Article
Computer Science, Artificial Intelligence
Ziyu Su, Thomas E. Tavolara, Gabriel Carreno-Galeano, Sang Jin Lee, Metin N. Gurcan, M. K. K. Niazi
Summary: This study proposes attention2majority, a weak multiple instance learning model, to automatically and efficiently process whole slide images (WSIs) of stained tissue sections for classification. By using intelligent sampling and a multi-head attention-based multiple instance learning model, slide-level classification based on high-confidence patches is achieved.
MEDICAL IMAGE ANALYSIS
(2022)
Review
Biochemistry & Molecular Biology
Anna Timakova, Vladislav Ananev, Alexey Fayzullin, Vladimir Makarov, Elena Ivanova, Anatoly Shekhter, Peter Timashev
Summary: The analysis of microvasculature and angiogenesis holds significant prognostic value in various diseases, including cancer. Traditional evaluation methods are time consuming and subject to variability among observers. Artificial intelligence, specifically computer vision solutions, can rapidly analyze blood vessel structures in whole slide images, leading to a new era in computational pathology.
Article
Computer Science, Artificial Intelligence
Ahmed Naglah, Fahmi Khalifa, Ayman El-Baz, Dibson Gondim
Summary: This study proposes a novel digital pathology system that accurately detects and quantifies the footprint of fibrous tissue in HE whole slide images. By using a deep learning model to generate virtual MT images, the system achieved significant improvement over traditional methods in the experiments.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Computer Science, Artificial Intelligence
Paul Tourniaire, Marius Ilie, Paul Hofman, Nicholas Ayache, Herve Delingette
Summary: Using mixed supervision, we improve the classification and localization performances of a weakly-supervised model based on attention-based deep Multiple Instance Learning. With a limited amount of patch-level labeled slides, we achieve performance close to fully-supervised models.
MEDICAL IMAGE ANALYSIS
(2023)
Article
Oncology
Nikita Shvetsov, Morten Gronnesby, Edvard Pedersen, Kajsa Mollersen, Lill-Tove Rasmussen Busund, Ruth Schwienbacher, Lars Ailo Bongo, Thomas Karsten Kilvaer
Summary: Tumor tissues from patients contain valuable prognostic and predictive information. Our approach utilizes AI pipelines and training data to identify tissue-infiltrating lymphocytes (TILs) in lung cancer tissue and provide prognostic information. This method has the potential to pave the way for large-scale deployment of digital pathology.
Article
Oncology
Hyun-Jong Jang, In-Hye Song, Sung-Hak Lee
Summary: The study demonstrated that a deep learning-based tissue classifier can be a useful tool for quantitative analysis of cancer tissue slides, showing the significant prognostic value of histomorphologic types in gastric cancer. This tool can efficiently support pathological workflow and provide insights into treatment planning.
Article
Computer Science, Information Systems
Yiping Jiao, Junhong Li, Shumin Fei
Summary: This study proposes an intuitive method to visualize the color style of Hematoxylin and Eosin (H&E) stained whole-slide images (WSIs) and validates its effectiveness in lung cancer research.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Quoc Dang Vu, Kashif Rajpoot, Shan E. Ahmed Raza, Nasir Rajpoot
Summary: Diagnostic, prognostic and therapeutic decision-making of cancer in pathology clinics can now be carried out based on analysis of multi-gigapixel tissue images, also known as whole-slide images (WSIs). Deep convolutional neural networks (CNNs) have been proposed to derive unsupervised WSI representations, but a major trade-off is the lack of interpretability. To address this, a handcrafted framework called Handcrafted Histological Transformer (H2T) is presented, which offers competitive performance and is faster than Transformer models.
