paper-with-me

Papers

Model-based Cleaning of the QUILT-1M Pathology Dataset for Text-Conditional Image Synthesis

2024-04-11 · Marc Aubreville, Jonathan Ganz, Jonas Ammeling, Christopher C. Kaltenecker, Christof A. Bertram

The QUILT-1M dataset is the first openly available dataset containing images harvested from various online sources. While it provides a huge data variety, the image quality and composition is highly heterogeneous, impacting its utility for text-conditional image synthesis. We propose an automatic pipeline that provides predictions of the most common impurities within the images, e.g., visibility of narrators, desktop environment and pathology software, or text within the image. Additionally, we propose to use semantic alignment filtering of the image-text pairs. Our findings demonstrate that by rigorously filtering the dataset, there is a substantial enhancement of image fidelity in text-to-image tasks.

📄 PDF Abstract BibTeX arXiv:2404.07676

Code (1)

deepmicroscopy/quiltcleaner 공식 구현 pytorch

Tasks

Image Generation

Similar Papers 제목 키워드 기반

Quilt-1M: One Million Image-Text Pairs for Histopathology

2023-06-20 · NeurIPS 2023 11 · Wisdom Oluchi Ikezogwo, Mehmet Saygin Seyfioglu, Fatemeh Ghezloo, Dylan Stefan Chan Geva 외

Recent accelerations in multi-modal applications have been made possible with the plethora of image and text data available online. However, the scarcity of analogous data in the medical field, specifically in histopatho…

Automatic Speech RecognitionCross-Modal RetrievalRepresentation Learningspeech-recognition+1

Quilt-LLaVA: Visual Instruction Tuning by Extracting Localized Narratives from Open-Source Histopathology Videos

2023-12-07 · CVPR 2024 1 · Mehmet Saygin Seyfioglu, Wisdom O. Ikezogwo, Fatemeh Ghezloo, Ranjay Krishna 외

Diagnosis in histopathology requires a global whole slide images (WSIs) analysis, requiring pathologists to compound evidence from different WSI patches. The gigapixel scale of WSIs poses a challenge for histopathology m…

DiagnosticImage CaptioningVisual Question Answering (VQA)whole slide images

Effortless Vision-Language Model Specialization in Histopathology without Annotation

2025-08-11 · Jingna Qiu, Nishanth Jain, Jonas Ammeling, Marc Aubreville 외 arxiv

Recent advances in Vision-Language Models (VLMs) in histopathology, such as CONCH and QuiltNet, have demonstrated impressive zero-shot classification capabilities across various tasks. However, their general-purpose desi…

PathGLS: Evaluating Pathology Vision-Language Models without Ground Truth through Multi-Dimensional Consistency

2026-03-17 · Minbing Chen, Zhu Meng, Fei Su arxiv

Vision-Language Models (VLMs) offer significant potential in computational pathology by enabling interpretable image analysis, automated reporting, and scalable decision support. However, their widespread clinical adopti…

Natural Language Inference

Hyperparameter Optimization and Reproducibility in Deep Learning Model Training

2025-10-16 · Usman Afzaal, Ziyu Su, Usama Sajjad, Hao Lu 외 arxiv

Reproducibility remains a critical challenge in foundation model training for histopathology, often hindered by software randomness, hardware non-determinism, and inconsistent hyperparameter reporting. To investigate the…

Hyperparameter Optimization