Cheng Jocelyn,  Leesmidt Kantheera,  Durand Rachelle,  Mongan John,  Cort Kayla,  Courtier Jesse

Final Pr. ID: Poster #: EDU-001

In recent years, the applications of artificial intelligence in radiology have increased exponentially. Randomized controlled trials demonstrating the potential of AI in radiology have been centered around deep learning techniques with artificial neural networks, such as convolutional neural networks (CNNs). Current applications of CNNs include the detection, segmentation and quantification of pathologic conditions. These tasks use deductive AI, which uses discriminative models to identify discrepancies between images. Generative AI is an emerging technique that has a number of exciting novel applications in the field of Pediatric Radiology. This exhibit will provide a practical overview to familiarize pediatric radiologists with this topic and review current and potential future use cases.

Generative adversarial networks (GAN) require two opposing networks, a generator that synthesizes new images, and a discriminator that differentiates between real and synthesized images. Through an iterative process, GAN creates highly realistic images that have exciting applications in image creation, translation between imaging modalities, acceleration of image synthesis and identification of pathologic abnormalities.

ChatGPT is a popular generative AI model that generates coherent responses to queries, and uses a Generative Pre-trained Transformer (GPT) network to encode and decode language data. ChatGPT is pre-trained on large volume text data to learn how language is used in different contexts. An encoder processes the input sequence and a decoder generates the output sequence. Both encoder and decoder have a multi-head self-attention mechanism allows differential weighing of the sequence to infer meaning and context. Potential applications in pediatric radiology are broad, including patient communication, providing appropriate imaging recommendations, report generation and review of radiologic safety guidelines and regulations. However, generative AI models require careful implementation, training and validation to reduce potential pitfalls in unintentionally synthesizing content that does not exist, or conversely, removing clinically relevant information from an image.

The purpose of this exhibit is to: (1) provide an introduction to Generative AI (GAN) and ChatGPT (2) review use cases of GAN and ChatGPT in pediatric radiology (3) provide practical methods for evaluating the performance of artificial intelligence algorithms in pediatric radiology applications. Read More

Authors:  Cheng Jocelyn , Leesmidt Kantheera , Durand Rachelle , Mongan John , Cort Kayla , Courtier Jesse

Keywords:  Generative Artificial Intelligence, Artificial Intelligence, ChatGPT

Jung Daniel

Final Pr. ID: Poster #: EDU-004

It hasn’t even been 1 year since chatGPT came out and changed the world. Yet, today, we already have Large Multimodal Models, similar to chatGPT, that can now process images and voice as well.

The capabilities of these models have specifically piqued the interest of Radiologists, who have been exploring use cases in both academic research and clinical practice. Researchers at the NIH have developed a GPT that can process radiology reports to process result findings, locally. This means that all the patient health information stays within your network, and you can use GPTs without compromising HIPAA or safety concerns.

What’s even greater about the local GPT that the NIH researchers have developed, is that there is no need for costly and time-consuming fine-tuning. The performance of the local GPT developed, just by utilizing some techniques, were on par with the state-of-the-art tool developed by Stanford.

The purpose of this educational exhibit is to highlight the capabilities that LMMs have in diagnosing and augmenting the radiologist workflow.

I will discuss how there is no need to become a technical expert to ‘fine tune’ models. I will discuss how you can implement LMMs locally to protect health information. I will highlight the advanced LLM techniques that is used to improve the accuracy of outputted results. I will convey the idea that we need to embrace these technologies that are being improved upon at an exponential rate.

Ultimately, I will tie this in to how pediatric radiologists can use the techniques I talk about to conduct academic research, and will get people to think about how they can utilize their own local GPTs for various applications in both augmenting the clinical work flow and in academics. Read More

Authors:  Jung Daniel

Keywords:  AI, chatGPT, LMM