18058884. CROSS DOMAIN SEGMENTATION WITH UNCERTAINTY-GUIDED CURRICULUM LEARNING simplified abstract (Siemens Healthcare GmbH)

CROSS DOMAIN SEGMENTATION WITH UNCERTAINTY-GUIDED CURRICULUM LEARNING

Organization Name

Siemens Healthcare GmbH

Inventor(s)

Yue Zhang of Jersey City NJ (US)

Caius Constantin Suliman

Florin-Cristian Ghesu of Skillman NJ (US)

Rui Liao of Princeton Junction NJ (US)

CROSS DOMAIN SEGMENTATION WITH UNCERTAINTY-GUIDED CURRICULUM LEARNING - A simplified explanation of the abstract

This abstract first appeared for US patent application 18058884 titled 'CROSS DOMAIN SEGMENTATION WITH UNCERTAINTY-GUIDED CURRICULUM LEARNING

Simplified Explanation

The abstract describes a method for training a machine learning based segmentation network using medical images. Here is a simplified explanation of the abstract:

• Medical images in one modality are received.
• Synthetic images in a different modality are generated based on the received medical images.
• Augmented images are created based on the synthetic images.
• Segmentations of anatomical objects are performed using a machine learning based reference network.
• Uncertainty associated with segmenting the anatomical object is computed.
• Suitability of medical images for training a segmentation network is determined based on the uncertainty.
• The segmentation network is trained using suitable medical images and annotations from a machine learning based teacher network.

---

1. 1. 1. Potential Applications

This technology can be applied in medical imaging for accurate segmentation of anatomical structures, aiding in diagnosis and treatment planning.

1. 1. 1. Problems Solved

This technology helps in automating the segmentation process in medical imaging, reducing manual effort and potential errors.

1. 1. 1. Benefits

- Improved accuracy in segmenting anatomical objects - Efficient training of machine learning based segmentation networks - Enhanced medical image analysis capabilities

1. 1. 1. Potential Commercial Applications

- Medical imaging software development - Healthcare AI solutions - Radiology and diagnostic imaging services

1. 1. 1. Possible Prior Art

Prior art may include similar methods for training machine learning models in medical imaging, such as using synthetic data for augmentation and uncertainty estimation for model training.

---

1. 1. 1. 1. Unanswered Questions

1. 1. 1. How does this method compare to traditional manual segmentation techniques in terms of accuracy and efficiency?

The article does not provide a direct comparison between this method and traditional manual segmentation techniques. Further studies or experiments may be needed to evaluate the performance of this technology against traditional methods.

1. 1. 1. What are the potential limitations or challenges of implementing this technology in real-world medical imaging settings?

The article does not address potential limitations or challenges of implementing this technology in real-world medical imaging settings. Factors such as data variability, computational resources, and integration with existing systems could be important considerations for practical implementation.

Original Abstract Submitted

Systems and methods for training a machine learning based segmentation network are provided. A set of medical images, each depicting an anatomical object, in a first modality is received. For each respective medical image of the set of medical images, a synthetic image, depicting the anatomical object, in a second modality is generated based on the respective medical image. One or more augmented images are generated based on the synthetic image. One or more segmentations of the anatomical object are performed from the one or more augmented images using a machine learning based reference network. An uncertainty associated with segmenting the anatomical object from the respective medical image is computed based on results of the one or more segmentations. It is determined whether the respective medical image is suitable for training a machine learning based segmentation network based on the uncertainty. The machine learning based segmentation network is trained based on 1) the suitable medical images of the set of medical images and 2) annotations of the anatomical object determined using a machine learning based teacher network.