MEDICAL IMAGE ANALYSIS
(2023)
Article
Computer Science, Artificial Intelligence
Ching-Wei Wang, Sheng-Chuan Huang, Yu-Ching Lee, Yu-Jie Shen, Shwu-Ing Meng, Jeff L. Gaol
Summary: This study developed an efficient and fully automatic hierarchical deep learning framework for bone marrow nucleated differential count (NDC) whole-slide image (WSI) analysis. The framework includes rapid localization, cell identification, and result integration steps, achieving high recall and accuracy in cell counting with superior performance compared to existing benchmark methods.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Chemistry, Analytical
Shakil Ahmed, Asadullah Shaikh, Hani Alshahrani, Abdullah Alghamdi, Mesfer Alrizq, Junaid Baber, Maheen Bakhtyar
Summary: The classification of pathology images is crucial for accurate disease analysis and patient treatment efficacy, but limited by a lack of large labeled datasets. The Kimia Path24 dataset was created specifically for histopathology image classification and retrieval, leading to significant accuracy improvements on the Inception-V3 and VGG-16 models through a transfer learning framework.
Article
Mathematics
Haixia Zheng, Yu Zhou, Xin Huang
Summary: This study proposes an effective spatially sensitive learning framework for cancer metastasis detection in whole-slide images, improving prediction consistency through a novel spatial loss function and achieving high precision and recall. By learning the spatial relationships between adjacent image patches, it provides more accurate detection results and is beneficial for early diagnosis of cancer metastasis.
Article
Computer Science, Interdisciplinary Applications
Hans Pinckaers, Wouter Bulten, Jeroen van der Laak, Geert Litjens
Summary: Prostate cancer is the most prevalent cancer among men in Western countries, and pathologists' evaluation is the gold standard for diagnosis. State-of-the-art convolutional neural networks are often patch-based and require detailed pixel-level annotations for effective training.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2021)
Article
Computer Science, Information Systems
Hakan Wieslander, Philip J. Harrison, Gabriel Skogberg, Sonya Jackson, Markus Friden, Johan Karlsson, Ola Spjuth, Carolina Wahlby
Summary: In this paper, a three-step pipeline is proposed to analyze biomedical image data using deep learning and conformal prediction. The process involves locating ROIs at low resolution, segmenting ROIs at mid-resolution, and extracting quantitative measurements at full resolution. Limiting the analysis to sub-regions with full confidence is shown to reduce noise and increase separability of observed biological effects.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2021)
Article
Biology
Youqing Mu, H. R. Tizhoosh, Taher Dehkharghanian, Clinton J. V. Campbell
Summary: The goal of AI-based computational pathology is to generate compact representations of whole slide images (WSIs) capturing essential diagnostic information. This study explores trainable mechanisms to generate compact slide-level representations in bone marrow cytology using deep learning. The results show promising performance in WSI retrieval and classification tasks, indicating the potential of this method to improve diagnostics in hematology and support AI-assisted computational pathology approaches.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Oncology
Xiaobo Zhang, Wei Ba, Xiaoya Zhao, Chen Wang, Qiting Li, Yinli Zhang, Shanshan Lu, Lang Wang, Shuhao Wang, Zhigang Song, Danhua Shen
Summary: This study aims to develop a deep learning system for endometrial cancer detection using whole-slide images (WSIs). The model achieved a high degree of accuracy in identifying EC, serving as an assisted diagnostic tool to improve working efficiency for pathologists.
FRONTIERS IN ONCOLOGY
(2022)
Article
Materials Science, Multidisciplinary
Meryem Uzun-Per, Gregory J. Gillispie, Thomas Erol Tavolara, James J. Yoo, Anthony Atala, Metin Nafi Gurcan, Sang Jin Lee, Muhammad Khalid Khan Niazi
Summary: The lack of suitable bioinks for bioprinting is a major limitation in tissue engineering and regenerative medicine, primarily due to the contradictory requirements of bioinks needing to exhibit desirable bioactivity while also being highly printable. This study proposes methods and tools for automatically quantifying the performance of bioinks, reducing the time and effort needed for analysis and providing a standardized set of tools for comparison.
ADVANCED ENGINEERING MATERIALS
(2021)
Article
Health Policy & Services
James E. Peacock, David M. Herrington, Sharon L. Edelstein, Austin L. Seals, Ian D. Plumb, Sharon Saydah, William H. Lagarde, Michael S. Runyon, Patrick D. Maguire, Adolfo Correa, William S. Weintraub, Thomas F. Wierzba, John W. Sanders
Summary: Prevention behaviors are crucial to limiting the spread of SARS-CoV-2, yet a survey of over 20,000 individuals in the US found that most did not fully adhere to recommended public health safety measures during holiday gatherings following Thanksgiving and the winter holidays. Women were more likely to gather with non-household members (NHM), while older individuals and non-Hispanic Whites were more likely to wear masks when NHM were present. The extent to which failure to follow these recommendations contributed to the COVID-19 surges observed post-holidays remains uncertain.
