both regions as input and predicts the bone age. This fuzzy system can process a dynamic number of inputs in case one or several ROIs detection fails.

This presentation demonstrates the use of a modular and extensible fuzzy inference system to integrate both phalangeal ROIs and carpal ROI for BAA.

This presentation demonstrates the applicability of the fuzzy system for BAA based on morphological bony features of multiple hand ROIs.

Zhang, A, Gertych, A, Tsao, S, Liu, B, Huang, H, Fuzzy System Design for Bone Age Assessment of Children in Mulitple Regions of the Hand.  Radiological Society of North America 2007 Scientific Assembly and Annual Meeting, November 25 - November 30, 2007 ,Chicago IL. http://archive.rsna.org/2007/5014896.html