A computer-aided detection and localization system could fine-tune diagnosis of small hepatocellular carcinomas with multislice CT, according to a study presented at the European Congress of Radiology in March.
A computer-aided detection and localization system could fine-tune diagnosis of small hepatocellular carcinomas with multislice CT, according to a study presented at the European Congress of Radiology in March.
Features of some liver lesions may be hard to spot or characterize with contrast-enhanced MSCT studies. An automated, accurate localization of these lesions during multiple scanning phases could make the evaluation of their enhancement patterns and diagnosis more efficient.
Guo-Qing Wei, Ph.D., and colleagues at the research division of Siemens Medical Solutions proposed a computer application to detect small liver cancers during multiple-phase contrast-enhanced CT scanning. Using 35 studies from cancer patients as a reference, they devised a technique to isolate suspected liver lesions smaller than 30 mm from nonliver tissue. This technique accurately registered small hepatocellular carcinomas across MSCT liver scans.
Three imaging specialists marked 41 small lesions with a suspicious appearance by consensus. They used a segmentation technique to reconstruct a liver volume in each contrast phase. They then applied an intensity-weighting method to suppress nonliver anatomies and employed a cross-correlation technique to find corresponding slices across different phases.
The investigators obtained detection rates of 82.9%, 90.2%, and 78% between precontrast/arterial, arterial/portal venous, and across simultaneous phases, respectively. They recorded a mean registration error of 0.4 slices.
Physicians could use the proposed intensity-weighted cross-correlation method to develop efficient detection tools to bolster diagnosis of small hepatocellular carcinomas, the researchers said.
Wei's study adds an alternative approach to the findings of Dr. Yu-Len Huang, a radiologist at the China Medical University Hospital in Taichung, Taiwan. Huang and colleagues tried a computerized model that accurately characterizes benign from malignant liver tumors without the need for contrast agents, which could avoid renal toxicity and allergic reactions in certain patients (Mid Taiwan J Med 2004;9:141-150).
For more information from the Diagnostic Imaging archives:
Recent research developments address RFA of the liver
CAD acquisition parameters prove critical to accuracy
CAD falls within cost-effectiveness range
Considering Breast- and Lesion-Level Assessments with Mammography AI: What New Research Reveals
June 27th 2025While there was a decline of AUC for mammography AI software from breast-level assessments to lesion-level evaluation, the authors of a new study, involving 1,200 women, found that AI offered over a seven percent higher AUC for lesion-level interpretation in comparison to unassisted expert readers.
Contrast-Enhanced Mammography and High-Concentration ICM Dosing: What a New Study Reveals
June 16th 2025New research showed a 96 to 97 percent sensitivity for contrast-enhanced mammography (CEM) with an increased iodine delivery rate facilitating robust contrast enhancement for women with aggressive breast cancer.
Mammography AI Platform for Five-Year Breast Cancer Risk Prediction Gets FDA De Novo Authorization
June 2nd 2025Through AI recognition of subtle patterns in breast tissue on screening mammograms, the Clairity Breast software reportedly provides validated risk scoring for predicting one’s five-year risk of breast cancer.