Mingrui Zhang

Noise Reduction and Atmospheric Correction for Coastal Applications of Landsat Thematic Mapper Imagery

Abstract The Landsat Thematic Mapper (TM) has three visible bands centered at 485 nm, 560 nm, and 660 nm which can be used for ocean applications. This article presents a method for deriving the bottom albedo from the TM in coastal waters. Our study of historical TM images shows degradation of the sensor through time. Pattern noise in the imagery in Bands 2 and 3 was analyzed and removed using a combination of Fourier filtering and edge-detection techniques. Noise was first examined over clear and deep nearby oceanic waters, and the filter algorithms developed there were applied to the entire image. To estimate water-leaving radiance from the satellite sensor, Rayleigh and aerosol path radiance were removed. Radiance due to aerosol scattering was calculated for offshore pixels assuming a marine aerosol type, and it was removed as a constant from the entire scene. The TM sensor calibration was validated by comparing water-leaving radiance values over the clear waters of the Florida current with known, normalized water-leaving radiance values. Corrections for water path radiance and water-column attenuation of bottom-reflected radiance were made for regions of known depth, allowing the bottom albedo and vegetative pixel fraction to be determined for shallow reef areas in the Florida Keys.

A generic knowledge-guided image segmentation and labeling system using fuzzy clustering algorithms

Segmentation of an image into regions and the labeling of the regions is a challenging problem. In this paper, an approach that is applicable to any set of multifeature images of the same location is derived. Our approach applies to, for example, medical images of a region of the body; repeated camera images of the same area; and satellite images of a region. The segmentation and labeling approach described here uses a set of training images and domain knowledge to produce an image segmentation system that can be used without change on images of the same region collected over time. How to obtain training images, integrate domain knowledge, and utilize learning to segment and label images of the same region taken under any condition for which a training image exists is detailed. It is shown that clustering in conjunction with image processing techniques utilizing an iterative approach can effectively identify objects of interest in images. The segmentation and labeling approach described here is applied to color camera images and two other image domains are used to illustrate the applicability of the approach.

Network-based approach identified cell cycle genes as predictor of overall survival in lung adenocarcinoma patients

Lung adenocarcinoma is the most common type of primary lung cancer. The purpose of this study was to delineate gene expression patterns for survival prediction in lung adenocarcinoma. Gene expression profiles of 82 (discovery set) and 442 (validation set 1) lung adenocarcinoma tumor tissues were analyzed using a systems biology-based network approach. We also examined the expression profiles of 78 adjacent normal lung tissues from 82 patients. We found a significant correlation of an expression module with overall survival (adjusted hazard ratio or HR=1.71; 95% CI=1.06-2.74 in discovery set; adjusted HR=1.26; 95% CI=1.08-1.49 in validation set 1). This expression module contained genes enriched in the biological process of the cell cycle. Interestingly, the cell cycle gene module and overall survival association were also significant in normal lung tissues (adjusted HR=1.91; 95% CI, 1.32-2.75). From these survival-related modules, we further defined three hub genes (UBE2C, TPX2, and MELK) whose expression-based risk indices were more strongly associated with poor 5-year survival (HR=3.85, 95% CI=1.34-11.05 in discovery set; HR=1.72, 95% CI=1.21-2.46 in validation set 1; and HR=3.35, 95% CI=1.08-10.04 in normal lung set). The 3-gene prognostic result was further validated using 92 adenocarcinoma tumor samples (validation set 2); patients with a high-risk gene signature have a 1.52-fold increased risk (95% CI, 1.02-2.24) of death than patients with a low-risk gene signature. These results suggest that a network-based approach may facilitate discovery of key genes that are closely linked to survival in patients with lung adenocarcinoma.

Interdisciplinary application tracks in an undergraduate computer science curriculum

The Computer Science Department at Winona State University revised its curriculum to include an interdisciplinary approach adapted to the study of computer science. The new curriculum consists of a traditional Computer Science option and an Applied Computer Science option consisting of four separate tracks, namely: bioinformatics, computer information systems, geographic information technology, and human computer interaction. This paper describes the design strategy and implementation plan as well as the content of our multi-track Applied Computer Science curriculum.

