Medical Image Analysis
CuraCloud™ A.I. will provide clinicians a diagnostic edge in multiple areas of image analysis, including:
Technological advantages of the platform:
- Proprietary algorithms and models built by experts in medical imaging and machine learning
- Storage efficient: Loss-less compression ratio ~ 20X
- Searchable: Single search in < 50 ms
- Comprehensive: Spans imaging, clinical notes, genomics
Radiology practices will be able to use CuraCloud A.I. solutions to speed radiologist review of critical cases, introduce new kinds of exams, and increase the throughput of image review without increasing costs or reducing quality.
Examples of exams being explored by CuraCloud:
- Intracranial Hemorrhage detection via CT
- Measuring stenosis or occlusion in blood vessels
- Pneumothorax detection via CT
- Pulmonary embolism identification via CT
This diagram shows how CuraCloud is developing AI-powered priority escalation to a radiologist’s worklist.
Once an image is acquired, it enters the worklist. The worklist delivers the image to CuraCloud AI for review.
CuraCloud AI identifies and flags any critical issues that require immediate radiologist review.
The AI also pre-populates the image result report for the radiologist and adds notes for the radiologist to consider (e.g., lung nodule measurement and description, malignancy prediction, etc).
The radiologist would use this information to prioritize and speed her image exams.
The AI also assists with smart worklist assignment, routing the right exam to the right clinician at the right time. This is particularly helpful for telehealth, emergency medicine and oncology applications.
Cancer Metastasis Detection
CuraCloud is developing A.I. models that enhance pathologist accuracy and efficiency in staging a variety of cancers.
Detecting metastasis of lymph nodes is critical to judging cancer prognosis, particularly for breast cancer. However, detection is challenging for the pathologist because slide images are high resolution (e.g., 100k X 200k pixels) and tissue is highly variable. CuraCloud is developing a spatially structured deep network (Spatio-Net) that improves detection of metastases by more than 5% over current state-of-the-art for breast cancer.
Spatio-Net achieves this advantage by taking spatial variation into account. In contrast, other approaches divide slide images into small areas and analyze each individually, discarding spatial structure information that is vital to detection inference. Spatio-Net gains a further advantage from its rich training dataset, which includes over 1 million image patches from 300 patients.