IP Library › Granted Patent US 12,580,063
Granted Patent B2
US 12,580,063 · App. 17/414,937 · Granted Mar 17, 2026

Methods and systems for radiotherapy treatment planning based on deep transfer learning

Inventors: Hannu Mikael Laaksonen (Espoo, FI); Sami Petri Perttu (Helsinki, FI); Tomi Ruokola (Espoo, FI); Jan Schreier (Helsinki, FI); Janne Nord (Espoo, FI)
Assignee: SIEMENS HEALTHINEERS INTERNATIONAL AG
G16H20/40A61N5/103G06N3/08
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 12,580,063
App. No.
17/414,937
Granted
Mar 17, 2026
Kind
B2
Abstract

Example methods and systems for deep transfer learning for radiotherapy treatment planning are provided. One example method may comprise: obtaining ( 310 ) a base deep learning engine that is pre-trained to perform a base radiotherapy treatment planning task; and based on the base deep learning engine, generating a target deep learning engine to perform a target radiotherapy treatment planning task. The target deep learning engine may be generated by configuring ( 330 ) a variable base layer among multiple base layers of the base deep learning engine, and generating ( 340 ) one of multiple target layers of the target deep learning engine by modifying the variable base layer. Alternatively or additionally, the target deep learning engine may be generated by configuring ( 350 ) an invariable base layer among the multiple base layers, and generating ( 360 ) one of multiple target layers of the target deep learning engine based on feature data generated using the invariable base layer.

Claims (64)

1 . A method for a computer system, interfaced with a treatment delivery system, to perform deep transfer learning for radiotherapy treatment planning, wherein the method comprises:

obtaining, by the computer system, a base deep learning engine that is pre-trained to perform a base radiotherapy treatment planning task, wherein the base deep learning engine includes multiple base layers; and

based on the base deep learning engine, generating, by the computer system, a target deep learning engine to perform a target radiotherapy treatment planning task by performing at least one of the following, wherein the base deep learning engine is trained based on base training data associated with the base radiotherapy treatment planning task in a first phase, the target deep learning engine is trained based on target training data associated with the target radiotherapy treatment planning task in a second phase later than the first phase, and the target radiotherapy treatment planning task and the base radiotherapy treatment planning task are two different radiotherapy treatment planning tasks:

configuring a variable base layer among the multiple base layers of the base deep learning engine, and generating one of multiple target layers of the target deep learning engine by modifying the variable base layer; and

configuring an invariable base layer among the multiple base layers of the base deep learning engine, and generating one of multiple target layers of the target deep learning engine based on feature data generated using the invariable base layer;

performing, by the computer system, the target radiotherapy treatment planning task to generate output data based on input data associated with a particular patient in a third phase using the target deep learning engine, wherein the third phase is later than the second phase; and

generating, by the computer system, a treatment plan based on the output data, wherein the treatment delivery system retrieves the treatment plan and delivers a radiotherapy treatment to the particular patient according to the treatment plan.

2 . The method of claim 1 , wherein generating the target deep learning engine comprises:

based on the base deep learning engine that is pre-trained to perform the base radiotherapy treatment planning task associated with a base anatomical site, generating the target deep learning engine to perform the target radiotherapy treatment planning task associated with a target anatomical site.

3 . The method of claim 1 , wherein generating the target deep learning engine comprises:

based on the base deep learning engine that is pre-trained to perform the base radiotherapy treatment planning task according to a base rule, generating the target deep learning engine to perform the target radiotherapy treatment planning task associated with a target rule.

4 . The method of claim 1 , wherein generating the target deep learning engine comprises:

generating the target deep learning engine to perform one of the following target radiotherapy treatment planning tasks: automatic segmentation to generate structure data based on image data; dose prediction to generate dose data based on structure data and image data; and treatment delivery data prediction to generate treatment delivery data.

5 . The method of claim 1 , wherein generating the target deep learning engine comprises one of the following:

configuring the multiple base layers of the base deep learning engine to be variable base layers; and

generating the multiple target layers by modifying the respective variable base layers based on the target training data.

6 . The method of claim 1 , wherein generating the target deep learning engine comprises:

configuring the multiple base layers of the base deep learning engine to be invariable base layers; and

generating the feature data using the invariable base layers based on the target training data.

7 . The method of claim 1 , wherein generating the target deep learning engine comprises:

configuring the multiple base layers of the base deep learning engine to include both invariable base layers and variable base layers.

