Grant Number:	1R01CA305903-01 Interpret this number
Primary Investigator:	Debes, Jose
Organization:	University Of Minnesota
Project Title:	Assessment of Sound Waves From Liver Vessels Via Artificial Intelligence to Detect Liver Cancer
Fiscal Year:	2026

Hepatocellular carcinoma (HCC) is the second leading cause of cancer-related death worldwide. Hepatitis B and C are the most common causes of HCC worldwide. The majority of deaths related to HCC occur in resource-limited settings such as South America where viral hepatitis is frequent. HCC mortality occurs primarily due to late detection, which precludes potentially curative interventions. Currently, individuals with liver disease are advised to undergo liver ultrasonography every 6 months with the goal of “visually” identifying a tumor small enough that might be amenable to cure. This visual screening approach is not optimal as ultrasound is operator dependent. Moreover, the sensitivity of ultrasound in screening for early HCC is approximately 50%, which is far from optimal. Therefore, there is a need to find easy-to-apply HCC screening methods which are effective, low-cost, scalable to a population, and require no advanced expertise. Artificial Intelligence (AI) is widely studied in its potential to diagnose cancer, usually attempting to enhance the detection accuracy of a mass that is already detectable by the human eye. The liver is mainly vascularized by the portal vein. As a primary liver cancer (HCC) develops small hepatic arteries grow within the tumor, providing arterial blood flow. Thus, the normal liver received blood from the portal vein while liver cancer (HCC) does so from the hepatic artery, effectively dividing liver and tumor into two vascular systems and changing blood flow of the main liver vessels. These changes are not perceptible to the human eye. We hypothesize that assessment of sound waves of the main liver vessels through Doppler ultrasonography, which can be detected with mobile-phone ultrasonography, and analyzed via AI will detect changes that can differentiate livers with HCC from those with no HCC. Our preliminary data show that a simple Doppler ultrasound assessment of the portal vein with our AI reading can detect the presence of HCC with a 90% sensitivity – an almost 2-fold improvement over ultrasound alone. Our aims are to #1 assess the role of Doppler ultrasonography sound waves analyzed via AI in differentiating livers with cancer versus controls, refining the algorithm in our 6000 ultrasound database; #2 determine the efficacy of Doppler ultrasonography in detecting HCC in resource-limited settings in a separate validation cohort, training and supervising local staff in Porto Alegre, Brazil to evaluate the algorithm in an existing 2000 ultrasound database and; #3 prospectively validate the real-life role of mobile-phone Doppler ultrasound analyzed via AI in screening for HCC in resource-limited settings, through our ongoing ESCALON HCC clinics in Brazil and Ecuador. If successful, our project has the potential to revolutionize the way we detect liver cancer. Moreover, this methodology could eventually be modified and applied to the screening of other cancers around the globe.





None