Techniques for detecting in real-time proton beam therapy three dimensional dose delivered

A non-transitory computer-readable medium carrying one or more sequences of instructions, wherein execution of the one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

• retrieving conversion data that indicates values for parameters that convert proton dose to acoustic pressure based on values at voxels in medical imagery;
• retrieving a 3D medical image of a first region of interest inside a first subject;
• receiving a real-time signal that indicates ultrasound pressure received at a two-dimensional ultrasound transducer array disposed outside the first subject in contact with the subject close to the region of interest;
• and
• determining a real-time 3D dose delivered in the region of interest inside the subject by inputting the real-time signal into a dose detection module that includes a neural network trained on a plurality of training set instances, each instance including simulated proton dose delivered to a similar region of interest inside a second subject and simulated pressure generation from that simulated dose based on the conversion data and a similar 3D medical image and simulated acoustic transmission from the similar region of interest to a similar two-dimensional ultrasound transducer array similarly disposed.

An apparatus comprising:

• at least one processor;
• and
• at least one computer-readable medium including one or more sequences of instructions and one or more data structures,
• the at least one computer-readable medium and the one or more sequences of instructions configured to, with the at least one processor, cause the apparatus to perform at least the following,
• retrieving, from the one or more data structures, conversion data that indicates values for parameters that convert proton dose to acoustic pressure based on values at voxels in medical imagery;
• retrieving, from the one or more data structures, a 3D medical image of a first region of interest inside a first subject;
• receiving a real-time signal that indicates ultrasound pressure received at a two-dimensional ultrasound transducer array disposed outside the first subject in contact with the subject close to the region of interest;
• and
• determining a real-time 3D dose delivered in the first region of interest inside the first subject by inputting the real-time signal into a dose detection module that includes a neural network trained on a plurality of training set instances, each instance including simulated proton dose delivered to a similar region of interest inside a second subject and simulated pressure generation from that simulated dose based on the conversion data and a similar 3D medical image and simulated acoustic transmission from the similar region of interest to a similar two-dimensional ultrasound transducer array similarly disposed.

A system for real-time detection of dose delivered to tissue by a proton beam, wherein real-time indicates within ten seconds of actual occurrence, the system comprising:

• a two-dimensional ultrasound transducer array disposed outside a first subject in contact with the first subject close to a first region of interest inside the first subject;
• a processor;
• a computer-readable medium carrying one or more sequences of instructions and storing one or more data structures, wherein execution of the one or more sequences of instructions by the processor causes the processor to perform the steps of
• retrieving, from the data structure, conversion data that indicates values for parameters that convert proton dose to acoustic pressure based on values at voxels in medical imagery;
• retrieving, from the data structure, a 3D medical image of the first region of interest inside the first subject;
• receiving a real-time signal that indicates ultrasound pressure received at the two-dimensional ultrasound transducer array;
• and
• determining a real-time 3D dose delivered in the first region of interest inside the first subject by inputting the real-time signal into a dose detection module that includes a neural network trained on a plurality of training set instances, each instance including simulated proton dose delivered to a similar region of interest inside a second subject and simulated pressure generation from that simulated dose based on the conversion data and a similar 3D medical image and simulated acoustic transmission from the similar region of interest to a similar two-dimensional ultrasound transducer array similarly disposed.