About the Customer
Lunit, a portmanteau of ‘Learning unit,’ is a medical AI software company devoted to providing AI-powered total cancer care. Their AI solutions help discover cancer and predict cancer treatment outcomes, achieving timely and individually tailored cancer treatment.
Founded in 2013, their accolade journey started in 2016 with 1st place in MICCAI Tumour Proliferation Assessment Challenge and gained impressive momentum. They were listed in the Top 5 “Social Impact” AI start-ups, by NVIDIA in 2017, AI 100 by CB Insights in 2017, and selected as Technology Pioneer by World Economic Forum in 2020.
Lunit values in building clinical evidence through publishing studies in major peer-reviewed journals. Their findings in AI detection on chest x-ray and mammography are published in RSNA Radiology Journal, Lancet Digital Health, JAMA Network Open, Clinical Infectious Diseases, and so on.
Lunit is headquartered in Seoul, South Korea with offices and representatives in the U.S.A. (Boston), Netherlands (Amsterdam), and China (Shanghai)
The Pain Point
They achieved CE mark for CXR (Chest Xray algorithm) and MMG (Mammography algorithm) in 2019 and 2020, respectively. The target for 2021 was to take their flagship Lunit INSIGHT products outside South Korea, initiating their global adoption through a US FDA clearance.
The journey towards acquiring an FDA clearance is long and tortuous, especially in the AI domain. Lunit had an ambitious target of acquiring 2 FDA clearances for their CXR and MMG products in 2021, which were eventually cleared at the end of the year. They had spent two years trying to gather resources to execute the validation studies needed for filing the approval.
The overwhelming list of resources they needed for executing their study included:
- High-Quality Data
- Clinical Experts
- Study Management Experts
- AI Performance Report Creation Tools
- A Robust Reading Platform with Annotation and Validation Tools
- Timely & Efficient Execution of Study
They were unable to find a single integrated solution to cater to all the listed requirements.
Lunit wanted to access all the required resources through a single collaborating platform and decided to partner with CARPL – A Single Window AI Platform that Helps Speed Up FDA Studies.
The INSIGHT CXR study was executed on the CARPL platform. From preliminary readings to final Ground Truth (GT) annotations, CARPL enabled seamless execution conducting multi-reader, feasibility studies.
CARPL team was responsible for
- Identification and verification of the data cohorts in accordance with the study criterion.
- Recruitment & training of principal investigator & other readers of the study
- Customization and study flow optimization of the CARPL Platform
- Execution of the study, verification of the study data, and audit of the statistical results
The entire study was performed on the CARPL Platform which offers
- Easy and secure data management
- Single window study level and pixel level annotations
- Comprehensive logging of study reader actions
- Intuitive statistical tools
- Executing end to end pivotal study
- Executing end to end pilot study
- Building validation & performance reports
These studies were enabled by platform features needed for FDA studies that CARPL is built on –
- Dataset Management – Data sourcing, management, and cohort creation – CARPL’s extensive feature-set addresses all data concerns for executing FDA studies
- Single interface for annotation – CARPL enables seamless annotations by multiple readers, in real time, with the ability to review/accept/reject annotations
- End-to-end AI validation platform – automated testing, bias estimation – doctors build trust in AI using CARPL
We stay true to our mission of bringing AI from Bench to Clinic and are glad that we partner with Lunit – one of the world’s leading AI developers – in their journey.
RSNA AI Theatre 2021
Brandon Suh (Chief Executive Officer, Lunit), gave a talk at the RSNA AI theatre about how CARPL has helped both, Lunit’s regulatory and commercial teams, accelerate time to market.