8 . A non-transitory computer-readable storage medium that includes a set of instructions which, in response to execution by a processor of a computer system, interfaced with a treatment delivery system, cause the processor to perform a method of deep transfer learning for radiotherapy treatment planning, wherein the method comprises:

obtaining, by the computer system, a base deep learning engine that is pre-trained to perform a base radiotherapy treatment planning task, wherein the base deep learning engine includes multiple base layers; and

based on the base deep learning engine, generating, by the computer system, a target deep learning engine to perform a target radiotherapy treatment planning task by performing at least one of the following, wherein the base deep learning engine is trained based on base training data associated with the base radiotherapy treatment planning task in a first phase, the target deep learning engine is trained based on target training data associated with the target radiotherapy treatment planning task in a second phase later than the first phase, and the target radiotherapy treatment planning task and the base radiotherapy treatment planning task are two different radiotherapy treatment planning tasks:

configuring a variable base layer among the multiple base layers of the base deep learning engine, and generating one of multiple target layers of the target deep learning engine by modifying the variable base layer; and

configuring an invariable base layer among the multiple base layers of the base deep learning engine, and generating one of multiple target layers of the target deep learning engine based on feature data generated using the invariable base layer;

performing, by the computer system, the target radiotherapy treatment planning task to generate output data based on input data associated with a particular patient in a third phase using the target deep learning engine, wherein the third phase is later than the second phase; and

generating, by the computer system, a treatment plan based on the output data, wherein the treatment delivery system retrieves the treatment plan and delivers a radiotherapy treatment to the particular patient according to the treatment plan.

9 . The non-transitory computer-readable storage medium of claim 8 , wherein generating the target deep learning engine comprises:

based on the base deep learning engine that is pre-trained to perform the base radiotherapy treatment planning task associated with a base anatomical site, generating the target deep learning engine to perform the target radiotherapy treatment planning task associated with a target anatomical site.

10 . The non-transitory computer-readable storage medium of claim 8 , wherein generating the target deep learning engine comprises:

based on the base deep learning engine that is pre-trained to perform the base radiotherapy treatment planning task according to a base rule, generating the target deep learning engine to perform the target radiotherapy treatment planning task associated with a target rule.

11 . The non-transitory computer-readable storage medium of claim 8 , wherein generating the target deep learning engine comprises:

generating the target deep learning engine to perform one of the following target radiotherapy treatment planning tasks: automatic segmentation to generate structure data based on image data; dose prediction to generate dose data based on structure data and image data; and treatment delivery data prediction to generate treatment delivery data.

12 . The non-transitory computer-readable storage medium of claim 8 , wherein generating the target deep learning engine comprises one of the following:

configuring the multiple base layers of the base deep learning engine to be variable base layers; and

generating the multiple target layers by modifying the respective variable base layers based on the target training data.

13 . The non-transitory computer-readable storage medium of claim 8 , wherein generating the target deep learning engine comprises:

configuring the multiple base layers of the base deep learning engine to be invariable base layers; and

generating the feature data using the invariable base layers based on the target training data.

14 . The non-transitory computer-readable storage medium of claim 8 , wherein generating the target deep learning engine comprises:

configuring the multiple base layers of the base deep learning engine to include both invariable base layers and variable base layers.

15 . A computer system interfaced with a treatment delivery system and configured to perform deep transfer learning for radiotherapy treatment planning, wherein the computer system comprises: a processor and a non-transitory computer-readable medium having stored thereon instructions that, when executed by the processor, cause the processor to:

obtain, by the computer system, a base deep learning engine that is pre-trained to perform a base radiotherapy treatment planning task, wherein the base deep learning engine includes multiple base layers; and

based on the base deep learning engine, generate, by the computer system, a target deep learning engine to perform a target radiotherapy treatment planning task by performing at least one of the following, wherein the base deep learning engine is trained based on base training data associated with the base radiotherapy treatment planning task in a first phase, the target deep learning engine is trained based on target training data associated with the target radiotherapy treatment planning task in a second phase later than the first phase, and the target radiotherapy treatment planning task and the base radiotherapy treatment planning task are two different radiotherapy treatment planning tasks:

configure a variable base layer among the multiple base layers of the base deep learning engine, and generating one of multiple target layers of the target deep learning engine by modifying the variable base layer; and

configure an invariable base layer among the multiple base layers of the base deep learning engine, and generating one of multiple target layers of the target deep learning engine based on feature data generated using the invariable base layer;

perform, by the computer system, the target radiotherapy treatment planning task to generate output data based on input data associated with a particular patient in a third phase using the target deep learning engine, wherein the third phase is later than the second phase; and

generate, by the computer system, a treatment plan based on the output data, wherein the treatment delivery system retrieves the treatment plan and delivers a radiotherapy treatment to the particular patient according to the treatment plan.

