New AI research proposes a simple but effective structure-based encoder for learning the representation of proteins based on their 3D structures

Mayo Clinic AI researchers present machine learning-based method for leveraging diffusion patterns to build a multitasking brain tumor inpainting algorithm

The number of publications on artificial intelligence and, in particular, machine learning (ML) related to medical imaging has increased significantly in recent years. A current PubMed search using the Mesh keywords “artificial intelligence” and “radiology” yielded 5,369 articles in 2021, more than five times the results found in 2011. ML models are constantly being developed to improve healthcare efficiencies and outcomes, from classification to semantic segmentation to object detection and image generation. Numerous published reports in diagnostic radiology, for example, indicate that ML models have the ability to perform as well or even better than medical experts in specific tasks, such as abnormality detection and pathology screening.

It is therefore undeniable that, if used correctly, AI can help radiologists and drastically reduce their workload. Despite growing interest in developing ML models for medical imaging, significant challenges may limit practical applications of such models or even predispose them to substantial bias. Data scarcity and data imbalance are two such challenges. On the one hand, medical imaging datasets are often much smaller than natural photography datasets such as ImageNet, and pooling institutional datasets or making them public may be impossible due to patient privacy concerns. On the other hand, the medical imaging datasets that data scientists have access to could also be more balanced.

In other words, the volume of medical imaging data for patients with specific pathologies is significantly less than for patients with common pathologies or healthy people. Using insufficiently large or imbalanced datasets to train or evaluate a machine learning model can lead to systemic biases in model performance. Synthetic image generation is one of the primary strategies to combat data scarcity and data imbalance, along with the public release of de-identified medical imaging datasets and the endorsement of strategies such as federated learning, which enables the development of machine learning (ML) models on multi-institutional datasets without data sharing.

Build high-quality training datasets with Kili Technology and solve NLP machine learning challenges to develop powerful ML applications

Generative ML models can learn to generate realistic medical imaging data that doesn’t belong to a real patient and can then be shared publicly without compromising patient privacy. Since the emergence of Generative Adversarial Networks (GANs) various generation models capable of synthesizing high-quality synthetic data have been introduced. Most of these models produce unlabeled imaging data, which can be useful in specific applications, such as self-supervised or semi-supervised downstream models. In addition, some other models are capable of conditional generation, which allows you to generate an image based on predetermined clinical, textual, or imaging variables.

Probabilistic noise diffusion models (DDPMs), also known as diffusion models, are a new class of imaging models that surpass GANs in synthetic image quality and output diversity. This latter class of generative models enables the generation of labeled synthetic data that advances research into machine learning, medical imaging quality, and patient care. Despite their enormous success in generating synthetic medical imaging data, GANs are often chastised for their lack of output diversity and unstable formation. Autoencoder deep learning models are a more traditional alternative to GANs because they are easier to train and produce more diverse output. However, their synthetic results lack the image quality of GANs.

Diffusion models based on Markov chain theory learn to generate their synthetic outputs by gradually de-noising an initial image filled with random Gaussian noise. This iterative process of denoising causes inference runs of diffusion models to be significantly slower than those of other generative models. However, it allows them to extract more representative features from their input data, allowing them to outperform other models. They present a proof-of-concept diffusion model that can be used for multitasking brain tumor inpainting on multi-sequential brain magnetic resonance imaging (MRI) studies in this methodological paper.

They have created a diffusion model that can receive a two -dimensional axial slice (2D) from a sequence weighed in T1 (T1), weighed in T1 with contrast (T1CE), weighed in T2 (T2) or Flair of a cerebral magnetic resonance imaging and inpaint a cutlery area defined by the user of that slice with a realistic and controllable impact of a high degree and high components and correspondents (for example, the surrounding edema) or cerebral fabrics without cancer (apparently normal).

In the United States, the incidence of high-grade glioma is 3.56 per 100,000 people, and there are only a few publicly available MRI datasets for brain tumors. Their model will allow ML researchers to modify (induce or remove) tumor-free or tumor-free synthetic tissues with configurable features on brain MRI slices in such limited data. The tool was implemented online for people to use. The model has been made open source along with its documentation on GitHub.

This Article is written as a research summary article by Marktechpost Staff based on the research paper 'MULTITASK BRAIN TUMOR INPAINTING WITH DIFFUSION MODELS: A METHODOLOGICAL REPORT'. All Credit For This Research Goes To Researchers on This Project. Check out the paper, code and tool.
Please Don't Forget To Join Our ML Subreddit


Aneesh Tickoo is a Consulting Intern at MarktechPost. She is currently pursuing her BA in Data Science and Artificial Intelligence from Indian Institute of Technology (IIT), Bhilai. She spends most of her time working on projects that harness the power of machine learning. Her research interest is image processing and she is passionate about building solutions around it. She loves connecting with people and collaborating on interesting projects.


Gain a competitive edge with data – actionable market insights for global brands, retailers, analysts and investors. (Sponsored)

#Mayo #Clinic #researchers #present #machine #learningbased #method #leveraging #diffusion #patterns #build #multitasking #brain #tumor #inpainting #algorithm
Image Source : www.marktechpost.com

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *