Showing papers by "Heng-Da Cheng published in 2002"
TL;DR: The experimental results demonstrate that the HOB can find homogeneous regions more effectively than HIB does, and can solve the problem of discriminating shading in color images to some extent.
189 citations
TL;DR: The experiments demonstrate that the proposed method can effectively enhance the contours and fine details of the mammographic features which will be useful for breast cancer diagnosis.
97 citations
Journal Article•
TL;DR: This Innovations Deserving Exploratory Analysis (IDEA) project refined and evaluated in the field an automated high resolution imaging system to detect and classify pavement cracks in real time at highway speeds and is ready for surveying pavement distress on highways.
Abstract: This Innovations Deserving Exploratory Analysis (IDEA) project refined and evaluated in the field an automated high resolution imaging system to detect and classify pavement cracks in real time at highway speeds. Work in a vehicle with a camera and accessories installed by the Utah Department of Transportation (DOT) allowed integration of the pavement crack analysis and detection system developed in the earlier IDEA project (NCHRP-81). While field tests demonstrated the system's capability of recording and processing of images at speeds up to 80 mph, the camera performed unsatisfactorily for the desired resolution. Consequently, a line camera with necessary specifications was procured and used to collect additional data. However, the line camera showed problems with synchronization, white light calibration, and interruption in image capture with change in scan rate. Use of wide angle lens with area camera produced distortion in the captured images. While an interpolation method appeared to help correct the distortion, it greatly increased the processing time. A satisfactory solution was to use two cameras without the wide-angle lens. This approach was used in field testing by Utah DOT. The testing program used five descriptive statistics (accuracy, sensitivity, specificity, positive predictive value and negative predictive value) to objectively evaluate the system's performance. The test results and feedback from Utah DOT were used to refine and upgrade the system. The final integrated system is ready for surveying pavement distress on highways. The detailed list of test images and results can be downloaded from the website http://cvprip.cs.usu.edu/idea.
9 citations
01 Jan 2002
TL;DR: Mammography has proven to be the most reliable method and the major diagnosis means for detecting and classifying breast cancer in the early stage and a decrease in both severe breast cancer and mortality in women who undergo regular mammographic screens.
Abstract: Breast cancer continues to be one of the most deadly diseases among American women, which is the second leading cause of cancer-related mortality among American women. Currently there are more than 50 million women over the age of 40 at risk of breast cancer and approximately 144,000 new cases of breast cancer are expected each year in the United States. One out of eight women will develop breast cancer at some point during her lifetime in this country [1,2]. Because of the high incidence of breast cancer, any improvement in the process of diagnosing the disease may have a significant impact on saving lives and cutting costs in the health care system. Since the cause of breast cancer remains unknown and the earlier stage tumors can be more easily and less expensively treated, early detection is the key to breast cancer control. Mammography has proven to be the most reliable method and the major diagnosis means for detecting and classifying breast cancer in the early stage. Studies have shown a decrease in both severe breast cancer and mortality in women who undergo regular mammographic screens [3].
2 citations
Proceedings Article•
01 Jan 2002
TL;DR: Fuzzy neural network provided an easy and scalable tool for malignant mass detection and is easy to incorporate experts' knowledge to the system for practical applications.
Abstract: This paper presents a new fuzzy neural network (FNN) approach to detect malignant masses on mammograms. The FNN has four layers. The first layer is the input layer consisting of 4 input fuzzy neurons. The second layer has 4 ordinary neurons. The third layer consists of m maximum fuzzy neurons. The number of fuzzy neurons in this layer is determined during the training process and varies with the network parameters set and data distribution. The fourth layer has 2 competitive fuzzy neurons. Mammograms were obtained from the digital database for screening mammography (DDSM). Regions of interest (ROIs) were extracted from each mammogram in six different sizes. All extracted ROIs in each size were randomly divided into two groups: training and testing groups. The co-occurrence matrix of each ROI of the image was calculated. Maximum probability, second order of element difference moment, entropy, and uniformity for image size of 256x256 were used as inputs after fuzzification of the image . A total of 90 images were divided into two sets, each with 19 normal images and 26 images with malignant masses. After training with the training set, the FNN can correctly detect malignant masses of the mammograms with accuracy of 85% to 96%. Fuzzy neural network provided an easy and scalable tool for malignant mass detection. It is easy to incorporate experts' knowledge to the system for practical applications.