16 . The computer system of claim 15 , wherein the instructions for generating the target deep learning engine cause the processor to:

based on the base deep learning engine that is pre-trained to perform the base radiotherapy treatment planning task associated with a base anatomical site, generate the target deep learning engine to perform the target radiotherapy treatment planning task associated with a target anatomical site.

17 . The computer system of claim 15 , wherein the instructions for generating the target deep learning engine cause the processor to:

based on the base deep learning engine that is pre-trained to perform the base radiotherapy treatment planning task according to a base rule, generate the target deep learning engine to perform the target radiotherapy treatment planning task associated with a target rule.

18 . The computer system of claim 15 , wherein the instructions for generating the target deep learning engine cause the processor to:

generate the target deep learning engine to perform one of the following target radiotherapy treatment planning tasks: automatic segmentation to generate structure data based on image data; dose prediction to generate dose data based on structure data and image data; and treatment delivery data prediction to generate treatment delivery data.

19 . The computer system of claim 15 , wherein the instructions for generating the target deep learning engine cause the processor to:

configure the multiple base layers of the base deep learning engine to be variable base layers; and

generate the multiple target layers by modifying the respective variable base layers based on the target training data.

20 . The computer system of claim 15 , wherein the instructions for generating the target deep learning engine cause the processor to:

configure the multiple base layers of the base deep learning engine to be invariable base layers; and

generate the feature data using the invariable base layers based on the target training data.

21 . The computer system of claim 15 , wherein the instructions for generating the target deep learning engine cause the processor to:

configure the multiple base layers of the base deep learning engine to include both invariable base layers and variable base layers.

22 . The method of claim 1 , wherein the treatment plan includes data to control the treatment delivery system.

Assignments (2)
CHANGE OF NAME Recorded Jan 4, 2023
From: VARIAN MEDICAL SYSTEMS INTERNATIONAL AG
To: SIEMENS HEALTHINEERS INTERNATIONAL AG
Reel/Frame 062266/0780 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jun 16, 2021
From: LAAKSONEN, HANNU; NORD, JANNE; PERTTU, SAMI PETRI; SCHREIER, JAN; RUOKOLA, TOMI
To: VARIAN MEDICAL SYSTEMS INTERNATIONAL AG
Reel/Frame 056569/0438 →
Continuity (2)
Provisional Application 62783217 · Dec 21, 2018
Related Publication 20220051781A1 · Feb 17, 2022
References Cited (12)
US 20170360404A1 · Gafner et al. · 2017 [cited by applicant]
US 20180052962A1 · Van Der Koijk · 2018 [cited by examiner]
US 20180315188A1 · Tegzes · 2018 [cited by examiner]
US 20200065940A1 · Tang · 2020 [cited by examiner]
CN 101505652A · 2009 [cited by applicant]
CN 107731298A · 2018 [cited by applicant]
WO WO2018048507A1 · 2018 [cited by examiner]
International Search Report and Written Opinion of the International Searching Authority, International application No. PCT/EP2019/064145, Aug. 28, 2019. [cited by applicant]
Xin Zhen et al., “Deep Convolutional Neural Network with Transfer Learning for Rectum Toxicity Prediction in Cervical Cancer Radiotherapy: A Feasibility Study”, Physics in Medicine & Biology, Institute of Physics Publis… [cited by applicant]
Will Koehrsen, “Transfer Learning with Convolutional Neural Networks in PyTorch”, Retrieved on Aug. 16, 2019 from the Internet at <URL: https://towardsdatascience.com/transfer-learning-with-convolutional-neural-networks… [cited by applicant]
Roland Becker, “Transfer Learning-So können neuronale Netze voneinander lernen”, Retrieved on Aug. 16, 2019 from the internet at <URL: https://jaai.de/transfer-learning-1739/>, Feb. 12, 2018. [cited by applicant]
Jason Brownlee, “A Gentle Introduction to Transfer Learning for Deep Learning”, Deep Learning for Computer Vision, Dec. 20, 2017, Retrieved on Aug. 16, 2019 from the Internet at <URL: https://machinelearningmastery.com/… [cited by applicant]