Beckman Researchers Developed AI Model to Identify Tumors & Diseases with Visual Maps

This AI model is designed to identify tumors and diseases within medical images aiming to detect disease in budding stages.

Researchers from the USA’s Beckman Institute for Advanced Science and Technology have developed an AI model that can accurately identify tumors and diseases with visual maps.

This model offers an advanced level of transparency by explaining each diagnosis with a visual map, allowing doctors to easily follow its line of reasoning, double-check for accuracy, and explain the results to patients.

Sharing thoughts, Sourya Sengupta, lead author of the research, said, "The idea is to help catch cancer and disease in its earliest stages -- like an X on a map -- and understand how the decision was made. Our model will help streamline that process and make it easier on doctors and patients alike."

While cancer detection is a primary focus, the model’s applications extend to various diseases. From neurological conditions to cardiovascular issues, its visual maps offer diagnostic precision.

Mapping Disease: How It Works

This AI model is designed to identify tumors and diseases within medical images aiming to detect disease in budding stages. Additionally, it offers an explanation for each diagnosis through visual maps.

When presented with medical images such as X-rays, MRIs, or CT scans, the AI model analyzes them. Thereafter, it pinpoints anomalies, highlights potential areas of concern, and provides a diagnosis.

Moreover, the model generates visual maps. These maps vividly illustrate the reasoning behind its diagnosis. Doctors can follow the map step by step, understanding precisely why the model arrived at a particular conclusion.

Applications & Impact

The new AI model is expected to have a significant impact, particularly in regions with a scarcity of doctors and long patient queues.

"When time and talent are in high demand, automated medical image screening can be deployed as an assistive tool - in no way replacing the skill and expertise of doctors," Sengupta added.

The model's capability to pre-scan medical images and identify abnormalities, such as tumors or early signs of disease, and then flag them for review by a doctor, is expected to streamline the diagnostic process, saving time and enhancing the efficiency of scan interpretation.

Further, researchers trained the model across three distinct disease diagnosis tasks, utilizing a dataset comprising over 20,000 images. Notably, the model exhibited impressive performance metrics, achieving accuracy rates of 77.8% for mammograms, 99.1% for retinal optical coherence tomography (OCT) images, and 83% for chest X-rays.

These high accuracy rates are attributed to the AI's deep neural network, which mimics the intricacies of human neurons in decision-making processes. This study was published in IEEE Transactions on Medical Imaging.

Founded in 1989 by Arnold Beckman and Prof Ted Brown, the Beckman Institute for Advanced Science and Technology is a premier interdisciplinary research institute in the United States. It aims to foster collaboration across scientific fields, blur traditional boundaries, and drive groundbreaking discoveries.

Going forward, the institute plans to continue pushing the envelope of scientific innovation. The Beckman Institute is affiliated with the University of Illinois Urbana-Champaign, US.

In November, Bengaluru-based startup Vizzhy Inc invested $12.5 million in the University of Illinois Urbana-Champaign to address the underlying issues in metabolic diseases.

Notably, a Memorandum of Understanding (MoU) between the University of Illinois Urbana-Champaign and Vizzhy Inc was recently signed. With this investment, the startup sought to explore new avenues in disease management.