An official website of the United States government
Digital pathology has transformed the traditional pathology practice of analyzing tissue under a microscope into a computer vision workflow. Whole-slide imaging allows pathologists to view and analyze microscopic images on a computer monitor, enabling computational pathology. By leveraging artificial intelligence (AI) and machine learning (ML), computational pathology has emerged as a promising field in recent years. Recently, task-specific AI/ML (eg, convolutional neural networks) has risen to the forefront, achieving above-human performance in many image-processing and computer vision tasks. The performance of task-specific AI/ML models depends on the availability of many annotated training datasets, which presents a rate-limiting factor for AI/ML development in pathology. Task-specific AI/ML models cannot benefit from multimodal data and lack generalization, eg, the AI models often struggle to generalize to new datasets or unseen variations in image acquisition, staining techniques, or tissue types. The 2020s are witnessing the rise of foundation models and generative AI. A foundation model is a large AI model trained using sizable data, which is later adapted (or fine-tuned) to perform different tasks using a modest amount of task-specific annotated data. These AI models provide in-context learning, can self-correct mistakes, and promptly adjust to user feedback. In this review, we provide a brief overview of recent advances in computational pathology enabled by task-specific AI, their challenges and limitations, and then introduce various foundation models. We propose to create a pathology-specific generative AI based on multimodal foundation models and present its potentially transformative role in digital pathology. We describe different use cases, delineating how it could serve as an expert companion of pathologists and help them efficiently and objectively perform routine laboratory tasks, including quantifying image analysis, generating pathology reports, diagnosis, and prognosis. We also outline the potential role that foundation models and generative AI can play in standardizing the pathology laboratory workflow, education, and training.
more »
« less
This content will become publicly available on July 24, 2026
Assisting human annotation of marine images with foundation models
Marine scientists have been leveraging supervised machine learning algorithms to analyze image and video data for nearly two decades. There have been many advances, but the cost of generating expert human annotations to train new models remains extremely high. There is broad recognition both in computer and domain sciences that generating training data remains the major bottleneck when developing ML models for targeted tasks. Increasingly, computer scientists are not attempting to produce highly-optimized models from general annotation frameworks, instead focusing on adaptation strategies to tackle new data challenges. Taking inspiration from large language models, computer vision researchers are now thinking in terms of “foundation models” that can yield reasonable zero- and few-shot detection and segmentation performance with human prompting. Here we consider the utility of this approach for ocean imagery, leveraging Meta’s Segment Anything Model to enrich ocean image annotations based on existing labels. This workflow yields promising results, especially for modernizing existing data repositories. Moreover, it suggests that future human annotation efforts could use foundation models to speed progress toward a sufficient training set to address domain specific problems.
more »
« less
- Award ID(s):
- 2230776
- PAR ID:
- 10656887
- Publisher / Repository:
- Frontiers of Marine Science
- Date Published:
- Journal Name:
- Frontiers in Marine Science
- Volume:
- 12
- ISSN:
- 2296-7745
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract Natural language processing (NLP) techniques can enhance our ability to interpret plant science literature. Many state-of-the-art algorithms for NLP tasks require high-quality labelled data in the target domain, in which entities like genes and proteins, as well as the relationships between entities, are labelled according to a set of annotation guidelines. While there exist such datasets for other domains, these resources need development in the plant sciences. Here, we present the Plant ScIenCe KnowLedgE Graph (PICKLE) corpus, a collection of 250 plant science abstracts annotated with entities and relations, along with its annotation guidelines. The annotation guidelines were refined by iterative rounds of overlapping annotations, in which inter-annotator agreement was leveraged to improve the guidelines. To demonstrate PICKLE’s utility, we evaluated the performance of pretrained models from other domains and trained a new, PICKLE-based model for entity and relation extraction (RE). The PICKLE-trained models exhibit the second-highest in-domain entity performance of all models evaluated, as well as a RE performance that is on par with other models. Additionally, we found that computer science-domain models outperformed models trained on a biomedical corpus (GENIA) in entity extraction, which was unexpected given the intuition that biomedical literature is more similar to PICKLE than computer science. Upon further exploration, we established that the inclusion of new types on which the models were not trained substantially impacts performance. The PICKLE corpus is, therefore, an important contribution to training resources for entity and RE in the plant sciences.more » « less
-
Recent progress in data-driven vision and language-based tasks demands developing training datasets enriched with multiple modalities representing human intelligence. The link between text and image data is one of the crucial modalities for developing AI models. The development process of such datasets in the video domain requires much effort from researchers and annotators (experts and non-experts). Researchers re-design annotation tools to extract knowledge from annotators to answer new research questions. The whole process repeats for each new question which is timeconsuming. However, since the last decade, there has been little change in how the researchers and annotators interact with the annotation process. We revisit the annotation workflow and propose a concept of an adaptable and scalable annotation tool. The concept emphasizes its users’ interactivity to make annotation process design seamless and efficient. Researchers can conveniently add newer modalities to or augment the extant datasets using the tool. The annotators can efficiently link free-form text to image objects. For conducting human-subject experiments on any scale, the tool supports the data collection for attaining group ground truth. We have conducted a case study using a prototype tool between two groups with the participation of 74 non-expert people. We find that the interactive linking of free-form text to image objects feels intuitive and evokes a thought process resulting in a high-quality annotation. The new design shows ≈ 35% improvement in the data annotation quality. On UX evaluation, we receive above-average positive feedback from 25 people regarding convenience, UI assistance, usability, and satisfaction.more » « less
-
NA (Ed.)Existing methods for last layer retraining that aim to optimize worst-group accuracy (WGA) rely heavily on well-annotated groups in the training data. We show, both in theory and practice, that annotation-based data augmentations using either downsampling or upweighting for WGA are susceptible to domain annotation noise. The WGA gap is exacerbated in highnoise regimes for models trained with vanilla empirical risk minimization (ERM). To this end, we introduce Regularized Annotation of Domains (RAD) to train robust last layer classifiers without needing explicit domain annotations. Our results show that RAD is competitive with other recently proposed domain annotation-free techniques. Most importantly, RAD outperforms state-of-the-art annotation-reliant methods even with only 5% noise in the training data for several publicly available datasets.more » « less
-
With the increased popularity of electronic textbooks, there is a growing interest in developing a new generation of “intelligent textbooks,” which have the ability to guide readers according to their learning goals and current knowledge. Intelligent textbooks extend regular textbooks by integrating machine-manipulable knowledge, and the most popular type of integrated knowledge is a list of relevant concepts mentioned in the textbooks. With these concepts, multiple intelligent operations, such as content linking, content recommendation, or student modeling, can be performed. However, existing automatic keyphrase extraction methods, even supervised ones, cannot deliver sufficient accuracy to be practically useful in this task. Manual annotation by experts has been demonstrated to be a preferred approach for producing high-quality labeled data for training supervised models. However, most researchers in the education domain still consider the concept annotation process as an ad-hoc activity rather than a carefully executed task, which can result in low-quality annotated data. Using the annotation of concepts for the Introduction to Information Retrieval textbook as a case study, this paper presents a knowledge engineering method to obtain reliable concept annotations. As demonstrated by the data we collected, the inter-annotator agreement gradually increased along with our procedure, and the concept annotations we produced led to better results in document linking and student modeling tasks. The contributions of our work include a validated knowledge engineering procedure, a codebook for technical concept annotation, and a set of concept annotations for the target textbook, which could be used as a gold standard in further intelligent textbook research.more » « less
