Given the high workload of screening programs, the benefit of using Transpara is further enhanced by the fact that it does not increase reading time.
CHICAGO (PRWEB) November 20, 2018
Radiologists significantly improved their cancer detection in mammography without increasing reading times when using Transpara™ software from ScreenPoint Medical, according to results of a study published with open access in Radiology.
In the study, “Detection of breast cancer using mammography: Impact of an Artificial Intelligence support system,” researchers compared breast cancer detection performance of radiologists conducting a single-read of mammography exams to performance when supported by the Transpara artificial intelligence (AI) system. Screening mammograms were interpreted by 14 radiologists, once with and once without AI support, providing a BI-RADS® score and probability of malignancy. Diagnostic performance and reading times were compared.
Radiologists improved their overall diagnostic performance when using Transpara, with the average AUC increasing from 0.87 to 0.89 (P = 0.002). Results showed that sensitivity increased from 83 to 86 percent and specificity increased from 77 to 79 percent when using Transpara as a decision support tool. Use of Transpara did not slow down the readers but decreased their average reading time per case. Overall the reduction was 4.5 percent, which was due to faster reading of exams marked with low-suspicion by Transpara.
The results suggest that the use of Transpara for decision support might prevent missed cancers and interpretation errors that are relatively common in the reading of mammography. Having a stand-alone sensitivity and specificity similar to that of radiologists, use of the Transpara software in combination with single-reading might achieve a performance similar to double human reading. According to the researchers, the improvement in diagnostic performance was most prevalent in the evaluation of equivocal cases, underscoring the potential clinical relevance of Transpara.
“Given the high workload of screening programs, the benefit of using Transpara is further enhanced by the fact that it does not increase reading time. The system is designed for concurrent use, which means that radiologists are supported while reading an exam rather than after finishing their interpretation, which is how traditional mammography CAD systems work. Radiologists like this approach and get confidence in the system as they work with it, when they recognize that the system may be as good as themselves.” said Prof. Nico Karssemeijer, founder and CEO of ScreenPoint Medical.
Utilizing state of the art image analysis and revolutionary deep learning technology, Transpara automatically identifies soft-tissue and calcification lesions and combines the findings of all available views into a single cancer suspiciousness score. Interactive decision support is a proven method to boost reading performance for soft tissue lesions. In addition, Transpara combines analysis of soft tissue lesions and calcifications, if present, from all available views of an exam to compute a single score for the case on a scale of 1 to 10. This represents categories with increasing occurrence of cancer. Breast imaging professionals can use this Transpara Score to automatically identify exams that are highly likely to be normal or the exams which are most likely to have suspicious findings.
Transpara has European regulatory approval (CE Mark) for use with digital mammography and digital breast tomosynthesis (DBT) and is compatible with images from multiple vendors Regulatory clearance for marketing in the U.S. is expected soon. The system is installed in leading centers throughout Europe.
About ScreenPoint Medical BV
ScreenPoint Medical develops image analysis technology for automated reading of mammograms and digital breast tomosynthesis exams, exploiting Big Data, Deep Learning and the latest developments in Artificial Intelligence. ScreenPoint Medical was founded in 2014 by Nico Karssemeijer and Michael Brady, two experts in breast imaging, machine learning, computer vision, and computer-aided detection. The main office is in Nijmegen, The Netherlands.
Chris K. Joseph