JOURNAL OF COMMUNITY HEALTH
(2022)
Review
Pathology
Claudio Luchini, Liron Pantanowitz, Volkan Adsay, Sylvia L. Asa, Pietro Antonini, Ilaria Girolami, Nicola Veronese, Alessia Nottegar, Sara Cingarlini, Luca Landoni, Lodewijk A. Brosens, Anna V. Verschuur, Paola Mattiolo, Antonio Pea, Andrea Mafficini, Michele Milella, Muhammad K. Niazi, Metin N. Gurcan, Albino Eccher, Ian A. Cree, Aldo Scarpa
Summary: Ki-67 assessment plays a key role in the diagnosis of neuroendocrine neoplasms (NENs) from all anatomic locations. Digital pathology combined with machine learning has shown to be highly accurate and reproducible for evaluating Ki-67 in NENs. In this systematic review, the advantages of digital image analysis (DIA) in assessing Ki-67 in pancreatic NENs (PanNENs) were highlighted, including improved standardization and reliability, as well as increased speed and practicality compared to manual counting. However, limitations such as higher costs and operator qualification issues need to be addressed. A comparative meta-analysis showed a high concordance between DIA and manual counting. These findings support the widespread adoption of validated DIA methods for Ki-67 assessment in PanNENs.
Article
Computer Science, Artificial Intelligence
Hamidullah Binol, M. Khalid Khan Niazi, Charles Elmaraghy, Aaron C. Moberly, Metin N. Gurcan
Summary: The lack of objective evaluation methods for the eardrum is a critical barrier to accurate diagnosis. This paper proposes a novel deep learning-based method called OtoXNet, which automatically learns features for eardrum classification from otoscope video clips. By utilizing multiple composite image generation methods, OtoXNet proves to outperform baseline approaches in qualitative results, showing the advantage of using multiple composite images in analyzing eardrum abnormalities.
NEURAL COMPUTING & APPLICATIONS
(2022)
Review
Otorhinolaryngology
Stephany Ngombu, Hamidullah Binol, Metin N. Gurcan, Aaron C. Moberly
Summary: This review discusses the state of the art applications of artificial intelligence (AI) techniques in diagnosing otitis media (OM) and highlights the potential benefits of using AI to automate and aid in diagnosis.
OTOLARYNGOLOGY-HEAD AND NECK SURGERY
(2023)
Article
Computer Science, Artificial Intelligence
Ziyu Su, Thomas E. Tavolara, Gabriel Carreno-Galeano, Sang Jin Lee, Metin N. Gurcan, M. K. K. Niazi
Summary: This study proposes attention2majority, a weak multiple instance learning model, to automatically and efficiently process whole slide images (WSIs) of stained tissue sections for classification. By using intelligent sampling and a multi-head attention-based multiple instance learning model, slide-level classification based on high-confidence patches is achieved.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Public, Environmental & Occupational Health
Lydia E. Calamari, Ashley H. Tjaden, Sharon L. Edelstein, William S. Weintraub, Roberto Santos, Michael Gibbs, Johnathan Ward, Michele Santacatterina, Alain G. Bertoni, Lori M. Ward, Sharon Saydah, Ian D. Plumb, Michael S. Runyon
Summary: This study investigated self-reported mask use among participants in the COVID-19 Community Research Partnership (CRP) and found that mask use was higher among vaccinated participants and those aged 65 years and older, female, racial or ethnic minority group, and healthcare workers. Lower mask use was associated with a history of self-reported prior COVID-19 illness.
PREVENTIVE MEDICINE REPORTS
(2022)
Proceedings Paper
Geosciences, Multidisciplinary
P. Bhugra, B. Bischke, C. Werner, R. Syrnicki, C. Packbier, P. Helber, C. Senaras, A. S. Rana, T. Davis, W. De Keersmaecker, D. Zanaga, A. Wania, R. Van de Kerchove, G. Marchisio
Summary: This paper evaluates the performance of satellite imagery in Land Use Land Cover (LULC) classification and finds that using multi-temporal images improves the accuracy of multi-label classification compared to using single-time-step images. This research is important for achieving efficient change detection and land monitoring methods.
