Not for publication in the USA
Siemens Healthineers now offers new AI-based features with its mammography reading and reporting solution Syngo.Breast Care*. It provides physicians with an innovative type of interactive decision support for faster, more accurate interpretation of mammography images. This enables radiologists to deliver results within a shorter period of time, transforms their reading workflows and ultimately improves care delivery.
AI can help make physicians’ work much easier, particularly in the area of cancer screening. Numerous mammograms are performed on a daily basis to screen for breast cancer, which means that radiologists have to interpret several hundred images a day accurately and under time pressure. Tomosynthesis, which takes 3D images of the breast, is being utilized more in screening and therefore adds to the number of images to be read. With the new version of Syngo.Breast Care, Siemens Healthineers can now offer radiologists interactive clinical decision support that makes reading mammography images both faster and more accurate.
AI-based algorithms help evaluate individual lesions more precisely in order to minimize the number of false positive findings. If radiologists detect an abnormality in a 2D mammogram or in a 3D tomosynthesis, they can simply click on the suspicious area to obtain an evaluation from the system on the probability that the lesion or mass is malignant. This so-called lesion score uses color coding to indicate the probability that a tissue alteration is malignant, thus helping radiologists to verify their suspicions. A peer-reviewed scientific study demonstrated convincing results for the AI-based support, showing an increase in sensitivity and specificity.1
In addition, the software automatically sorts the cases according to their probability for breast cancer. It calculates the case score from an overall ranking of any existing lesions, microcalcifications, and other abnormalities. The score ranges from 1 to 10, where a score of 1 indicates a very low probability of cancer and 10 indicates a relatively high probability. This high correlation between case score and probability of breast cancer was proven in a clinical study1.
Syngo.Breast Care’s new SmartSort technology enables radiologists to rank exams according to their preferences based on these case scores. For example, critical cases can be immediately moved to the top so that they are given highest priority. The prioritizing of cases could also be helpful for optimizing the workflows for a double-blind reading by a second expert or for interdisciplinary consensus.
Siemens Healthineers partnered with ScreenPoint Medical to integrate interactive decision support in Syngo.Breast Care. This company’s highly innovative mammography reading software, Transpara**, is based on deep learning and has been trained with over a million images.
Link to the related press feature
For further information on AI at Siemens Healthineers, please see www.healthcare.siemens.com/infrastructure-it/artificial-intelligence
* Syngo.Breast Care VB40 - powered by TransparaTM, ScreenPoint Medical - is currently under development. It is not for sale in the U.S. Its future availability cannot be guaranteed.
** Transpara is currently 510(k) pending for 2D reading.
Rodriguez-Ruiz A., Lång K., Gubern-Merida A., et al.: Detecting breast cancer in mammography: a deep learning-based computer system versus 101 radiologists (Journal of the National Cancer Institute 2018 (Accepted).