AI Detects Over Half of Metastases Overlooked by Radiologists on CT

Early detection of liver metastases is critical for improving patient outcomes, but physicians may sometimes miss the most common malignant liver tumors, resulting in delayed or missed treatment opportunities. Now, a new study has discovered that artificial intelligence (AI) can detect over 50% of liver metastases that radiologists overlook on contrast-enhanced CT scans.

Scientists at Kyoto University Graduate School of Medicine (Kyoto, Japan) aimed to determine whether AI could assist radiologists in reducing the number of missed cases. They reviewed the records of close to 750 patients diagnosed with liver metastases at their institution between 2010-2017. Two expert abdominal radiologists categorized cases as either missed or correctly identified by physicians at the time. This led to a final sample of approximately 135 patients, with 68 classified as having overlooked metastases.

The AI software was then used to process the 100-plus images and successfully detected liver metastases in 54% of patients whose findings were missed by human readers. The per-lesion sensitivity was 70.1% across all liver lesion types, 70.8% for metastases, and 55% for those overlooked by radiologists. Overall, the AI tool identified metastases in 92.7% of all cases and 53.7% of overlooked instances, with an average of around 0.48 false positives per patient.

Both the software and radiologists commonly failed to detect metastases with smaller size, low contrast, and background fatty liver. The researchers suggest that the gap between AI and physicians could be attributed to factors such as the purpose of the examination, with CT scans for non-liver-related concerns sometimes resulting in misses, and the radiologists' physical and mental condition, including fatigue, anxiety, or lack of concentration. The researchers also noted that the software failed to detect 16% of liver metastases identified by radiologists, emphasizing its role as a support tool rather than a replacement.

“Our results suggest the potential of AI-powered software in reducing the frequency of overlooked small liver metastases when used in conjunction with the radiologists’ clinical interpretation,” concluded Hirotsugu Nakai, with the Kyoto University Graduate School of Medicine.

Related Links:
Kyoto University Graduate School of Medicine

Visit expo >

New

Gold Member

X-Ray QA Meter

T3 AD Pro

New

Portable Color Doppler Ultrasound System

S5000

Portable X-ray Unit

AJEX130HN

New

Mobile Barrier

Tilted Mobile Leaded Barrier