Embedding computer science concepts in K-12 science curricula

To engage a broader audience in computer science, we have developed a set of curriculum units embedded with computer science concepts for K-12 science education. We chose bioinformatics as a vehicle to deliver these units. Our curriculum development cycle began with the identification of a set of computer science concepts which are potentially relevant to life sciences. Problems in life sciences as well as bioinformatics tools to be used for solving these problems were carefully examined for the delivery of identified computer concepts. They were later presented to groups of regional K-12 science teachers in our summer workshop on bioinformatics. With their help, we adapted and polished these curriculum units to meet Minnesota state standards for K-12 science education. This paper describes our approach in developing the curriculum units.

A new validity measure for a correlation-based fuzzy c-means clustering algorithm

One of the major challenges in unsupervised clustering is the lack of consistent means for assessing the quality of clusters. In this paper, we evaluate several validity measures in fuzzy clustering and develop a new measure for a fuzzy c-means algorithm which uses a Pearson correlation in its distance metrics. The measure is designed with within-cluster sum of square, and makes use of fuzzy memberships. In comparing to the existing fuzzy partition coefficient and a fuzzy validity index, this new measure performs consistently across six microarray datasets. The newly developed measure could be used to assess the validity of fuzzy clusters produced by a correlation-based fuzzy c-means clustering algorithm.

Porting a cancer treatment prediction to a mobile device

The Lung Cancer Survivability Prediction Tool (LCSPT) is a web-based system that predicts the survival possibility of a lung cancer patient based on the status of the patient and the treatments provided. In order to make the LCSPT more accessible and convenient to doctors working in a clinical setting, we have developed a new interface for mobile devices. The display size along with wireless data transfer speeds pose the most significant challenges to porting a software application to a mobile device. We have addressed these issues by redefining the interface and limiting the amount of data that is required. The resultant tool provides doctors with the flexibility of mobility while maintaining the effectiveness of the desktop version of the LCSPT.

A bioinformatics track with outreach components

Bioinformatics is a discipline that uses computational tools and computer technologies to model, analyze, present, and visualize biological data. In this paper, we discuss the design of a bioinformatics track within the computer science curriculum at Winona State Universitys Rochester campus. We also developed a bioinformatics course and used it as a mechanism for computer science outreach. Our outreach program is designed to target 6th to 12th grade science teachers and help them develop K-12 science projects with bioinformatics components.

Fuzzy analysis of satellite images to find phytoplankton blooms

A knowledge-based approach to automatic classification of Coastal Zone Color Scanner (CZCS) images of the west Florida Shelf is described. The approach is utilized to monitor red tides and phytoplankton plumes, such as green river off the west Florida shelf. CZCS images are initially segmented by the unsupervised mrFCM algorithm, then a knowledge based system is applied to the centroid values of resultant clusters to label case I and case II waters, green river and red tide. Our knowledge base consists of a rule based system and an embedded neural network. Our results show, among 25 ground truth images, this system can correctly recognize all 15 images with green river and 10 images without. The system can correctly classify 75% of the pixels belonging to green river. A time series of red tide in 1978 is also successfully identified.

Knowledge-based classification of CZCS images and monitoring of red tides off the west Florida shelf

Red tides on the west Florida shelf have significant economic and public health effects. Tracking the phytoplankton bloom, known as red tide, is important to understanding the phenomena. In this paper, a knowledge-based approach to automatic classification of Coastal Zone Color Scanner satellite images is developed. The Coastal Zone Color Scanner or CZCS images are initially segmented by the unsupervised mr-FCM algorithm then an expert system utilizes rules, and an iterative clustering process, to recognize case I (deep) water, case II (shallow) water and red tide by searching for expected features. The results show that this system is effective in recognizing images with red tide and segmenting the red tide.