- Computer vision to identify whole slide images, highlight the problem areas and auto-detect visual indicators of deisease.
- Natural Language Processing to summarize patient history for quick reading and analysis.
- AI based Clinical Decision Support for development of cancer diagnosis models – automating tedious tasks like mitotic cell counts, IHC Scoring and enhancing sensitivity of tumor detection and quantification (prostate, breast etc)
- Provide Cloud based storage to reduce the CAPEX on IT infrastructure and provide the capabilities for auto-scaling the configuration depending upon workload
- Data encryption and secure communication protocols to ensure security of data during transmission while at rest
Technologies Used:
Problem: AI based Histopathology workflow automation and diagnosis
- Cancer diagnosis being complex, takes 2-3 weeks for the test results. Even a routine test requires counting of individual cells and taking the ratio of positive (brown) to negative (blue) stained cells.
- Although widely touted as a replacement for glass slides and microscopes in pathology, digital slides present major challenges in interoperability, data storage, security, transmission and processing.
- In the absence of a standardised data format, each vendor defines its own proprietary data formats, analysis tools, viewers and software libraries, which creates interoperability challenges for the pathologists.
Solutions :
- There is need for a solution that could help reduce the burden on the histopathologists, by automating the mundane tasks.
- The solution needs to be vendor-agnostic digital pathology platform that enables digitization of the entire pathology lab workflow.
- It should automate the analysis of Whole Slide Scanner images and support annotations on images for clinical use
- The solution needs to seamlessly connect with LIMS and HIMS systems
- The solution should be compatible with PACS systems and DICOM protocol
- It should produce a rich diagnostic report and have the capabilities for workflow automation and telepathology
- The solution should be able to create datasets for AI based diagnosis
Pilot Deployment :
The solution was deployed at Rajiv Gandhi Cancer Institute and Research Center (RGCIRC). Normally, it takes weeks for a cancer patient to get lab results. The AI based solution reduces the TAT from a week to a day to produce a report for Prostate Cancer and improves accuracy of diagnosis. Pathologist has to largely focus on verifying the report generated by the solution.
About the Author: admin
Related posts
5th Edition of LHIF: Digital Propulsion Of Universal Healthcare
We are pleased to announce the 5th edition of NASSCOM CoE IoT’s flagship healthcare initiative Lifesciences & Healthcare Innovation Forum, which will be focused on Digital ...
