New ultrasound CAD scheme achieves high accuracy

Sunday, November 27 | 12:05 p.m.-12:15 p.m. | SSA21-09 | Room S504AB
In this scientific session, a Japanese research team will discuss its computer-aided detection (CAD) scheme, which is based on histologic classifications of breast ultrasound images.

Reducing the number of unnecessary breast biopsies may be possible if decisions to biopsy or follow up breast masses could take into account possible histological classifications on ultrasound images, said presenter Akiyoshi Hizukuri of Mie University.

For example, patients with breast masses associated with invasive carcinomas should be biopsied immediately, while patients with a noninvasive carcinoma type that grows rapidly should undergo biopsy or follow-up within a few months, Hizukuri said. Those with cysts and fibroadenomas can have follow-up at a relatively long interval of six months.

"Computerized analysis for evaluating the likelihood of histological classifications and also the likelihood of malignancy for breast masses would be helpful to clinicians for their decisions on patient management," he said.

The researchers developed a CAD scheme for determining histological classification of masses on ultrasound images and mammograms. In contrast with other traditional CAD methods, the group's CAD algorithm makes use of objective features based on clinicians' subjective impressions of image features.

The end result is a CAD scheme that achieved high classification accuracies, according to the researchers. Learn all the details in their presentation.

Page 1 of 509
Next Page