View source for 17768419. Synthetic Generation of Clinical Skin Images in Pathology simplified abstract (GOOGLE LLC)

=Synthetic Generation of Clinical Skin Images in Pathology=


==Organization Name==

[[:Category:GOOGLE LLC|GOOGLE LLC]]


[[Category:GOOGLE LLC]]

==Inventor(s)==

[[:Category:Vivek Natarajan of Mountain View CA (US)|Vivek Natarajan of Mountain View CA (US)]][[Category:Vivek Natarajan of Mountain View CA (US)]]

[[:Category:Yuan Liu of Mountain View CA (US)|Yuan Liu of Mountain View CA (US)]][[Category:Yuan Liu of Mountain View CA (US)]]

[[:Category:David Coz of Mountain View CA (US)|David Coz of Mountain View CA (US)]][[Category:David Coz of Mountain View CA (US)]]

[[:Category:Amirata Ghorbani of Mountain View CA (US)|Amirata Ghorbani of Mountain View CA (US)]][[Category:Amirata Ghorbani of Mountain View CA (US)]]

==Synthetic Generation of Clinical Skin Images in Pathology - A simplified explanation of the abstract==


This abstract first appeared for US patent application 17768419 titled 'Synthetic Generation of Clinical Skin Images in Pathology

==Simplified Explanation==

The patent application describes the use of Generative Adversarial Networks (GAN) to generate synthetic clinical images with various skin conditions. These synthetic images can be used for data augmentation and training skin condition classifiers.

* GANs are used to create synthetic clinical images with different skin conditions.
* The synthetic images can vary in size, location, and skin color, while maintaining high fidelity.
* These synthetic images can be used to train skin condition classifiers and improve their ability to detect rare but malignant conditions.

== Potential Applications ==
The technology can be applied in the fields of dermatology, medical imaging, and artificial intelligence for healthcare.

== Problems Solved ==
The technology addresses the need for diverse and high-quality clinical images for training skin condition classifiers, especially for rare and malignant conditions.

== Benefits ==
The use of synthetic images generated by GANs can improve the accuracy and performance of skin condition classifiers, leading to better diagnosis and treatment outcomes for patients.

== Potential Commercial Applications ==
This technology can be utilized by healthcare providers, medical imaging companies, and AI developers to enhance their diagnostic tools and services.

== Possible Prior Art ==
Prior art may include research on GANs for image generation in various domains, as well as studies on data augmentation techniques for medical imaging.

== Unanswered Questions ==
1. How does the performance of skin condition classifiers trained on synthetic images compare to those trained on real clinical images?
2. Are there any limitations or challenges in using synthetic images for training classifiers in clinical settings?


==Original Abstract Submitted==

We disclose the generation and training of Generative Adversarial Networks (GAN) to synthesize clinical images with skin conditions. Synthetic images for a pre-specified skin condition are generated, while being able to vary its size, location and the underlying skin color. We demonstrate that the generated images are of high fidelity using objective GAN evaluation metrics. The synthetic images are not only visually similar to real images, but also embody the respective skin conditions. Additionally, synthetic skin images can be used as a data augmentation technique for training a skin condition classifier, and improve the ability of the classifier to detect rare but malignant conditions.



[[Category:G06T7/00]]
[[Category:G06T11/00]]
[[Category:G16H10/60]]
[[Category:G16H50/70]]