2022 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM (IGARSS 2022)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Aysim Toker, Lukas Kondmann, Mark Weber, Marvin Eisenberger, Andres Camero, Jingliang Hu, Ariadna Pregel Hoderlein, Caglar Senaras, Timothy Davis, Daniel Cremers, Giovanni Marchisio, Xiao Xiang Zhu, Laura Leal-Taixe
Summary: Researchers propose the DynamicEarthNet dataset, which contains daily satellite observations and monthly pixel-wise semantic segmentation labels for 75 selected areas of interest around the world. They compare different algorithms and introduce a new evaluation metric, SCS, for time-series semantic change segmentation.
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022)
(2022)
Article
Oncology
Thomas E. Tavolara, Metin N. Gurcan, M. Khalid Khan Niazi
Summary: This study proposes an unsupervised method to learn meaningful features from histopathological imaging data. The method achieves high accuracy and correlation in classifying non-small cell lung cancer subtypes and scoring breast cancer proliferation. The significance of this method lies in its ability to learn meaningful features from raw imaging data without slide-level annotations.
Proceedings Paper
Computer Science, Information Systems
Thomas E. Tavolara, M. Khalid Khan Niazi, Gary Tozbikian, Robert Wesolowski, Metin N. Gurcan
Summary: This study developed an automated method to predict HER2 scores in breast cancer, using immunohistochemical staining images and tissue sections. The preliminary results showed potential for localizing and scoring HER2 using H&E images.
MEDICAL IMAGING 2022: DIGITAL AND COMPUTATIONAL PATHOLOGY
(2022)
Proceedings Paper
Computer Science, Information Systems
Thomas E. Tavolara, Arijit Dutta, Martin Burks, Wei Chen, Wendy Frankel, Metin N. Gurcan, M. Khalid Khan Niazi
Summary: This study successfully developed an automated algorithm that combines routine H&E staining with pan-cytokeratin staining to generate ground truth for tumor budding. The results demonstrated the potential feasibility of this method in identifying tumor buds.
MEDICAL IMAGING 2022: DIGITAL AND COMPUTATIONAL PATHOLOGY
(2022)
Review
Engineering, Biomedical
Diana Lim, Eric S. Renteria, Drake S. Sime, Young Min Ju, Ji Hyun Kim, Tracy Criswell, Thomas D. Shupe, Anthony Atala, Frank C. Marini, Metin N. Gurcan, Shay Soker, Joshua Hunsberger, James J. Yoo
Summary: Regenerative medicine and tissue engineering provide new therapeutic options for restoring, maintaining, or improving tissue function. To optimize the biological function of tissue-engineered clinical products, specific conditions must be maintained in a bioreactor to allow product maturation and mimic the in vivo environment. Real-time monitoring of product functional capacity is critical for quality management during manufacturing.
BIO-DESIGN AND MANUFACTURING
(2022)
Article
Health Policy & Services
Martin S. Kohn, Umit Topaloglu, Eric S. Kirkendall, Ajay Dharod, Brian J. Wells, Metin Gurcan
Summary: The nature of information in medicine has changed with the availability of massive, diverse data streams; a Learning Health System facilitates the development of medical decision-making tools and demonstrates enhanced value in decision-making; clinicians need to acquire skills necessary to work with big data in this era.
LEARNING HEALTH SYSTEMS
(2022)
Article
Medicine, Research & Experimental
Hamidullah Binol, Muhammad Khalid Khan Niazi, Garth Essig, Jay Shah, Jameson K. Mattingly, Michael S. Harris, Charles Elmaraghy, Theodoros Teknos, Nazhat Taj-Schaal, Lianbo Yu, Metin N. Gurcan, Aaron C. Moberly
Summary: This study investigated whether a single composite image stitched from a digital otoscopy video can provide sufficient diagnostic information for accurate diagnosis, and found that it can indeed provide accurate diagnostic information.