SBIR/STTR Research Award
Development of An Integrated High Throughput Imaging and Image Analysis Platform for Muscle
Award amount
$1,001,570Program and phase
STTR · Phase IIAward year
2016Award details
- Awarding agency
- Department of Health and Human Services
- Branch
- National Institutes of Health
- Recipient company
- CytoInformatics LLC
- Proposal award date
- 2016-09-30
- Contract end date
- 2019-05-31
- UEI
- Not reported
- Solicitation number
- PA14-072
- Topic code
- NIA
Research abstract
DESCRIPTION provided by applicant There is growing awareness that weakness of muscle is a significant biomedical health issue associated with many different chronic diseases and aging There are a variety of different diseases that affect muscle including muscular dystrophy fibromyalgia cerebral palsy amyotrophic lateral sclerosis and myasthenia gravis each of which carries its own unique set of symptoms accompanied by a myriad of genetic and molecular abnormalities There is agreement within the basic research and clinical communities that important morphological characteristics of muscle fibers such as fiber cross sectional area fiber type the number and position of myonuclei cellular infiltration and fibrosis are among many critical factors that determine the health and functionality of the muscle Until now the imaging equipment available is relatively low throughput and the quantification is still largely based on manual or at best semi automatic methods that require extensive human intervention In Phase I CytoInformatics LLC has developed a software called CytoQuant which can accurately perform automatic segmentation and provide accurate boundaries of each muscle fiber for a cropped image patch The goal of CytoInformatics LLC in Phase II is to deliver the first version of the CytoQuant software to the market develop the second version of CytoQuant by continuously improving its performance especially in handling large scale whole slide images and also develop an integrated multiplexing imaging hardware and analysis software framework as a seamless solution for high throughput automated image analysis of muscle specimens After careful investigation in Phase I please refer to the details in business plan we conclude that this integrated software hardware platform with cutting edge technologies in advanced imaging large scale machine learning and big data is commercially attractive to the entire muscle community including research institutes hospitals and pharmaceutical and athletic companies demonstrated by strong enthusiasm among end users and many supporting letters The specific aims are Deliver the first version of CytoQuant developed in Phase I to the market continue to develop the second version of CytoQuant which will focus on improving the robustness and speed in handling whole slide images Develop an integrated platform that provides unified imaging hardware and image analysis software to handle whole slide images labeled with multiple bio relevant markers This unified whole slide based software and hardware solution will deliver an efficient and specialized imaging platform that seamlessly integrates with a high throughput multi marker and automatic image analysis software specifically designed for muscle PUBLIC HEALTH RELEVANCE In Phase II CytoInformatics is proposing a unified whole slide multispectral imaging MSI system that seamlessly integrates with a high throughput accurate and automated software to provide fast and efficient multiplexed imaging and quantitative muscle image analysis simultaneously This will be the first integrated hardware software solution that is specifically designed for muscle
Technology keywords
Not reported
Awards with the same topic code
Verified by an exact topic code match. This is not a text-similarity match.
- Evaluate the Safety, Tolerability, and Immunogenicity of Adjuvanted Preventive Dual Aβ/tau Vaccine, Duvax in Healthy Volunteers
- A Drug Discovery Platform to Prioritize Development of Natural Product Neuromodulators for Treating Cognitive Impairment
- VR-CARES: Feasibility of Virtual Reality for Caregiver Assembly, Relief, Empowerment, and Support to improve social connection, health, and at-home dementia care
- Developing an Integrated Brain State Monitor for Personalized Anesthesia Decision Support: Preventing Postoperative Delirium in Elderly and Patients At Risk for Alzheimer's or Related Dementias
- Lead Optimization of a Therapeutic Candidate for Alzheimer’s Disease