Image Analysis Core

点击翻译按钮获取中文摘要

IMAGE ANALYSIS CORE PROJECT SUMMARY Evaluating changes that precede frailty and end of life using histological characterization of age-related lesions augments molecular, cellular, and physiologic data, and provides an understanding of early-onset mechanisms that underlie age-related changes that may eventually have clinical relevance. The overall goal of the Image Analysis Core is to develop and provide resources for the geroscience community to aid in computer-assisted histopathological analysis and discovery of age-related histological features. Recently, the NIA-funded Geropathology Research Network (GRN), established to enhance the translational value of geropathology for preclinical research studies in anti-aging clinical trials, developed and validated a grading system, designated the geropathology grading platform (GGP), for quantification and comparison of histological lesion scores in tissues from aging mice. While implementation of this grading platform by a trained pathologist may be feasible for experiments with small numbers of animals, an automated approach is necessary for experiments consisting of large sample numbers. An automated approach that can provide unbiased analysis of large sample numbers will lead to a more timesaving and cost-effective analysis and generation of more robust data. A quantitative image analysis pipeline that uses machine learning to accurately identify specific features in scanned slides of stained kidneys was recently developed. This quantitative tool can be easily adjusted to allow quantification using the GGP. The Specific Aims of the Image Analysis Core are: Aim 1. Adapt a quantitative pipeline for the analysis of aged heart, liver, and lung tissues by training and establishing classifiers. Currently, scanned slides of mouse kidneys are uploaded and processed into a large number of tiles in TIF format, and then histological features specific for the kidney are identified and automatically fed into ImageJ for quantification. This pipeline will be adapted for aging research by introducing a training set to identify tissue-specific histological features and develop filters for scoring the lesions according to the GGP. Aim 2. Validate the quantitative pipeline using an annotated set of aged mouse tissues from the Geropathology Research Network. Once pipelines specific for heart, liver, and lung are developed and trained, their accuracy and robustness will be validated by analyzing a set of annotated slides provided by the GRN. Aim 3. Develop and distribute to the geroscience community open-source, user-friendly packages for both the quantitative and discovery pipelines with online training. In addition to providing image analysis as a Core service, the pipelines will be made available to the geroscience community so that other investigators can do their own analysis and customize the pipelines for their own research. These quantitative and discovery tools can be trained for use on any tissue or organ and, once adapted, will be invaluable to the geroscience community. Computer-assisted geropathology will be a powerful tool to measure study endpoints as well as determining the effects of intervention in aging studies.