machine vision and learning implementations

Projects demonstrating this capability:

HAMMS - Hand-Arm Movement Monitoring System

HAMMS was originally conceived by NIDA as a system for pre-symptomatic diagnosis of Parkinson’s disease by monitoring a patient’s fine locomotor abilities. SPIS designed HAMMS to be a clinic-friendly (i.e., low cost, portable, and quick routine diagnostic) system consisting of a camera and custom video processing algorithm.  The system tracks the hand of a patient performing a set of motions as defined by a path outlined on a computer screen, and also provides visual feedback (i.e., real-time machine vision) to the participant as to how well they are maintaining the desired path.

Photo of HAMMS system in use

Image-Based Robotic Targeting System to Control Micromanipulators for Single Cell Studies in Living Biological Tissues

We are developing a second-generation software system for image-based robotic automation of micromanipulators, which entails seamless integration and control of combined optical-mechanical systems used for NINDS biological investigation.

3D model of micromanipulator and micropipette

mHealth Smartphone Application to Measure Risky Driving Behavior and Predict Crashes

The risk of fatal crashes per driven mile is highest among teen drivers compared to other age groups. According to the CDC statistics only in 2014, 221,313 teens drivers were admitted to the emergency department and 2,333 were killed. This amounts to an average of six teen driver deaths every day due to automotive crashes. Improved safety policies are known to reduce the number of motor vehicle crashes among teens.

mHealth Smartphone Application to Measure Risky Driving Behavior

SCORHE: System for Continuous Observation of Rodents in Home-cage Environment

In collaboration with NCI, NIDDK, NIMH, and FDA, SCORHE is designed to meet the demands of NIH IRP researchers and animal care staff to automatically assess laboratory mice ethology in large-scale settings. SCORHE is comprised of custom video-acquisition and analysis tools developed to quantify mice activity and behavior for short and long (multi-day) durations while housed within a typical home-cage. The specialized hardware was designed to be space efficient, compatible with vivarium cage racks, and animal-facility user friendly.

SCORHE 3D CAD and prototype in ventilated rack

Tissue MicroArrayer for High Throughput Analysis of Pathology Tissue Samples

NIH researchers introduced an innovative technique for high density arraying of archival clinical tissue in the research and clinical laboratory. Consisting of an array of cylindrical cores extracted from formalin-fixed paraffin embedded tissue samples, tissue microarrays (TMA) have become widely used as a powerful validation tool for high throughput genomic screens.

Automated robotic tissue microarrayer

Two-Photon Excitation Fluorescence Microscopy Motion Tracking to Study in-vivo Subcellular Structures

SPIS collaborated with the NHLBI Laboratory of Cardiac Energetics to develop a methodology enabling the study in vivo sub-cellular structures and signaling processes in real-time.  Multi-photon fluorescence imaging provides improved tissue penetration, sensitivity, and information content when investigating dynamic intercellular events within living tissue, but physiological motion degrades the quality of these images and makes temporal observations challenging.  Our system implementation, designed to be functional in tandem with the commercial two-photon microscope system, adjusts

Time-lapse volume of with and without motion tracking correction