Artificial Intelligence on the impact of CT dose reduction

Posted on: 2020-03-02

Artificial intelligence has already impacted several areas of medicine; the time has come for exposure to radiation from computed tomography images. According to an article published in Health Imaging, about a study published in Nature Machine Intelligence, “a new approach to deep learning reduced the radiation exposure of computed tomography images and produced higher quality scans than traditional iterative reconstruction techniques”.

In addition, the AI approach – which uses a modularized neural network – was also much faster than traditional methods. In the study, the co-author Ge Wang’s method was compared with the commercial interactive reconstruction techniques of three leading low-dose CT providers. And yet, according to the researchers, the study emphasizes the importance of seeking deep learning to obtain more efficient and safe images.

There were 60 exams of patients removed from Massachusetts General Hospital in Boston; half underwent routine abdominal CT and the rest received a routine chest CT scan on one of the three commercial scanners. Three radiologists analyzed and classified the images for two characteristics: structural fidelity and noise suppression in the image.

When observing abdominal images, radiologists assigned higher scores to images created using the modularized neural network in two of the three scanners; the third device was considered comparable to the iterative construction method. On chest images, readers determined that the image quality was comparable between CT methods and across all devices.

"Promote through education safe diagnostic imaging in Latin America with emphasis on radiological protection"

News Sign our Commitment Letter now!