eduzhai > Applied Sciences > Engineering >

Learning to Scan A Deep Reinforcement Learning Approach for Personalized Scanning in CT Imaging

  • Save

... pages left unread,continue reading

Document pages: 18 pages

Abstract: Computed Tomography (CT) takes X-ray measurements on the subjects toreconstruct tomographic images. As X-ray is radioactive, it is desirable tocontrol the total amount of dose of X-ray for safety concerns. Therefore, wecan only select a limited number of measurement angles and assign each of themlimited amount of dose. Traditional methods such as compressed sensing usuallyrandomly select the angles and equally distribute the allowed dose on them. Inmost CT reconstruction models, the emphasize is on designing effective imagerepresentations, while much less emphasize is on improving the scanningstrategy. The simple scanning strategy of random angle selection and equal dosedistribution performs well in general, but they may not be ideal for eachindividual subject. It is more desirable to design a personalized scanningstrategy for each subject to obtain better reconstruction result. In thispaper, we propose to use Reinforcement Learning (RL) to learn a personalizedscanning policy to select the angles and the dose at each chosen angle for eachindividual subject. We first formulate the CT scanning process as an MDP, andthen use modern deep RL methods to solve it. The learned personalized scanningstrategy not only leads to better reconstruction results, but also shows stronggeneralization to be combined with different reconstruction algorithms.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...