Publications by authors named "Felix Dorfner"
Rationale And Objectives: Training Convolutional Neural Networks (CNN) requires large datasets with labeled data, which can be very labor-intensive to prepare. Radiology reports contain a lot of potentially useful information for such tasks. However, they are often unstructured and cannot be directly used for training.
View Article and Find Full Text PDFArticle Synopsis
- Recent advancements in deep learning (DL) are enhancing clinical tools for analyzing brain tumors in MRI, aiding in tumor segmentation, quantification, and classification.
- DL provides objective and consistent measurements essential for accurate diagnosis, treatment planning, and tracking disease progression.
- Additionally, DL can help personalize medicine by predicting tumor characteristics and patient prognoses, with the review assessing both current uses and future possibilities.
Purpose: To examine whether incorporating anatomy-centred deep learning can improve generalisability and enable prediction of disease progression.
Methods: This retrospective multicentre study included conventional pelvic radiographs of four different patient cohorts focusing on axial spondyloarthritis collected at university and community hospitals. The first cohort, which consisted of 1483 radiographs, was split into training (n=1261) and validation (n=222) sets.
Article Synopsis
- * The study tested GPT-4o as a virtual advisor, providing tailored recommendations for machine learning (ML) and deep learning (DL) algorithms based on researchers' specific data needs.
- * Results showed GPT-4o effectively recommended suitable algorithms for various radiology tasks, signaling its potential to bridge knowledge gaps and enhance research quality in the field.
Article Synopsis
- Advances in large language models (LLMs) have led to numerous commercial and open-source models, but there has been no real-world comparison of OpenAI's GPT-4 against these models for extracting information from radiology reports.
- The study aimed to compare GPT-4 with several leading open-source LLMs in extracting relevant findings from chest radiograph reports using datasets from the ImaGenome and Massachusetts General Hospital.
- Results showed that GPT-4 slightly outperformed the best open-source model, Llama 2-70B, in terms of accuracy scores, with both showing strong performance in extracting findings from the reports.