A dataset of primary nasopharyngeal carcinoma MRI with multi-modalities segmentation

2024-04-04

arXiv.org

AI Summary

A publicly available, high-quality multimodal MRI dataset for treatment-naïve nasopharyngeal carcinoma (NPC) is currently lacking, hindering the development of intelligent diagnostic, precise tumor segmentation, and staging algorithms. Method: We constructed the first open multimodal MRI benchmark dataset specifically for NPC, comprising 831 axial T1-weighted, T2-weighted, and contrast-enhanced T1-weighted scans from 277 treatment-naïve patients. All scans are accompanied by pixel-level tumor segmentations manually delineated by radiologists and structured clinical metadata. Imaging acquisition followed standardized protocols, and annotations underwent rigorous double-blinded quality control, ensuring tight alignment among imaging sequences, lesion annotations, and clinical phenotypes. Contribution/Results: This dataset fills a critical gap in publicly accessible, multimodal NPC benchmarks. It serves as foundational infrastructure for training deep learning models, validating algorithms across institutions, and developing clinical decision-support systems.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Security and Privacy: Data transparency and provenance
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Multi-modality magnetic resonance imaging data with various sequences facilitate the early diagnosis, tumor segmentation, and disease staging in the management of nasopharyngeal carcinoma (NPC). The lack of publicly available, comprehensive datasets limits advancements in diagnosis, treatment planning, and the development of machine learning algorithms for NPC. Addressing this critical need, we introduce the first comprehensive NPC MRI dataset, encompassing MR axial imaging of 277 primary NPC patients. This dataset includes T1-weighted, T2-weighted, and contrast-enhanced T1-weighted sequences, totaling 831 scans. In addition to the corresponding clinical data, manually annotated and labeled segmentations by experienced radiologists offer high-quality data resources from untreated primary NPC.

Problem

• Lack of public MRI datasets for nasopharyngeal carcinoma research
• Need for multi-modality NPC data to improve diagnosis and treatment
• Absence of annotated segmentation data for machine learning development

Innovation

• Multi-modality MRI dataset for NPC
• Includes T1, T2, contrast-enhanced scans
• Manually annotated by expert radiologists