New AI Identifies Nearly 100% Of Cancer, Outperforming Doctors

Artificial intelligence is quickly becoming a powerful tool in detecting cancer with greater precision than the human eye. A recent AI model has demonstrated an impressive near-perfect success rate, offering a glimpse into the future of cancer diagnosis.

Global Team Develops ECgMPL AI Model for Endometrial Cancer Detection

A global team of scientists, including experts from Charles Darwin University (CDU) in Australia, has developed an advanced AI model called ECgMPL. This model analyzes microscopic images of cells and tissues to identify endometrial cancer, one of the most common types of reproductive cancers, with remarkable 99.26% accuracy.The researchers also believe that this model can detect a wide range of other cancers, including colorectal and oral cancer.

According to Dr. Asif Karim, co-author of the study from CDU, “The ECgMLP model outperforms existing methods by achieving 99.26% accuracy, surpassing transfer learning and custom models while remaining computationally efficient.” He adds, “Optimized through studies and self-attention mechanisms, ECgMLP generalizes well across multiple histopathology datasets, making it a robust and clinically applicable solution for diagnosing endometrial cancer.”

Essentially, the AI model analyzes microscopic scans – histopathology images – and enhances their quality to detect early signs of cancer. It focuses on specific areas of the scans to identify growths that might be difficult for human eyes to see. At present, human-led diagnostic methods have an accuracy range of 78.91% to 80.93%. Early detection of endometrial cancer is critical, as treatment success rates are high when it’s detected early. However, once the cancer spreads, treatment becomes much harder, highlighting the importance of timely diagnosis.

Broad Potential of AI in Detecting Other Diseases and Cancers

Endometrial cancer has affected over 600,000 Americans. Although this form of cancer may not impact everyone equally, the scientists note that the ECgMLP model has broad potential for other diseases as well.

Co-author Niusha Shafiabady, an associate professor at ACU, states, “We can apply this methodology for rapid and accurate early detection of various diseases, ultimately improving patient outcomes.” The team also evaluated the model on several other cancer datasets, achieving high accuracy: 98.57% for colorectal cancer, 98.20% for breast cancer, and 97.34% for oral cancer.

It’s important to note that this AI model does not replace medical professionals but complements their expertise by helping cancer specialists detect and monitor the disease more accurately. Moreover, this AI-driven method provides a faster, more accessible, and cost-effective way to diagnose cancer.

Shafiabady adds, “We can integrate this core AI model into a software system to assist doctors in decision-making during cancer diagnosis.

The researchers emphasize that early and accurate detection of endometrial cancer is vital for effective treatment. Using deep learning algorithms to classify histopathological images has proven to deliver superior accuracy and processing time.

Read the original article on: New Atlas

Read more: AI Assists Doctors in Identifying More Cases of Breast Cancer in the Largest Real-World Study