IP Library › Granted Patent US 12711356
Granted Patent B2
US 12711356 · App. 18/194,216 · Granted Aug 18, 2026

Generation and application of radiation dosage based on neural network architecture

Inventor: Esa Kuusela (Espoo, FI)
Assignee: Siemens Healthineers International AG
G06N3/045G06N3/084
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12711356
App. No.
18/194,216
Granted
Aug 18, 2026
Kind
B2
Abstract

Aspects of this technical solution can include generating, by a processor, a non-linear output of a layer of a first model including a first neural network, the layer of the first model including a non-linear operator and corresponding to a distribution of matter, generating, by the processor, a linear output based on a layer of a second model including a second neural network and the non-linear output, the layer of the second model including a linear operator and corresponding to a plurality of beams respectively configured to generate radiation, outputting, by the processor and based on the linear response, an indication of a distribution of energy output by the plurality of beams to correspond to the distribution of matter, and causing, by the processor, one or more of the plurality of beams to output radiation according to the distribution of energy output.

Claims (45)

1 . A method, comprising:

generating, by a processor, a non-linear output of a layer of a first model including a first neural network, the layer of the first model including a non-linear operator and corresponding to a distribution of matter;

generating, by the processor, a linear output based on a layer of a second model including a second neural network and the non-linear output, the layer of the second model including a linear operator and corresponding to a plurality of beams respectively configured to generate radiation, wherein the linear output is proportional to a fluence input to the second neural network;

outputting, by the processor and based on the linear output, an indication of a distribution of energy output by the plurality of beams to correspond to the distribution of matter; and

causing, by the processor, one or more of the plurality of beams to output radiation according to the distribution of energy output.

2 . The method of claim 1 , further comprising:

inputting, by the processor to the layer of the first model, a non-linear feedback corresponding to the non-linear output.

3 . The method of claim 1 , further comprising:

inputting, by the processor to the layer of the second model, a linear feedback corresponding to the linear output.

4 . The method of claim 1 , further comprising;

inputting, by the processor to the layer of the second model, a non-linear feedback corresponding to the non-linear output, the non-linear feedback corresponding to a response by the distribution of matter to the distribution of energy.

5 . The method of claim 1 , wherein the layer of the first model corresponds to a two-dimensional portion of the distribution of matter at a first distance from source within the distribution of matter.

6 . The method of claim 5 , wherein the layer of the second model corresponds to a two-dimensional portion of the distribution of energy at the first distance from source within the distribution of matter.

7 . The method of claim 5 , wherein the layer of the second model corresponds to a two-dimensional portion of the distribution of energy at a second distance from source within the distribution of matter.

8 . The method of claim 7 , wherein the first distance from source is greater than the second distance from source.

9 . The method of claim 1 , wherein one or more of the first neural network and the second neural network correspond to a recursive neural network.

10 . A system, comprising:

a server comprising a processor and a non-transitory computer-readable medium containing instructions that when executed by the processor causes the processor to:

generate a non-linear output of a layer of a first model including a first neural network, the layer of the first model including a non-linear operator and corresponding to a distribution of matter;

generate a linear output based on a layer of a second model including a second neural network and the non-linear output, the layer of the second model including a linear operator and corresponding to a plurality of beams respectively configured to generate radiation, wherein the linear output is proportional to a fluence input to the second neural network;

output, based on the linear output, an indication of a distribution of energy output by the plurality of beams to correspond to the distribution of matter; and

cause one or more of the plurality of beams to output radiation according to the distribution of energy output.

11 . The system of claim 10 , the processor further configured to:

input, to the layer of the first model, a non-linear feedback corresponding to the non-linear output.

12 . The system of claim 10 , the processor further configured to:

input, to the layer of the second model, a linear feedback corresponding to the linear output.

13 . The system of claim 10 , the processor further configured to:

input, to the layer of the second model, a non-linear feedback corresponding to the non-linear output, the non-linear feedback corresponding to a response by the distribution of matter to the distribution of energy.

14 . The system of claim 10 , wherein the layer of the first model corresponds to a two-dimensional portion of the distribution of matter at a first distance from source within the distribution of matter.

15 . The system of claim 14 , wherein the layer of the second model corresponds to a two-dimensional portion of the distribution of energy at the first distance from source within the distribution of matter.

16 . The system of claim 14 , wherein the layer of the second model corresponds to a two-dimensional portion of the distribution of energy at a second distance from source within the distribution of matter.

17 . The system of claim 16 , wherein the first distance from source is greater than the second distance from source.

18 . The system of claim 10 , wherein one or more of the first neural network and the second neural network correspond to a recursive neural network.

19 . A system comprising:

a computer in communication with a server and configured to display a graphical user interface;

a radiotherapy machine in communication with the server; and

the server configured to:

generate a non-linear output of a layer of a first model including a first neural network, the layer of the first model including a non-linear operator and corresponding to a distribution of matter;

generate a linear output based on a layer of a second model including a second neural network and the non-linear output, the layer of the second model including a linear operator and corresponding to a plurality of beams respectively configured to generate radiation, wherein the linear output is proportional to a fluence input to the second neural network;

output, based on the linear output, an indication of a distribution of energy output by the plurality of beams to correspond to the distribution of matter; and

cause one or more of the plurality of beams to output radiation according to the distribution of energy output.

20 . The system of claim 19 , the computer further configured to:

input, to the layer of the first model, a non-linear feedback corresponding to the non-linear output;

input, to the layer of the second model, a linear feedback corresponding to the linear output; and

input, to the layer of the second model, a non-linear feedback corresponding to the non-linear output, the non-linear feedback corresponding to a response by the distribution of matter to the distribution of energy.